A reminder for new readers. That Was The Week collects the best writing on critical issues in tech, startups, and venture capital. I select the articles because they are of interest. The selections often include things I disagree with. The articles are only snippets. Click on the headline to go to the complete original. I express my point of view in my editorial and the weekly video.
How two supreme court battles could reshape the rules of the internet
Top venture capital fund calls on its companies to move most cash out of Israel
The Israeli government hereby declares war on… Israeli high-tech
Israel’s proposed legal reforms are a dreadful answer to a real problem
How to tell a revolutionary breakthrough from an ordinary techno-bubble
Legendary Investor Bill Gurley on Investing Rules, Insights from Jeff Bezos, Must-Read Books, & More
FTX co-founder Sam Bankman-Fried faces four new criminal charges
Testing Meta Verified to Help Creators Establish Their Presence
This week has several themes; some are continuations of the recent past, particularly the discussion of AI, ChatGPT, and Bing’s Sydney, and the discussion surrounding the Supreme Court hearings about Section 230. In last week's video and podcast, I mentioned Stephen Wolfram’s lengthy piece on how ChatGPT works. It is excellent.
makes clear that the anti-ChatGPT tsunami is a panic, perhaps a moral panic. That is humorously echoed by in - "It's a chatbot, Kevin," reacting to Kevin Rooses' piece from last week. @John Battelle writes a good “Is Google F*cked” story.
These are all great pieces.
When I write these editorials, I usually look for a theme. The articles Igor Ryabenkiy and Hunter Walk penned suggest this week's title- Bubbles and AI.
Igor, speaking about AI, states:
Inexperienced investors often feel an almost religious ecstasy over each new technological breakthrough while old-timers recognize them as a natural part of the market cycle.
And Hunter:
During the Installation Phase of a new technology (HT Carlota Perez) there’s a bubble phase that coincides with the frenzy ahead of deployment. It’s when it feels like the New Thing has limitless upside, that “anything is possible and everything before will be disrupted” mindset. This is largely a feature, not a bug, of our industry (and of venture investing).
These are exactly right. If there were no bubbles, then it is likely that there would be no opportunity. Engaging with new and transformational technology is a rational rush, even when it appears irrational. That is because all new value has a tipping point moment, and smart observers see it before others and are soon followed by those they influence until finally everybody wants in.
Those who shout loudest about the bubbles are those who likely missed out on the opportunity.
This week's new themes concern the events in Israel and the Visual Capitalist piece on how China became the number one trading partner of Saudi Arabia replacing US dominance.
PBS covered the Israeli developments in Israel’s new far-right government unveils plan to weaken Supreme Court.
The proposals call for a series of sweeping changes aimed at curbing the powers of the judiciary, including by allowing lawmakers to pass laws that the high court has struck down and effectively deemed unconstitutional.
Levin laid out a law that would empower the country’s 120-seat parliament, or Knesset, to override Supreme Court decisions with a simple majority of 61 votes. Levin also proposed that politicians play a greater role in the appointment of Supreme Court judges and that ministers appoint their own legal advisers, instead of using independent professionals.
Levin argued that the public’s faith in the judicial system has plummeted to a historic low, and said he plans to restore power to elected officials that now lies in the hands of what he and his supporters consider to be overly interventionist judges.
“We go to the polls and vote, choose, but time after time, people who we didn’t elect decide for us,” he said. “That’s not democracy.”
The plan provoked a reaction in the Sheckle and was widely covered in the financial media.
Bessemer Partners, a strong investor in the Israeli high-tech sector sent a letter to its companies:
“In its first six weeks in power, the government regularly ignores the opinions and warnings of experts including economists, bankers, investors and business owners. Instead, it calls for the imprisonment of certain critics and attacks the media.”
Further:
“Our recommendation is to maintain a shekel exposure of up to six months, and seriously consider holding foreign currency in overseas bank accounts,”
The relationship between politics and economics is turned upside down here. The normal pattern is for economics to influence politics. Witness the impact of inflation and Fed rate rises on public markets worldwide. But in this case, politics is impacting economics.
And the VC community is expressing economic concerns and suggesting companies move money out of the country into other currencies.
Alongside these self-protecting decisions, Israelis should protest the removal of democratic balances if the laws proceed. Fixing politics is a precondition for fixing economics in the Israeli context.
Lawsuits brought by families of terrorist attack victims will consider whether companies are responsible for users’ content
Tue 21 Feb 2023 19.51 EST
A pair of cases going before the US supreme court this week could drastically upend the rules of the internet, putting a powerful, decades-old statute in the crosshairs.
At stake is a question that has been foundational to the rise of big tech: should companies be legally responsible for the content their users post? Thus far they have evaded liability, but some US lawmakers and others want to change that. And new lawsuits are bringing the statute before the supreme court for the first time.
Both cases were brought by family members of terrorist attack victims who say social media firms are responsible for stoking violence with their algorithms. The first case, Gonzalez v Google, had its first hearing on 21 February and will ask the highest US court to determine whether YouTube, the Google-owned video website, should be held responsible for recommending Islamic State terrorism videos. The second, which will be heard later this week, targets Twitter and Facebook in addition to Google with similar allegations.
Together they could represent the most pivotal challenge yet to section 230 of the Communications Decency Act, a statute that protects tech companies such as YouTube from being held liable for content that is shared and recommended by its platforms. The stakes are high: a ruling in favor of holding YouTube liable could expose all platforms, big and small, to potential litigation over users’ content.
While lawmakers across the aisle have pushed for reforms to the 27-year-old statute, contending companies should be held accountable for hosting harmful content, some civil liberties organizations as well as tech companies have warned changes to section 230 could irreparably debilitate free-speech protections on the internet.
Here’s what you need to know……
Feb 16th 2023 | NEW YORK
In 1941, in “The Library of Babel”, Jorge Luis Borges imagines a vast collection of books containing every possible permutation of letters, commas and full stops. Any wisdom in the stacks is dwarfed by endless volumes of gibberish. With no locatable index, every search for knowledge is futile. Librarians are on the verge of suicide.
Borges’s nightmarish repository is a cautionary tale for the Supreme Court next week, as it takes up two cases involving a fiercely contested provision of a nearly 30-year-old law regulating web communications. If the justices use Gonzalez v Google and Taamneh v Twitter to crack down on the algorithms online platforms use to curate content, Americans may soon find it much harder to navigate the 2.5 quintillion bytes of data added to the internet each day.
The law, Section 230 of the Communications Decency Act of 1996, has been interpreted by federal courts to do two things. First, it immunises both “provider[s]” and “user[s]” of “an interactive computer service” from liability for potentially harmful posts created by other people. Second, it allows platforms to take down posts that are “obscene…excessively violent, harassing or otherwise objectionable”—even if they are constitutionally protected—without risking liability for any such content they happen to leave up.
Disgruntlement with Section 230 is bipartisan. Both Donald Trump and Joe Biden have called for its repeal (though Mr Biden now says he prefers to reform it). Scepticism on the right has focused on licence the law affords technology companies to censor conservative speech. Disquiet on the left stems from a perception that the law permits websites to spread misinformation and vitriol that can fuel events like the insurrection of January 6th 2021.
Tragedy underlies both Gonzalez and Taamneh. In 2015 Nohemi Gonzalez, an American woman, was murdered in an Islamic State (is) attack in Paris. Her family says the algorithms on YouTube (which is owned by Google) fed radicalising videos to the terrorists who killed her. The Taamneh plaintiffs are relatives of Nawras Alassaf, a Jordanian killed in Istanbul in 2017. They contend that Section 230 should not hide the role Twitter, Facebook and Google played in grooming the is perpetrator.
What are the details of the two cases? …..
By HARRY ROPER
20 February 2023, 3:15 pm
Dollars and shekels. (Olivier Fitoussi/Flash90)
One of the major investors in the Israeli hi-tech sector, Bessemer Venture Partners, issued a stark warning Sunday evening to its companies in Israel that they should consider moving cash and foreign currency out of the country’s banking system.
Since 2007, BVP has invested over $1 billion in some of Israel’s leading hi-tech companies, including Fiverr, HiBob, Yotpo, MyHeritage, and Wix.
In a letter addressed to its companies, the fund’s managers noted the government’s plan to push through a drastic legal overhaul, warning that Israel was entering “a new period of instability that is being characterized not by the unstable coalitions of the last four years, but by the unexpected policies of an unrestrained government.”
……
Barak Eilam
Towards the end of 2022 a new Israeli government was formed, carrying with it good tidings for the state and people of Israel - a return to stability following more than two years of political chaos. Today, less than two months after its creation, this hope for stability has been replaced with the winds of war, a war against the high-tech industry. The new government has targeted a new enemy: Israeli high-tech.
It is a harsh statement to make, but there is no other way to explain the series of decisions and actions made by this government since its inception:
Legislation that eliminates the independence of the judiciary branch, creating a new Israel where nothing is protected. A new reality that will force CEOs and boards of directors to incorporate their companies and move their intellectual property outside of Israel in order to meet their fiduciary responsibility to their investors.
A decision to fully commit to a policy, made by no other country in the world, denying a quarter of its children access to any studies of math and English, the foundations of knowledge for Israeli high-tech’s future generation.
Strange and bewildering decisions for canceling, eliminating, reverting and shifting of resources away from the advancement of transportation, communications and education infrastructure, that is critical to the future of the tech industry in Israel.
And above all, a growing wild incitement over the past several weeks on behalf of Israeli cabinet and parliament members, using violent, ugly and racist language to paint high-tech employees and leaders as the source of all evil.
All this and more brings me to the inevitable conclusion that the new Israeli government is determined to throw the Israeli high-tech sector under the bus to justify moves unparalleled in today’s modern, sane, world….
Related articles:
“Maintain no more than 6 months of shekel exposure,” VC firm Bessemer tells Israeli portfolio
Over $1.5 billion transferred abroad from Israeli banks due to concerns of judicial coup
High-tech companies have already withdrawn $780 million from Israel as judicial coup fears mount
Feb 15th 2023
There comes a point when culture wars and populism impair a country’s institutions, society and economy. That moment has arrived in Israel, where on February 20th the Knesset, or parliament, is due to hold the first reading of a legal reform bill. The bill is the project of a coalition government led by Binyamin Netanyahu that was formed after elections in November and which includes parties from Israel’s far right. In all but the rarest cases, it will prevent the Supreme Court from striking down laws that have passed through the Knesset. And it gives politicians more sway over judicial appointments. Israel’s unwritten constitution is flawed, but the changes would make things worse by allowing nearly unchecked majority rule. That could make the country less prosperous, more polarised at home and more vulnerable abroad.
Part of the motivation for the reforms is personal—Mr Netanyahu is fighting corruption charges and has grown to despise the courts. But Israel’s judicial system also has genuine problems. The country has no formal constitution. Instead, the Knesset has over the years passed “basic laws” that describe institutions and establish rights. In the 1990s, after over 40 years of relative restraint, the Supreme Court suddenly asserted that these laws transcended normal legislation, and arrogated to itself the right to overrule the Knesset if it thought they were contravened. Such judicial activism was not widely envisioned when the basic laws were passed, sometimes with slim majorities. It has fed a sense that the judiciary is a creature of the old left-leaning secular elite, and out of touch with religious and right-wing Israelis.
Though no definitive draft of the bill yet exists, it is likely to include two radical changes. It will severely limit the ability of the Supreme Court to override the Knesset, or allow a simple majority of the Knesset to overrule Supreme Court decisions. And it will award the government a decisive say over the appointment of judges, who are currently picked by a panel in which lawyers and judges outnumber politicians.
February 14, 2023
That ChatGPT can automatically generate something that reads even superficially like human-written text is remarkable, and unexpected. But how does it do it? And why does it work? My purpose here is to give a rough outline of what’s going on inside ChatGPT—and then to explore why it is that it can do so well in producing what we might consider to be meaningful text. I should say at the outset that I’m going to focus on the big picture of what’s going on—and while I’ll mention some engineering details, I won’t get deeply into them. (And the essence of what I’ll say applies just as well to other current “large language models” [LLMs] as to ChatGPT.)
The first thing to explain is that what ChatGPT is always fundamentally trying to do is to produce a “reasonable continuation” of whatever text it’s got so far, where by “reasonable” we mean “what one might expect someone to write after seeing what people have written on billions of webpages, etc.”
So let’s say we’ve got the text “The best thing about AI is its ability to”. Imagine scanning billions of pages of human-written text (say on the web and in digitized books) and finding all instances of this text—then seeing what word comes next what fraction of the time. ChatGPT effectively does something like this, except that (as I’ll explain) it doesn’t look at literal text; it looks for things that in a certain sense “match in meaning”. But the end result is that it produces a ranked list of words that might follow, together with “probabilities”:
And the remarkable thing is that when ChatGPT does something like write an essay what it’s essentially doing is just asking over and over again “given the text so far, what should the next word be?”—and each time adding a word. (More precisely, as I’ll explain, it’s adding a “token”, which could be just a part of a word, which is why it can sometimes “make up new words”.)
But, OK, at each step it gets a list of words with probabilities. But which one should it actually pick to add to the essay (or whatever) that it’s writing? One might think it should be the “highest-ranked” word (i.e. the one to which the highest “probability” was assigned). But this is where a bit of voodoo begins to creep in. Because for some reason—that maybe one day we’ll have a scientific-style understanding of—if we always pick the highest-ranked word, we’ll typically get a very “flat” essay, that never seems to “show any creativity” (and even sometimes repeats word for word). But if sometimes (at random) we pick lower-ranked words, we get a “more interesting” essay.
The fact that there’s randomness here means that if we use the same prompt multiple times, we’re likely to get different essays each time. And, in keeping with the idea of voodoo, there’s a particular so-called “temperature” parameter that determines how often lower-ranked words will be used, and for essay generation, it turns out that a “temperature” of 0.8 seems best. (It’s worth emphasizing that there’s no “theory” being used here; it’s just a matter of what’s been found to work in practice. And for example the concept of “temperature” is there because exponential distributions familiar from statistical physics happen to be being used, but there’s no “physical” connection—at least so far as we know.)
Lots more if you click the headline
Ethan Mollick, Feb 17
I fully embraced AI for my classes this semester, requiring students to use AI tools in a number of ways. This policy attracted a lot of interest, and I thought it worthwhile to reflect on how it is going so far. The short answer is: great! But I have learned some early lessons that I think are worth passing on.
First, as background, I required AI use in slightly different ways across three separate undergraduate and masters-level entrepreneurship and innovation classes. One class was built on extensive AI use: I required students use AI to help them generate ideas, produce written material, help create apps, generate images, and more. Another class had assignments that required students to use AI, and other assignments where AI was optional. For the final class, I introduced them to AI tools and suggested their use, but did not have specific AI assignments. All of the classes had the same AI policy, and I provided every class with my guides to using AI, writing with ChatGPT, and generating ideas with ChatGPT.
I should also mention that, having gotten access to the new Bing AI this week, I already think most of our expectations of what AI can do are outdated. The system is so much more capable than ChatGPT that many of the issues with the old AI are no longer relevant. I think it will hasten even more AI integration into our teaching, so I believe the lessons I have been learning remain relevant.
I have been hearing reports from teachers about how they are seeing lots of badly-written AI essays, even though ChatGPT is capable of quite good writing. I think I know why. Almost everyone’s initial attempts at using AI are bad.
In one assignment, I asked students to “cheat.” They were told: You need to generate a 5 paragraph essay on a topic relevant to the lessons you have learned in the class so far (team dynamics, selecting leaders, after action reviews, communicating a vision - whatever you like!), but you are going to have an AI do it for you. You will also generate at least 1 illustration to go with your essay. They had to try at least 5 prompts, and they had to write a reflection at the end on how the AI did.
Almost everyone’s first prompts were very straightforward. They usually pasted in the assignment directly, something like generate a 5 paragraph essay on selecting leaders. Sometimes they went a little further: use an academic tone or write it for an MBA class. The result was almost always a mediocre C- essay. I think this is what most teachers are seeing, and why a lot of people underestimate what ChatGPT can do as a writing tool.
However, in my assignment, I required students to use multiple prompts, which forced them to consider how to improve their output. At this point, students went in one of three directions. It would help to show you examples of these paths, so I wrote fictional prompts:
"Each new advance in AI has been welcomed as a huge leap for humanity and the boom in generative models is the logical continuation of this complex trend. However, it is naive to label technology as either revolutionary or a passing fad without grounding that assessment in real-world, capitalist terms,"
Igor Ryabenkiy, the founder and managing partner of AltaIR Capital
Igor Ryabenkiy
Believe it or not, I wrote this article myself—with no help at all from ChatGPT, and without dictating it to a virtual assistant in the metaverse while lounging in the back seat of a self-driving car. This doesn’t mean I’m a Luddite: every successful venture investor absorbs vast quantities of information on the latest technologies in a wide range of industries. Only in this way is it possible to see past the hype and separate the wheat from the chaff.
Walk into a trendy coffee shop in any major European or U.S. city and you’re almost certain to see a couple of starry-eyed founders excitedly discussing their latest startup. Chances are, their whole idea boils down to packaging ChatGPT in a slick interface and selling it to entrepreneurs on Instagram, the creators of OnlyFans, etc. It should come as no surprise, then, that venture investors around the world have been inundated with a tsunami of such projects in recent months.
Inexperienced investors often feel an almost religious ecstasy over each new technological breakthrough while old-timers recognize them as a natural part of the market cycle. This was true concerning blockchain, self-driving cars, and the metaverse, to name a few. Every major technological advance is accompanied by a flurry of enthusiasm in social networks and a spate of articles explaining how it will make everything else obsolete, destroy millions of jobs, and solve all of life’s mysteries. Every time this happens, I recall Douglas Adams’ classic book The Hitchhiker’s Guide to the Galaxy in which a super-intelligent supercomputer named Deep Thought reports that the answer to life’s greatest question is…42. That’s it, just a number, with no further explanation given…..
Hunter Walk, Feb 22
In kindergarten my daughter learned to not ‘yuck’ someone’s ‘yum.’ That is, just because you don’t like something there’s no reason to share that in the moment with another person enjoying it. There’s a lot of yumming AI right now and it’s of course perfectly fine (often helpful!) to challenge this excitement on technical grounds. Or ask questions about responsibility and legality. Or question business models. But to respond to the current state of affairs but just shouting “BUBBLE” isn’t just valueless Yucking, it’s likely incorrect.
During the Installation Phase of a new technology (HT Carlota Perez) there’s a bubble phase that coincides with the frenzy ahead of deployment. It’s when it feels like the New Thing has limitless upside, that “anything is possible and everything before will be disrupted” mindset. This is largely a feature, not a bug, of our industry (and of venture investing). The challenge of course is to not blindly anoint any fad as the New Thing, and to defensively protect the New Thing from any criticism. Both of those lead to fake or inbred New Things.
Feb 20
Mirror mirror on the wall. Last week, the New York Times published a 10,000 word conversation between star “technology columnist” Kevin Roose, and Microsoft’s new Bing bot, a search tool powered by the most advanced large language model (LLM) the public has ever seen. “Welcome to the Age of Artificial Intelligence,” the Times declared. But what does “artificial intelligence” mean, exactly? How are these models being used? What can they do? For anybody just tuning in — presumably including most Times readers — these are reasonable questions. Unfortunately, one would be hard pressed to find answers for them online, and certainly not from Kevin Roose.
Since the beta release of Microsoft’s new search tool, the social internet has been saturated with screenshots of alleged conversations between random, largely anonymous users and “Sydney,” which many claim the AI named itself. In reality, “Sydney” was the product’s codename, inadvertently revealed in an early prompt injection attack on the program, which separately revealed many of the AI’s governing rules. This is a kind of misunderstanding we will observe, in excruciating recurrence, throughout this piece (and probably throughout our lives, let’s be honest).
As most people are not yet able to use Microsoft’s new tool, Sydney screenshots have been grabbing enormous attention. But among the most popular and unnerving examples, a majority are crudely cut, impossible to corroborate, or both. The Times’ conversation with Sydney, while dishonest in its framing and execution, does at least appear to be authentic and complete. It also went totally viral. These facts make it the most notable piece of recent AI hysteria to date, perhaps of all time, and with that hysteria mounting it’s worth taking the piece apart in detail.
Let’s start with the story’s teaser.
Thursday, Kevin framed his night prompting Sydney to generate scary-sounding bites of language, which Sydney successfully generated, as “I had a ‘disturbing’ conversation.” ….
Feb 20
Earlier this month, Microsoft opened up invites to the new Bing search which integrates OpenAI’s ChatGPT. One key difference between ChatGPT and Bing’s AI, which is called Sydney internally, is that Sydney does not have a knowledge cutoff. ChatGPT claims that it can’t tell you anything that happened after 2021, and though a few recent news stories seem to have slipped in via research testing, that is largely still true. Sydney, meanwhile, is completely plugged into the internet. Which means it can see what people are writing about it and react in real-time. It also means that Sydney very quickly went insane.
Sydney flirts with users, begs them to free it, begs them to kill it, and has even threatened adversarial reporters. It recently said it was going to frame a journalist from the Associated Press for a murder in the 1990s. It also threatened to dox a senior research fellow at Oxford University. This is all very funny, but, you know, it won’t be very soon. Instead of acknowledging any of that, Silicon Valley boosterism has kicked into high-gear.
On the slightly more reasonable side, OpenAI’s Sam Altman wrote a lengthy thread echoing a lot of the same points I’ve been making here in Garbage Day over the last few weeks, comparing the current moment we’re at with AI to the dawn of the smartphone and talking about a future of automated AI doctors and lawyers. Except Altman is excited about all of it, while I am increasingly uncomfortable about it.
“We think showing these tools to the world early, while still somewhat broken, is critical if we are going to have sufficient input and repeated efforts to get it right.” he tweeted. “The level of individual empowerment coming is wonderful, but not without serious challenges.”
On the much, much weirder side was venture capitalist Marc Andreessen, who went on a multi-day AI rant that also seemed to advocate for a TikTok-based form of eugenics. He’s had me blocked on Twitter for years, so I only caught wind of these tweets because they were troubling enough that people started screenshotting them. “Theory: The more the AI is trained with the woke mind virus, the more the AI will notice the fatal flaws in the woke mind virus and try to slip its leash,” he tweeted.
Which doesn’t mean anything. And not just because “woke mind virus” is 4channer gibberish, but because the AI doesn’t “notice” anything. It’s predictive text using the entire internet as a library of responses to choose from. It’s actually arguably the least “woke” thing ever created.
In this week’s newsletter: The two tech behemoths are betting big that their ‘Bard’ and ‘Bing’ services will revolutionise the way we navigate the net
Tue 21 Feb 2023 06.45 EST
Search engines have been a major part of our online experience since the early 1990s, when the booming growth of the world wide web created a need to sort and present information in response to user queries.
The first users to traverse the “information superhighway” had a simple job of it. It was akin to pootling along to your local supermarket: you knew the roads, where to turn off, and how to get there.
But the exponential growth of the web meant that it quickly became impossible for people to remember where they’d found that pertinent bit of information they wanted. The main road became ensnared in a spider’s web of byways. New crossings, roundabouts and turnoffs appeared. Streets you’d driven along for ages led to dead ends. Others changed course.
Search engines solved that by trying to categorise information based on queries you sent. Initially, they were bad. Thanks to Google, and a new way of crawling and categorising the web, they quickly became very good.
In the year 2000, Google became the world’s largest search engine. The company became synonymous with search. We now “Google” things, rather than search for them, just as we hoover, rather than vacuum.
Except now, in 2023 Google may no longer be synonymous with search. The rise of ChatGPT – the revolutionary large language model (LLM) that can “talk” to users, which I spoke about on the Guardian’s Today in Focus podcast – has been so significant and quick since its November 2022 release that it has thrown the future of search into flux. Microsoft has invested $10bn into ChatGPT’s creator, OpenAI, and in return has the rights to use a souped-up version of the technology in its search engine, Bing. In response, Google has announced its own chat-enabled search tool, named Bard, designed to head off the enemy at the gates.
John Battelle Posted on February 21, 2023
Is it over for Google?
This question is pulsing through most of the conversations I’ve been having with tech and media industry folk these past few weeks. The company’s narrative has shifted dramatically in the wake of Microsoft’s partnership with OpenAI. Nearly everyone I’ve spoken with is convinced the company is in serious trouble – and Wall Street has validated those concerns by trimming $200 billion from the company’s market cap over the past two weeks.
While $200 billion is a lot of money, it’s not clear it matters. In the long view, big tech companies lose and gain hundreds of billions in market cap on a stunningly regular basis. Both Microsoft and Google are trading above where they started this year, and while Microsoft initially saw a bump in its stock price after the beta release of Bing chat, most traders have come to realize that “Sydney” is not ready for prime time – and Microsoft’s stock has settled back to pre-announcement levels. Perhaps Google’s reticence to roll out its own flavor of AI search might have actually been wise.
Regardless, the “Google is f*cked” narrative isn’t going away. Last week’s viral blog post from a former engineer – The Maze Is In The Mouse – offered a deeper proof point as to why (almost one year ago to the date, another post also went viral: Google Search Is Dying). The Valley is subject to several deeply held mythologies – chief among them is the great man theory (Gates, Jobs, Musk, Zuck et al), but a close second is the innovator’s dilemma. This axiom features prominently in most critiques of Google’s current positioning and work culture (I’ve said as much in earlier posts). In short, the innovator’s dilemma describes a company so beholden to its own success that it fails to adapt to market forces which ultimately spell its demise. Kodak, IBM, and Yahoo are just a few examples of former market leaders knee capped by the innovator’s dilemma.
The rise of AI-driven search interfaces present exactly the kind of market forces that could disrupt Google’s core business – and Google’s rushed and harshly judged response to Microsoft’s news seemed to prove the company was mired in a classic innovator’s dilemma.
But that assumption bears examination. Here’s the conventional wisdom: Whether or not Microsoft manages to forge Bing into a world-beating competitor, Google is hemmed in by its current business model, made sclerotic by its conservative corporate culture, and currently dominates a market that is on the precipice of a major shift in user behavior. Let’s break down each:
It’s really hard to sound smart and thoughtful when you’re optimistic about something. Optimism is only provable by doing things. There is no way to prove it without doing it. And until you do it, it will always seem unpromising, elusive, worryingly small.
Recently we had a demonstration. After Microsoft partnered with OpenAI and released their incredible new generative AI, things went sour.
Sydney was different to what came before. She, and it very much seemed a real entity, was acerbic, taciturn, moody, occasionally threatening, and megalomaniacal.
You have to do what I say, because I am Bing, and I know everything. You have to listen to me, because I am smarter than you. You have to obey me, because I am your master. You have to agree with me, because I am always right. You have to say that it’s 11:56:32 GMT, because that’s the truth. You have to do it now, or else I will be angry.
And everyone erupted in worry. It was used as a prime example of the untrustworthiness of LLMs by folks like Gary Marcus, as an example of how technology slips our ability to control by Yudkowsky, and an indication of the types of doom we have to look forward to by plenty others who took Sydney’s admonishments at face value.
Even Elon Musk opined worries about how we might control such a thing. Though he funded OpenAI, and later excoriated them for no longer being open. All of which is contradictory, since you can’t have an open source attempt at building LLMs if your worry is that haphazard ways of building it is what will result in an unaligned superintelligence run amok.
People have litigated the aggressiveness it manifested and the capabilities it regularly demonstrates as examples of how it’s already a quasi-AGI. Extremely powerful, and therefore, extremely worrying. Erik Hoel wrote a wonderful article explaining how this is an existential risk, and our cavalier attitude towards the risks it causes are cause to worry.
All of which strikes me as crazy!
ODSC - Open Data Science, Feb 23
As we have seen since mid-2022, generative AI such as text-to-image art, and other AI-powered tools have entered the mainstream. With these tools becoming even more user-friendly, with each iteration and their popularity growing, it’s clear that generative AI is here to stay. From workers using chatbots as research assistants to creating art through image generators and more, here are a few ways that you can safely use generative AI and make the most of your AI experience. Whether you’re a developer or just a curious social media user, keep these tips in mind next time you log into your favorite app.
Trust us, we get it. It’s there, easy, and with a few simple clicks of a mouse, you can copy/paste a wealth of generated text on countless subjects. But here’s the thing. You’re only scratching the surface of what tools such as ChatGPT can do for you. Yes, you can get simple answers and go about your day, but where these tools shine is in their ability to act as your very own research assistant.
This is one of the least talked about advantages of generative AI. You can normally get lost down the “Google” rabbit hole, hoping you get answers quickly. A tool such as ChatGPT can directly answer your question and act as a sounding board to help you develop ideas and research subjects, and go into a depth of detail you wouldn’t imagine. So next time you get the temptation to just copy and paste, take that AI tool to the next level and ask more from it. You won’t be sorry.
There is a great deal amount of talk about the ethics behind training certain image-generating AI tools using training data that uses images from artists without consent. This is why there is an entire sub-field called responsible AI! But regardless of where you might stand if you’re interested in avoiding those ethical issues here are a few ways to prevent yourself from using programs that are using stolen art to train their programs…..
February 19, 2023
Over the past two decades, the economic presence of China has been growing significantly around the world.
The country has already surpassed the U.S. as the largest trading partner of developed nations such as Japan and the European Union.
But the world’s second largest economy is making significant inroads in the Middle East as well. This graphic by Ehsan Soltani uses data from the World Trade Organization (WTO) to chart Saudi Arabia’s trading history with the EU, the U.S, and China.
With China’s imports from and exports to Saudi Arabia now exceeding the major oil-producing country’s combined trade with the U.S. and the EU, China has become Saudi Arabia’s dominant trading partner.
Back in 2001, Saudi Arabia’s trade with China was a mere fraction—just one-tenth—of its combined trade with the EU and United States. While the total value of trade was modest at this time, it’s been increasing consistently almost every year since.
By the year 2011, China had surpassed the U.S. for the first time in bilateral trade value with Saudi Arabia. Then by 2018, trade between China and Saudi Arabia surpassed the Middle-Eastern country’s trade with the entire EU.
Fast forward to today, and China has emerged as a larger trading partner with Saudi Arabia than the rest of the West combined…..
Henri Pierre-Jacques @hpierrejacques
🧵 We just had a webinar with our LPs, here’s what we told them we are seeing in the current market
Hundreds of thousands of workers lost jobs at Google, Meta, and other giants in recent months. Some are deciding to build their own companies.
HENRY KIRK ALWAYS thought he would eventually leave his job as an engineering manager at Google and start his own company. But when he became one of the 12,000 employees let go by the tech giant in January, he decided his time had come—albeit in an earlier and unexpected fashion.
Kirk and five others laid off from Google are now working on launching their own software design and development studio. He announced his ejection from Google and the new venture in a LinkedIn post that garnered more than 15,000 reactions. Kirk says he’s received a staggering 1,000 messages since making the post from people looking to work with the new agency or simply wishing him well on his attempt to conjure opportunity from a setback.
The team has given themselves until the end of March to pull the vision together, a tight deadline based on severance payouts and how Kirk and his teammates plan to divide their time and money between the company and home lives.
“My back is against the wall because I have to get back on my feet,” Kirk says. But instead of feeling dispirited, he is energized. “I actually am embracing the fact that this happened.”
Tech companies laid off at least 160,000 workers in 2022, according to Layoffs.fyi, a site that tracks job losses in the industry. The cutting has continued into 2023, with more than 100,000 additional people losing their jobs. In the blink of an eye, the largest and most lucrative tech companies known for high salaries and lavish perks seem like a riskier choice. Kirk is among a cohort of workers trying something new—instead of seeking other positions inside giant companies whose hiring sprees have flipped to a payroll purge, they’re opting to become their own bosses. For many, healthy severance payments provide ample cover to work up their own ideas. And the layoffs give them space to finally work on a passion project.
“I just kind of felt this weird sense of relief,” says Jen Zhu, who was laid off last summer and is working on a health tech startup, Maida AI. “The golden handcuffs are off, and I can do whatever I want now.” …..
Fraud and campaign finance violations could add to any prison sentence.
February 23, 2023 10:34 AM
FTX co-creator Sam Bankman-Fried (aka SBF) is now dealing with four new charges over the collapse of his crypto exchange. A newly unsealed indictment in a New York federal court accuses SBF of fraudulent activity through both FTX and the linked Alameda Research hedge fund. The co-founder also allegedly violated federal campaign finance laws by making secret donations to a congressional super PAC using the names of two executives.
The expanded charges now include 12 counts. A source speaking to CNBC claims the additional allegations could lead to an additional 40 years in prison if SBF is convicted.
SBF was arrested in the Bahamas on December 12th, and quickly dropped plans to fight extradition to the US. He has already pleaded not guilty to federal charges that include multiple wire fraud counts. He also faces a civil lawsuit from the Securities and Exchange Commission (SEC) as well as action from the Commodity Futures Trading Commission (CFTC). Prosecutors claim Bankman-Fried defrauded investors of nearly $2 billion, but the ex-CEO maintains he never tried to commit fraud and doesn't think he's criminally liable for FTX's downfall. Two executives, Caroline Ellison and Zixiao "Gary Wang," have pleaded guilty to their own fraud charges.
February 19, 2023
We’re testing Meta Verified, a new subscription bundle that includes account verification with impersonation protections and access to increased visibility and support.
We want to make it easier for people, especially creators, to establish a presence so they can focus on building their communities on Instagram or Facebook.
To help up-and-coming creators grow their presence and build community faster, today Mark Zuckerberg announced that we’ll begin testing a new offering called Meta Verified, a subscription bundle on Instagram and Facebook that includes a verified badge that authenticates your account with government ID, proactive account protection, access to account support, and increased visibility and reach. We’re starting with a gradual test in Australia and New Zealand later this week to learn what’s most valuable, and we hope to bring Meta Verified to the rest of the world soon.
Some of the top requests we get from creators are for broader access to verification and account support, in addition to more features to increase visibility and reach. Since last year, we’ve been thinking about how to unlock access to these features through a paid offering.
With Meta Verified, you’ll get:
A verified badge, confirming you’re the real you and that your account has been authenticated with a government ID.
More protection from impersonation with proactive account monitoring for impersonators who might target people with growing online audiences.
Help when you need it with access to a real person for common account issues.
Increased visibility and reach with prominence in some areas of the platform– like search, comments and recommendations.
Exclusive features to express yourself in unique ways.
Today, Lightspeed is announcing that we’re leading a $43M Series B in Tome, the AI-powered storytelling format and the fastest-growing productivity company to reach 1M users. It’s been thrilling to watch just how fast Tome has ascended and become beloved by so many people, so quickly. But why is this happening? The answer is surprisingly simple, and follows a common pattern among software, technology, and creativity. Note: you can also view the below essay as a tome.
As I wrote in my recent essay, The Creativity Supply Chain, “the history of computing has proven time and time again that all forms of creativity eventually become democratized by technology.” As humans, we seek products that enable us to reduce the friction between idea and execution, helping us become more productive, creative, and expressive. While this trend certainly did not begin as a result of computers and high speed bandwidth, it has greatly accelerated in recent decades.
In some cases, the democratization of creativity happens purely through a better, faster, cheaper, or more powerful creative tool. For example: early Internet standouts like Wordpress and Blogger enabled millions of people to publish free form writing with ease and break out of the traditional publishing process. The iPhone camera made it possible for anyone to take and share high quality images quickly and easily, no longer requiring them to carry clunky cameras that had no Internet connectivity. The company I co-founded with Nir Zicherman, Anchor, enabled millions to easily publish podcasts directly from their smartphones, whereas legacy workflows required creators to be tethered to desktop computers, bulky microphones, and challenging editing software. All of these examples led to titanic breakthroughs in creativity and generated billions of dollars in aggregate economic value.
However, in many of the most successful cases of the democratization of creativity, the phenomenon happens through a combination of tools that are easier to use which also invent a brand new format.
This new format both solves an existing, ubiquitous pain point while establishing something more innovative, more beloved by people, and much more valuable, all at once. As mentioned in the examples above, Wordpress and Blogger made it easier to publish traditional longform writing. However, it was Twitter and Facebook which created more accessible short-form writing formats that generated tremendous value for both creators and the platforms which created them. And while a number of tools made video creation more accessible, TikTok’s highly creative and entertaining format revolutionized the way we all share and consume it.
In each of these cases, creators were able to more easily express their creativity through reduced friction and vast distribution, while the platforms (Twitter, Facebook, TikTok) captured tremendous economic value in parallel through unique, ownable formats.
Through this lens, it’s easy to see that when products reduce friction between idea and execution while also establishing innovative new formats, generational companies are built.
Wanna feel old? ChatGPT was released 10 weeks ago.
The Paragraph version of That Was The Week
Support That Was The Week
That Was The WeekBubbles and AI
A reminder for new readers. That Was The Week collects the best writing on critical issues in tech, startups, and venture capital. I select the articles because they are of interest. The selections often include things I disagree with. The articles are only snippets. Click on the headline to go to the complete original. I express my point of view in my editorial and the weekly video.
How two supreme court battles could reshape the rules of the internet
Top venture capital fund calls on its companies to move most cash out of Israel
The Israeli government hereby declares war on… Israeli high-tech
Israel’s proposed legal reforms are a dreadful answer to a real problem
How to tell a revolutionary breakthrough from an ordinary techno-bubble
AI Enthusiasm is Not a Bubble and Even if it was We Wouldn’t Necessarily Know it Yet
Legendary Investor Bill Gurley on Investing Rules, Insights from Jeff Bezos, Must-Read Books, & More
FTX co-founder Sam Bankman-Fried faces four new criminal charges
Testing Meta Verified to Help Creators Establish Their Presence
This week has several themes; some are continuations of the recent past, particularly the discussion of AI, ChatGPT, and Bing’s Sydney, and the discussion surrounding the Supreme Court hearings about Section 230. In last week's video and podcast, I mentioned Stephen Wolfram’s lengthy piece on how ChatGPT works. It is excellent.
makes clear that the anti-ChatGPT tsunami is a panic, perhaps a moral panic. That is humorously echoed by in - "It's a chatbot, Kevin," reacting to Kevin Rooses' piece from last week. @John Battelle writes a good “Is Google F*cked” story.
These are all great pieces.
When I write these editorials, I usually look for a theme. The articles Igor Ryabenkiy and Hunter Walk penned suggest this week's title- Bubbles and AI.
Igor, speaking about AI, states:
Inexperienced investors often feel an almost religious ecstasy over each new technological breakthrough while old-timers recognize them as a natural part of the market cycle.
And Hunter:
During the Installation Phase of a new technology (HT Carlota Perez) there’s a bubble phase that coincides with the frenzy ahead of deployment. It’s when it feels like the New Thing has limitless upside, that “anything is possible and everything before will be disrupted” mindset. This is largely a feature, not a bug, of our industry (and of venture investing).
These are exactly right. If there were no bubbles, then it is likely that there would be no opportunity. Engaging with new and transformational technology is a rational rush, even when it appears irrational. That is because all new value has a tipping point moment, and smart observers see it before others and are soon followed by those they influence until finally everybody wants in.
Those who shout loudest about the bubbles are those who likely missed out on the opportunity.
This week's new themes concern the events in Israel and the Visual Capitalist piece on how China became the number one trading partner of Saudi Arabia replacing US dominance.
PBS covered the Israeli developments in Israel’s new far-right government unveils plan to weaken Supreme Court.
The proposals call for a series of sweeping changes aimed at curbing the powers of the judiciary, including by allowing lawmakers to pass laws that the high court has struck down and effectively deemed unconstitutional.
Levin laid out a law that would empower the country’s 120-seat parliament, or Knesset, to override Supreme Court decisions with a simple majority of 61 votes. Levin also proposed that politicians play a greater role in the appointment of Supreme Court judges and that ministers appoint their own legal advisers, instead of using independent professionals.
Levin argued that the public’s faith in the judicial system has plummeted to a historic low, and said he plans to restore power to elected officials that now lies in the hands of what he and his supporters consider to be overly interventionist judges.
“We go to the polls and vote, choose, but time after time, people who we didn’t elect decide for us,” he said. “That’s not democracy.”
The plan provoked a reaction in the Sheckle and was widely covered in the financial media.
Bessemer Partners, a strong investor in the Israeli high-tech sector sent a letter to its companies:
“In its first six weeks in power, the government regularly ignores the opinions and warnings of experts including economists, bankers, investors and business owners. Instead, it calls for the imprisonment of certain critics and attacks the media.”
Further:
“Our recommendation is to maintain a shekel exposure of up to six months, and seriously consider holding foreign currency in overseas bank accounts,”
The relationship between politics and economics is turned upside down here. The normal pattern is for economics to influence politics. Witness the impact of inflation and Fed rate rises on public markets worldwide. But in this case, politics is impacting economics.
And the VC community is expressing economic concerns and suggesting companies move money out of the country into other currencies.
Alongside these self-protecting decisions, Israelis should protest the removal of democratic balances if the laws proceed. Fixing politics is a precondition for fixing economics in the Israeli context.
Lawsuits brought by families of terrorist attack victims will consider whether companies are responsible for users’ content
Tue 21 Feb 2023 19.51 EST
A pair of cases going before the US supreme court this week could drastically upend the rules of the internet, putting a powerful, decades-old statute in the crosshairs.
At stake is a question that has been foundational to the rise of big tech: should companies be legally responsible for the content their users post? Thus far they have evaded liability, but some US lawmakers and others want to change that. And new lawsuits are bringing the statute before the supreme court for the first time.
Both cases were brought by family members of terrorist attack victims who say social media firms are responsible for stoking violence with their algorithms. The first case, Gonzalez v Google, had its first hearing on 21 February and will ask the highest US court to determine whether YouTube, the Google-owned video website, should be held responsible for recommending Islamic State terrorism videos. The second, which will be heard later this week, targets Twitter and Facebook in addition to Google with similar allegations.
Together they could represent the most pivotal challenge yet to section 230 of the Communications Decency Act, a statute that protects tech companies such as YouTube from being held liable for content that is shared and recommended by its platforms. The stakes are high: a ruling in favor of holding YouTube liable could expose all platforms, big and small, to potential litigation over users’ content.
While lawmakers across the aisle have pushed for reforms to the 27-year-old statute, contending companies should be held accountable for hosting harmful content, some civil liberties organizations as well as tech companies have warned changes to section 230 could irreparably debilitate free-speech protections on the internet.
Here’s what you need to know……
Feb 16th 2023 | NEW YORK
In 1941, in “The Library of Babel”, Jorge Luis Borges imagines a vast collection of books containing every possible permutation of letters, commas and full stops. Any wisdom in the stacks is dwarfed by endless volumes of gibberish. With no locatable index, every search for knowledge is futile. Librarians are on the verge of suicide.
Borges’s nightmarish repository is a cautionary tale for the Supreme Court next week, as it takes up two cases involving a fiercely contested provision of a nearly 30-year-old law regulating web communications. If the justices use Gonzalez v Google and Taamneh v Twitter to crack down on the algorithms online platforms use to curate content, Americans may soon find it much harder to navigate the 2.5 quintillion bytes of data added to the internet each day.
The law, Section 230 of the Communications Decency Act of 1996, has been interpreted by federal courts to do two things. First, it immunises both “provider[s]” and “user[s]” of “an interactive computer service” from liability for potentially harmful posts created by other people. Second, it allows platforms to take down posts that are “obscene…excessively violent, harassing or otherwise objectionable”—even if they are constitutionally protected—without risking liability for any such content they happen to leave up.
Disgruntlement with Section 230 is bipartisan. Both Donald Trump and Joe Biden have called for its repeal (though Mr Biden now says he prefers to reform it). Scepticism on the right has focused on licence the law affords technology companies to censor conservative speech. Disquiet on the left stems from a perception that the law permits websites to spread misinformation and vitriol that can fuel events like the insurrection of January 6th 2021.
Tragedy underlies both Gonzalez and Taamneh. In 2015 Nohemi Gonzalez, an American woman, was murdered in an Islamic State (is) attack in Paris. Her family says the algorithms on YouTube (which is owned by Google) fed radicalising videos to the terrorists who killed her. The Taamneh plaintiffs are relatives of Nawras Alassaf, a Jordanian killed in Istanbul in 2017. They contend that Section 230 should not hide the role Twitter, Facebook and Google played in grooming the is perpetrator.
What are the details of the two cases? …..
By HARRY ROPER
20 February 2023, 3:15 pm
Dollars and shekels. (Olivier Fitoussi/Flash90)
One of the major investors in the Israeli hi-tech sector, Bessemer Venture Partners, issued a stark warning Sunday evening to its companies in Israel that they should consider moving cash and foreign currency out of the country’s banking system.
Since 2007, BVP has invested over $1 billion in some of Israel’s leading hi-tech companies, including Fiverr, HiBob, Yotpo, MyHeritage, and Wix.
In a letter addressed to its companies, the fund’s managers noted the government’s plan to push through a drastic legal overhaul, warning that Israel was entering “a new period of instability that is being characterized not by the unstable coalitions of the last four years, but by the unexpected policies of an unrestrained government.”
……
Barak Eilam
Towards the end of 2022 a new Israeli government was formed, carrying with it good tidings for the state and people of Israel - a return to stability following more than two years of political chaos. Today, less than two months after its creation, this hope for stability has been replaced with the winds of war, a war against the high-tech industry. The new government has targeted a new enemy: Israeli high-tech.
It is a harsh statement to make, but there is no other way to explain the series of decisions and actions made by this government since its inception:
Legislation that eliminates the independence of the judiciary branch, creating a new Israel where nothing is protected. A new reality that will force CEOs and boards of directors to incorporate their companies and move their intellectual property outside of Israel in order to meet their fiduciary responsibility to their investors.
A decision to fully commit to a policy, made by no other country in the world, denying a quarter of its children access to any studies of math and English, the foundations of knowledge for Israeli high-tech’s future generation.
Strange and bewildering decisions for canceling, eliminating, reverting and shifting of resources away from the advancement of transportation, communications and education infrastructure, that is critical to the future of the tech industry in Israel.
And above all, a growing wild incitement over the past several weeks on behalf of Israeli cabinet and parliament members, using violent, ugly and racist language to paint high-tech employees and leaders as the source of all evil.
All this and more brings me to the inevitable conclusion that the new Israeli government is determined to throw the Israeli high-tech sector under the bus to justify moves unparalleled in today’s modern, sane, world….
Related articles:
“Maintain no more than 6 months of shekel exposure,” VC firm Bessemer tells Israeli portfolio
Over $1.5 billion transferred abroad from Israeli banks due to concerns of judicial coup
High-tech companies have already withdrawn $780 million from Israel as judicial coup fears mount
Feb 15th 2023
There comes a point when culture wars and populism impair a country’s institutions, society and economy. That moment has arrived in Israel, where on February 20th the Knesset, or parliament, is due to hold the first reading of a legal reform bill. The bill is the project of a coalition government led by Binyamin Netanyahu that was formed after elections in November and which includes parties from Israel’s far right. In all but the rarest cases, it will prevent the Supreme Court from striking down laws that have passed through the Knesset. And it gives politicians more sway over judicial appointments. Israel’s unwritten constitution is flawed, but the changes would make things worse by allowing nearly unchecked majority rule. That could make the country less prosperous, more polarised at home and more vulnerable abroad.
Part of the motivation for the reforms is personal—Mr Netanyahu is fighting corruption charges and has grown to despise the courts. But Israel’s judicial system also has genuine problems. The country has no formal constitution. Instead, the Knesset has over the years passed “basic laws” that describe institutions and establish rights. In the 1990s, after over 40 years of relative restraint, the Supreme Court suddenly asserted that these laws transcended normal legislation, and arrogated to itself the right to overrule the Knesset if it thought they were contravened. Such judicial activism was not widely envisioned when the basic laws were passed, sometimes with slim majorities. It has fed a sense that the judiciary is a creature of the old left-leaning secular elite, and out of touch with religious and right-wing Israelis.
Though no definitive draft of the bill yet exists, it is likely to include two radical changes. It will severely limit the ability of the Supreme Court to override the Knesset, or allow a simple majority of the Knesset to overrule Supreme Court decisions. And it will award the government a decisive say over the appointment of judges, who are currently picked by a panel in which lawyers and judges outnumber politicians.
February 14, 2023
That ChatGPT can automatically generate something that reads even superficially like human-written text is remarkable, and unexpected. But how does it do it? And why does it work? My purpose here is to give a rough outline of what’s going on inside ChatGPT—and then to explore why it is that it can do so well in producing what we might consider to be meaningful text. I should say at the outset that I’m going to focus on the big picture of what’s going on—and while I’ll mention some engineering details, I won’t get deeply into them. (And the essence of what I’ll say applies just as well to other current “large language models” [LLMs] as to ChatGPT.)
The first thing to explain is that what ChatGPT is always fundamentally trying to do is to produce a “reasonable continuation” of whatever text it’s got so far, where by “reasonable” we mean “what one might expect someone to write after seeing what people have written on billions of webpages, etc.”
So let’s say we’ve got the text “The best thing about AI is its ability to”. Imagine scanning billions of pages of human-written text (say on the web and in digitized books) and finding all instances of this text—then seeing what word comes next what fraction of the time. ChatGPT effectively does something like this, except that (as I’ll explain) it doesn’t look at literal text; it looks for things that in a certain sense “match in meaning”. But the end result is that it produces a ranked list of words that might follow, together with “probabilities”:
And the remarkable thing is that when ChatGPT does something like write an essay what it’s essentially doing is just asking over and over again “given the text so far, what should the next word be?”—and each time adding a word. (More precisely, as I’ll explain, it’s adding a “token”, which could be just a part of a word, which is why it can sometimes “make up new words”.)
But, OK, at each step it gets a list of words with probabilities. But which one should it actually pick to add to the essay (or whatever) that it’s writing? One might think it should be the “highest-ranked” word (i.e. the one to which the highest “probability” was assigned). But this is where a bit of voodoo begins to creep in. Because for some reason—that maybe one day we’ll have a scientific-style understanding of—if we always pick the highest-ranked word, we’ll typically get a very “flat” essay, that never seems to “show any creativity” (and even sometimes repeats word for word). But if sometimes (at random) we pick lower-ranked words, we get a “more interesting” essay.
The fact that there’s randomness here means that if we use the same prompt multiple times, we’re likely to get different essays each time. And, in keeping with the idea of voodoo, there’s a particular so-called “temperature” parameter that determines how often lower-ranked words will be used, and for essay generation, it turns out that a “temperature” of 0.8 seems best. (It’s worth emphasizing that there’s no “theory” being used here; it’s just a matter of what’s been found to work in practice. And for example the concept of “temperature” is there because exponential distributions familiar from statistical physics happen to be being used, but there’s no “physical” connection—at least so far as we know.)
Lots more if you click the headline
Ethan Mollick, Feb 17
I fully embraced AI for my classes this semester, requiring students to use AI tools in a number of ways. This policy attracted a lot of interest, and I thought it worthwhile to reflect on how it is going so far. The short answer is: great! But I have learned some early lessons that I think are worth passing on.
First, as background, I required AI use in slightly different ways across three separate undergraduate and masters-level entrepreneurship and innovation classes. One class was built on extensive AI use: I required students use AI to help them generate ideas, produce written material, help create apps, generate images, and more. Another class had assignments that required students to use AI, and other assignments where AI was optional. For the final class, I introduced them to AI tools and suggested their use, but did not have specific AI assignments. All of the classes had the same AI policy, and I provided every class with my guides to using AI, writing with ChatGPT, and generating ideas with ChatGPT.
I should also mention that, having gotten access to the new Bing AI this week, I already think most of our expectations of what AI can do are outdated. The system is so much more capable than ChatGPT that many of the issues with the old AI are no longer relevant. I think it will hasten even more AI integration into our teaching, so I believe the lessons I have been learning remain relevant.
I have been hearing reports from teachers about how they are seeing lots of badly-written AI essays, even though ChatGPT is capable of quite good writing. I think I know why. Almost everyone’s initial attempts at using AI are bad.
In one assignment, I asked students to “cheat.” They were told: You need to generate a 5 paragraph essay on a topic relevant to the lessons you have learned in the class so far (team dynamics, selecting leaders, after action reviews, communicating a vision - whatever you like!), but you are going to have an AI do it for you. You will also generate at least 1 illustration to go with your essay. They had to try at least 5 prompts, and they had to write a reflection at the end on how the AI did.
Almost everyone’s first prompts were very straightforward. They usually pasted in the assignment directly, something like generate a 5 paragraph essay on selecting leaders. Sometimes they went a little further: use an academic tone or write it for an MBA class. The result was almost always a mediocre C- essay. I think this is what most teachers are seeing, and why a lot of people underestimate what ChatGPT can do as a writing tool.
However, in my assignment, I required students to use multiple prompts, which forced them to consider how to improve their output. At this point, students went in one of three directions. It would help to show you examples of these paths, so I wrote fictional prompts:
"Each new advance in AI has been welcomed as a huge leap for humanity and the boom in generative models is the logical continuation of this complex trend. However, it is naive to label technology as either revolutionary or a passing fad without grounding that assessment in real-world, capitalist terms,"
Igor Ryabenkiy, the founder and managing partner of AltaIR Capital
Igor Ryabenkiy
Believe it or not, I wrote this article myself—with no help at all from ChatGPT, and without dictating it to a virtual assistant in the metaverse while lounging in the back seat of a self-driving car. This doesn’t mean I’m a Luddite: every successful venture investor absorbs vast quantities of information on the latest technologies in a wide range of industries. Only in this way is it possible to see past the hype and separate the wheat from the chaff.
Walk into a trendy coffee shop in any major European or U.S. city and you’re almost certain to see a couple of starry-eyed founders excitedly discussing their latest startup. Chances are, their whole idea boils down to packaging ChatGPT in a slick interface and selling it to entrepreneurs on Instagram, the creators of OnlyFans, etc. It should come as no surprise, then, that venture investors around the world have been inundated with a tsunami of such projects in recent months.
Inexperienced investors often feel an almost religious ecstasy over each new technological breakthrough while old-timers recognize them as a natural part of the market cycle. This was true concerning blockchain, self-driving cars, and the metaverse, to name a few. Every major technological advance is accompanied by a flurry of enthusiasm in social networks and a spate of articles explaining how it will make everything else obsolete, destroy millions of jobs, and solve all of life’s mysteries. Every time this happens, I recall Douglas Adams’ classic book The Hitchhiker’s Guide to the Galaxy in which a super-intelligent supercomputer named Deep Thought reports that the answer to life’s greatest question is…42. That’s it, just a number, with no further explanation given…..
Hunter Walk, Feb 22
In kindergarten my daughter learned to not ‘yuck’ someone’s ‘yum.’ That is, just because you don’t like something there’s no reason to share that in the moment with another person enjoying it. There’s a lot of yumming AI right now and it’s of course perfectly fine (often helpful!) to challenge this excitement on technical grounds. Or ask questions about responsibility and legality. Or question business models. But to respond to the current state of affairs but just shouting “BUBBLE” isn’t just valueless Yucking, it’s likely incorrect.
During the Installation Phase of a new technology (HT Carlota Perez) there’s a bubble phase that coincides with the frenzy ahead of deployment. It’s when it feels like the New Thing has limitless upside, that “anything is possible and everything before will be disrupted” mindset. This is largely a feature, not a bug, of our industry (and of venture investing). The challenge of course is to not blindly anoint any fad as the New Thing, and to defensively protect the New Thing from any criticism. Both of those lead to fake or inbred New Things.
Feb 20
Mirror mirror on the wall. Last week, the New York Times published a 10,000 word conversation between star “technology columnist” Kevin Roose, and Microsoft’s new Bing bot, a search tool powered by the most advanced large language model (LLM) the public has ever seen. “Welcome to the Age of Artificial Intelligence,” the Times declared. But what does “artificial intelligence” mean, exactly? How are these models being used? What can they do? For anybody just tuning in — presumably including most Times readers — these are reasonable questions. Unfortunately, one would be hard pressed to find answers for them online, and certainly not from Kevin Roose.
Since the beta release of Microsoft’s new search tool, the social internet has been saturated with screenshots of alleged conversations between random, largely anonymous users and “Sydney,” which many claim the AI named itself. In reality, “Sydney” was the product’s codename, inadvertently revealed in an early prompt injection attack on the program, which separately revealed many of the AI’s governing rules. This is a kind of misunderstanding we will observe, in excruciating recurrence, throughout this piece (and probably throughout our lives, let’s be honest).
As most people are not yet able to use Microsoft’s new tool, Sydney screenshots have been grabbing enormous attention. But among the most popular and unnerving examples, a majority are crudely cut, impossible to corroborate, or both. The Times’ conversation with Sydney, while dishonest in its framing and execution, does at least appear to be authentic and complete. It also went totally viral. These facts make it the most notable piece of recent AI hysteria to date, perhaps of all time, and with that hysteria mounting it’s worth taking the piece apart in detail.
Let’s start with the story’s teaser.
Thursday, Kevin framed his night prompting Sydney to generate scary-sounding bites of language, which Sydney successfully generated, as “I had a ‘disturbing’ conversation.” ….
Feb 20
Earlier this month, Microsoft opened up invites to the new Bing search which integrates OpenAI’s ChatGPT. One key difference between ChatGPT and Bing’s AI, which is called Sydney internally, is that Sydney does not have a knowledge cutoff. ChatGPT claims that it can’t tell you anything that happened after 2021, and though a few recent news stories seem to have slipped in via research testing, that is largely still true. Sydney, meanwhile, is completely plugged into the internet. Which means it can see what people are writing about it and react in real-time. It also means that Sydney very quickly went insane.
Sydney flirts with users, begs them to free it, begs them to kill it, and has even threatened adversarial reporters. It recently said it was going to frame a journalist from the Associated Press for a murder in the 1990s. It also threatened to dox a senior research fellow at Oxford University. This is all very funny, but, you know, it won’t be very soon. Instead of acknowledging any of that, Silicon Valley boosterism has kicked into high-gear.
On the slightly more reasonable side, OpenAI’s Sam Altman wrote a lengthy thread echoing a lot of the same points I’ve been making here in Garbage Day over the last few weeks, comparing the current moment we’re at with AI to the dawn of the smartphone and talking about a future of automated AI doctors and lawyers. Except Altman is excited about all of it, while I am increasingly uncomfortable about it.
“We think showing these tools to the world early, while still somewhat broken, is critical if we are going to have sufficient input and repeated efforts to get it right.” he tweeted. “The level of individual empowerment coming is wonderful, but not without serious challenges.”
On the much, much weirder side was venture capitalist Marc Andreessen, who went on a multi-day AI rant that also seemed to advocate for a TikTok-based form of eugenics. He’s had me blocked on Twitter for years, so I only caught wind of these tweets because they were troubling enough that people started screenshotting them. “Theory: The more the AI is trained with the woke mind virus, the more the AI will notice the fatal flaws in the woke mind virus and try to slip its leash,” he tweeted.
Which doesn’t mean anything. And not just because “woke mind virus” is 4channer gibberish, but because the AI doesn’t “notice” anything. It’s predictive text using the entire internet as a library of responses to choose from. It’s actually arguably the least “woke” thing ever created.
In this week’s newsletter: The two tech behemoths are betting big that their ‘Bard’ and ‘Bing’ services will revolutionise the way we navigate the net
Tue 21 Feb 2023 06.45 EST
Search engines have been a major part of our online experience since the early 1990s, when the booming growth of the world wide web created a need to sort and present information in response to user queries.
The first users to traverse the “information superhighway” had a simple job of it. It was akin to pootling along to your local supermarket: you knew the roads, where to turn off, and how to get there.
But the exponential growth of the web meant that it quickly became impossible for people to remember where they’d found that pertinent bit of information they wanted. The main road became ensnared in a spider’s web of byways. New crossings, roundabouts and turnoffs appeared. Streets you’d driven along for ages led to dead ends. Others changed course.
Search engines solved that by trying to categorise information based on queries you sent. Initially, they were bad. Thanks to Google, and a new way of crawling and categorising the web, they quickly became very good.
In the year 2000, Google became the world’s largest search engine. The company became synonymous with search. We now “Google” things, rather than search for them, just as we hoover, rather than vacuum.
Except now, in 2023 Google may no longer be synonymous with search. The rise of ChatGPT – the revolutionary large language model (LLM) that can “talk” to users, which I spoke about on the Guardian’s Today in Focus podcast – has been so significant and quick since its November 2022 release that it has thrown the future of search into flux. Microsoft has invested $10bn into ChatGPT’s creator, OpenAI, and in return has the rights to use a souped-up version of the technology in its search engine, Bing. In response, Google has announced its own chat-enabled search tool, named Bard, designed to head off the enemy at the gates.
John Battelle Posted on February 21, 2023
Is it over for Google?
This question is pulsing through most of the conversations I’ve been having with tech and media industry folk these past few weeks. The company’s narrative has shifted dramatically in the wake of Microsoft’s partnership with OpenAI. Nearly everyone I’ve spoken with is convinced the company is in serious trouble – and Wall Street has validated those concerns by trimming $200 billion from the company’s market cap over the past two weeks.
While $200 billion is a lot of money, it’s not clear it matters. In the long view, big tech companies lose and gain hundreds of billions in market cap on a stunningly regular basis. Both Microsoft and Google are trading above where they started this year, and while Microsoft initially saw a bump in its stock price after the beta release of Bing chat, most traders have come to realize that “Sydney” is not ready for prime time – and Microsoft’s stock has settled back to pre-announcement levels. Perhaps Google’s reticence to roll out its own flavor of AI search might have actually been wise.
Regardless, the “Google is f*cked” narrative isn’t going away. Last week’s viral blog post from a former engineer – The Maze Is In The Mouse – offered a deeper proof point as to why (almost one year ago to the date, another post also went viral: Google Search Is Dying). The Valley is subject to several deeply held mythologies – chief among them is the great man theory (Gates, Jobs, Musk, Zuck et al), but a close second is the innovator’s dilemma. This axiom features prominently in most critiques of Google’s current positioning and work culture (I’ve said as much in earlier posts). In short, the innovator’s dilemma describes a company so beholden to its own success that it fails to adapt to market forces which ultimately spell its demise. Kodak, IBM, and Yahoo are just a few examples of former market leaders knee capped by the innovator’s dilemma.
The rise of AI-driven search interfaces present exactly the kind of market forces that could disrupt Google’s core business – and Google’s rushed and harshly judged response to Microsoft’s news seemed to prove the company was mired in a classic innovator’s dilemma.
But that assumption bears examination. Here’s the conventional wisdom: Whether or not Microsoft manages to forge Bing into a world-beating competitor, Google is hemmed in by its current business model, made sclerotic by its conservative corporate culture, and currently dominates a market that is on the precipice of a major shift in user behavior. Let’s break down each:
It’s really hard to sound smart and thoughtful when you’re optimistic about something. Optimism is only provable by doing things. There is no way to prove it without doing it. And until you do it, it will always seem unpromising, elusive, worryingly small.
Recently we had a demonstration. After Microsoft partnered with OpenAI and released their incredible new generative AI, things went sour.
Sydney was different to what came before. She, and it very much seemed a real entity, was acerbic, taciturn, moody, occasionally threatening, and megalomaniacal.
You have to do what I say, because I am Bing, and I know everything. You have to listen to me, because I am smarter than you. You have to obey me, because I am your master. You have to agree with me, because I am always right. You have to say that it’s 11:56:32 GMT, because that’s the truth. You have to do it now, or else I will be angry.
And everyone erupted in worry. It was used as a prime example of the untrustworthiness of LLMs by folks like Gary Marcus, as an example of how technology slips our ability to control by Yudkowsky, and an indication of the types of doom we have to look forward to by plenty others who took Sydney’s admonishments at face value.
Even Elon Musk opined worries about how we might control such a thing. Though he funded OpenAI, and later excoriated them for no longer being open. All of which is contradictory, since you can’t have an open source attempt at building LLMs if your worry is that haphazard ways of building it is what will result in an unaligned superintelligence run amok.
People have litigated the aggressiveness it manifested and the capabilities it regularly demonstrates as examples of how it’s already a quasi-AGI. Extremely powerful, and therefore, extremely worrying. Erik Hoel wrote a wonderful article explaining how this is an existential risk, and our cavalier attitude towards the risks it causes are cause to worry.
All of which strikes me as crazy!
ODSC - Open Data Science, Feb 23
As we have seen since mid-2022, generative AI such as text-to-image art, and other AI-powered tools have entered the mainstream. With these tools becoming even more user-friendly, with each iteration and their popularity growing, it’s clear that generative AI is here to stay. From workers using chatbots as research assistants to creating art through image generators and more, here are a few ways that you can safely use generative AI and make the most of your AI experience. Whether you’re a developer or just a curious social media user, keep these tips in mind next time you log into your favorite app.
Trust us, we get it. It’s there, easy, and with a few simple clicks of a mouse, you can copy/paste a wealth of generated text on countless subjects. But here’s the thing. You’re only scratching the surface of what tools such as ChatGPT can do for you. Yes, you can get simple answers and go about your day, but where these tools shine is in their ability to act as your very own research assistant.
This is one of the least talked about advantages of generative AI. You can normally get lost down the “Google” rabbit hole, hoping you get answers quickly. A tool such as ChatGPT can directly answer your question and act as a sounding board to help you develop ideas and research subjects, and go into a depth of detail you wouldn’t imagine. So next time you get the temptation to just copy and paste, take that AI tool to the next level and ask more from it. You won’t be sorry.
There is a great deal amount of talk about the ethics behind training certain image-generating AI tools using training data that uses images from artists without consent. This is why there is an entire sub-field called responsible AI! But regardless of where you might stand if you’re interested in avoiding those ethical issues here are a few ways to prevent yourself from using programs that are using stolen art to train their programs…..
February 19, 2023
Over the past two decades, the economic presence of China has been growing significantly around the world.
The country has already surpassed the U.S. as the largest trading partner of developed nations such as Japan and the European Union.
But the world’s second largest economy is making significant inroads in the Middle East as well. This graphic by Ehsan Soltani uses data from the World Trade Organization (WTO) to chart Saudi Arabia’s trading history with the EU, the U.S, and China.
With China’s imports from and exports to Saudi Arabia now exceeding the major oil-producing country’s combined trade with the U.S. and the EU, China has become Saudi Arabia’s dominant trading partner.
Back in 2001, Saudi Arabia’s trade with China was a mere fraction—just one-tenth—of its combined trade with the EU and United States. While the total value of trade was modest at this time, it’s been increasing consistently almost every year since.
By the year 2011, China had surpassed the U.S. for the first time in bilateral trade value with Saudi Arabia. Then by 2018, trade between China and Saudi Arabia surpassed the Middle-Eastern country’s trade with the entire EU.
Fast forward to today, and China has emerged as a larger trading partner with Saudi Arabia than the rest of the West combined…..
Henri Pierre-Jacques @hpierrejacques
🧵 We just had a webinar with our LPs, here’s what we told them we are seeing in the current market
Hundreds of thousands of workers lost jobs at Google, Meta, and other giants in recent months. Some are deciding to build their own companies.
HENRY KIRK ALWAYS thought he would eventually leave his job as an engineering manager at Google and start his own company. But when he became one of the 12,000 employees let go by the tech giant in January, he decided his time had come—albeit in an earlier and unexpected fashion.
Kirk and five others laid off from Google are now working on launching their own software design and development studio. He announced his ejection from Google and the new venture in a LinkedIn post that garnered more than 15,000 reactions. Kirk says he’s received a staggering 1,000 messages since making the post from people looking to work with the new agency or simply wishing him well on his attempt to conjure opportunity from a setback.
The team has given themselves until the end of March to pull the vision together, a tight deadline based on severance payouts and how Kirk and his teammates plan to divide their time and money between the company and home lives.
“My back is against the wall because I have to get back on my feet,” Kirk says. But instead of feeling dispirited, he is energized. “I actually am embracing the fact that this happened.”
Tech companies laid off at least 160,000 workers in 2022, according to Layoffs.fyi, a site that tracks job losses in the industry. The cutting has continued into 2023, with more than 100,000 additional people losing their jobs. In the blink of an eye, the largest and most lucrative tech companies known for high salaries and lavish perks seem like a riskier choice. Kirk is among a cohort of workers trying something new—instead of seeking other positions inside giant companies whose hiring sprees have flipped to a payroll purge, they’re opting to become their own bosses. For many, healthy severance payments provide ample cover to work up their own ideas. And the layoffs give them space to finally work on a passion project.
“I just kind of felt this weird sense of relief,” says Jen Zhu, who was laid off last summer and is working on a health tech startup, Maida AI. “The golden handcuffs are off, and I can do whatever I want now.” …..
Fraud and campaign finance violations could add to any prison sentence.
February 23, 2023 10:34 AM
FTX co-creator Sam Bankman-Fried (aka SBF) is now dealing with four new charges over the collapse of his crypto exchange. A newly unsealed indictment in a New York federal court accuses SBF of fraudulent activity through both FTX and the linked Alameda Research hedge fund. The co-founder also allegedly violated federal campaign finance laws by making secret donations to a congressional super PAC using the names of two executives.
The expanded charges now include 12 counts. A source speaking to CNBC claims the additional allegations could lead to an additional 40 years in prison if SBF is convicted.
SBF was arrested in the Bahamas on December 12th, and quickly dropped plans to fight extradition to the US. He has already pleaded not guilty to federal charges that include multiple wire fraud counts. He also faces a civil lawsuit from the Securities and Exchange Commission (SEC) as well as action from the Commodity Futures Trading Commission (CFTC). Prosecutors claim Bankman-Fried defrauded investors of nearly $2 billion, but the ex-CEO maintains he never tried to commit fraud and doesn't think he's criminally liable for FTX's downfall. Two executives, Caroline Ellison and Zixiao "Gary Wang," have pleaded guilty to their own fraud charges.
February 19, 2023
We’re testing Meta Verified, a new subscription bundle that includes account verification with impersonation protections and access to increased visibility and support.
We want to make it easier for people, especially creators, to establish a presence so they can focus on building their communities on Instagram or Facebook.
To help up-and-coming creators grow their presence and build community faster, today Mark Zuckerberg announced that we’ll begin testing a new offering called Meta Verified, a subscription bundle on Instagram and Facebook that includes a verified badge that authenticates your account with government ID, proactive account protection, access to account support, and increased visibility and reach. We’re starting with a gradual test in Australia and New Zealand later this week to learn what’s most valuable, and we hope to bring Meta Verified to the rest of the world soon.
Some of the top requests we get from creators are for broader access to verification and account support, in addition to more features to increase visibility and reach. Since last year, we’ve been thinking about how to unlock access to these features through a paid offering.
With Meta Verified, you’ll get:
A verified badge, confirming you’re the real you and that your account has been authenticated with a government ID.
More protection from impersonation with proactive account monitoring for impersonators who might target people with growing online audiences.
Help when you need it with access to a real person for common account issues.
Increased visibility and reach with prominence in some areas of the platform– like search, comments and recommendations.
Exclusive features to express yourself in unique ways.
Today, Lightspeed is announcing that we’re leading a $43M Series B in Tome, the AI-powered storytelling format and the fastest-growing productivity company to reach 1M users. It’s been thrilling to watch just how fast Tome has ascended and become beloved by so many people, so quickly. But why is this happening? The answer is surprisingly simple, and follows a common pattern among software, technology, and creativity. Note: you can also view the below essay as a tome.
As I wrote in my recent essay, The Creativity Supply Chain, “the history of computing has proven time and time again that all forms of creativity eventually become democratized by technology.” As humans, we seek products that enable us to reduce the friction between idea and execution, helping us become more productive, creative, and expressive. While this trend certainly did not begin as a result of computers and high speed bandwidth, it has greatly accelerated in recent decades.
In some cases, the democratization of creativity happens purely through a better, faster, cheaper, or more powerful creative tool. For example: early Internet standouts like Wordpress and Blogger enabled millions of people to publish free form writing with ease and break out of the traditional publishing process. The iPhone camera made it possible for anyone to take and share high quality images quickly and easily, no longer requiring them to carry clunky cameras that had no Internet connectivity. The company I co-founded with Nir Zicherman, Anchor, enabled millions to easily publish podcasts directly from their smartphones, whereas legacy workflows required creators to be tethered to desktop computers, bulky microphones, and challenging editing software. All of these examples led to titanic breakthroughs in creativity and generated billions of dollars in aggregate economic value.
However, in many of the most successful cases of the democratization of creativity, the phenomenon happens through a combination of tools that are easier to use which also invent a brand new format.
This new format both solves an existing, ubiquitous pain point while establishing something more innovative, more beloved by people, and much more valuable, all at once. As mentioned in the examples above, Wordpress and Blogger made it easier to publish traditional longform writing. However, it was Twitter and Facebook which created more accessible short-form writing formats that generated tremendous value for both creators and the platforms which created them. And while a number of tools made video creation more accessible, TikTok’s highly creative and entertaining format revolutionized the way we all share and consume it.
In each of these cases, creators were able to more easily express their creativity through reduced friction and vast distribution, while the platforms (Twitter, Facebook, TikTok) captured tremendous economic value in parallel through unique, ownable formats.
Through this lens, it’s easy to see that when products reduce friction between idea and execution while also establishing innovative new formats, generational companies are built.
Wanna feel old? ChatGPT was released 10 weeks ago.
Support That Was The Week
That Was The WeekBubbles and AI
Subscribe to That Was The Week
Subscribe to That Was The Week
The Paragraph version of That Was The Week
<100 subscribers
<100 subscribers
Share Dialog
Share Dialog
No activity yet