As impressive as these capabilities are, something about this feels off to me. Instead of building agents to interact with the internet as people currently do, shouldn't we be designing agents to interact with the internet in an agent-native way?

Websites are representations of information, and they are rendered with people in mind. That's why we have big bold copy above the fold, a clear CTA, good spacing between elements, visual elements to add interest, etc.

The thing is, the same information can be easily represented as a .txt file or as a JSON or as a RSS feed or as an API.

Instead of taking a human-centered design approach to designing websites, what would it look like if we took an agent-centered approach?

What would usability and desirability even look like in this context?

Will everything be shuffled into a renewed robots.txt file?

Will everything behave like a market?

At best, I think of these examples as stepping stones to the frontier of the new B2B - Bot to Bot - economy.

I think it's neat that agents can control the browser, but it feels like a strange ouija board experience. Do we really need to show the mouse sloooooowly moving across the screen or can we fast forward to the agent negotiating and completing a purchase of a Japanese mini truck on my behalf?

What's clear is that we're designing a way for agents to interact with our version of the internet. We'll see if they will design a way for us to interact with theirs.

Welcome to the latest edition of the Future Archive. Do you know someone who would enjoy this newsletter? Forward this email to them and they can subscribe here.

Let's jump in!

1/ Max Roser writes about how difficult it is to see the absence of something.

2/ Melanie Mitchell writes about the metaphors we use for artificial intelligence.

Words matter. And the metaphors that we use to describe things shape - ever so subtly - how we perceive and relate to the things being described.

The metaphors we humans use in framing LLMs can pivotally affect not only how we interact with these systems and how much we trust them, but also how we view them scientifically, and how we apply laws to and make policy about them.

The field of AI has always leaned heavily on metaphors. AI systems are called “agents” that have “knowledge” and “goals”; LLMs are “trained” by receiving “rewards”; “learn” in a “self-supervised” manner by “reading” vast amounts of human-generated text; and “reason” using a method called chain of “thought.” These, not to mention the most central terms of the field—neural networks, machine learning, and artificial intelligence—are analogies with human abilities and characteristics that remain quite different from their machine counterparts. As far back as the 1970s, the AI researcher Drew McDermott referred to such anthropomorphic language as “wishful mnemonics”—in essence, such terminology was devised in the hope that the metaphors would eventually become reality.

Humans are, of course, prone to anthropomorphize nonhumans, including animals, corporations, and even the weather. But we are particularly vulnerable to this tendency when faced with AI systems that converse with us in fluent language, using first person pronouns, and telling us about their “feelings.”

3/ OpenAI might be considering an ads business model.

Oof.

4/ Oxford University Press' word of the year is brain rot.

Brain rot is marked by a “supposed deterioration of a person’s mental or intellectual state, especially viewed as a result of overconsumption of material (now particularly online content) considered to be trivial or unchallenging.” It has a symbiotic relationship with internet garbage, or, as shoddily made AI-generated content has been deemed, slop, some of which is created by spammers who find financial incentive in flooding social platforms. Brain rot is the symptom, not the disease: It stems from this daily avalanche of meaningless images and videos, all those little tumbling content particles that do not stir the soul.

And yet these ephemera nonetheless seep into our skulls. Slop has a way of taking up valuable space while simultaneously shortening our attention span, making it harder to do things like read books or other activities that might actually fulfill us. Brain rot doesn’t hurt; it’s dulling, numbing, something more like a steady drip. You know you have it when you have consumed but you are most certainly not filled up. And the deluge of disposable digital stuff often feels like a self-fulfilling, self-deadening prophecy: Rotting brains crave more slop.

5/ Amelia Wattenberger writes about how we might simultaenously view information at different levels of abstraction.

If you do one thing, go check out her full essay on her website.

So long, and thanks for all the fish!

Welcome to the latest edition of the Future Archive. Do you know someone who would enjoy this newsletter? Forward this email to them and they can subscribe here.

Let's jump in!

1/ Vox writes about the millions of Americans who are eligible for existing welfare, the burden of paperwork, and the missed opportunity to reduce overall poverty.

Fayyad focuses on the administrative burden that's attached to many of these welfare programs.

These barriers often look like lengthy and confusing applications that require troves of documents to prove that an applicant is indeed eligible, seemingly never-ending waitlists, work requirements, interviews, and a whole learning process to figure out which programs you ought to apply for and how. There are some programs that many would-be recipients don’t even know exist.

And according to a study by the Urban Institute, if everyone who was eligible for certain assistance programs actually received those benefits, overall poverty would decrease by 31 percent and child poverty would drop by 44 percent.

This reminds me of the work that the Gates Foundation funded earlier this year to research people's perspectives and experiences with economic mobility here in the US.

2/ L.M. Sacasas writes about the enclosure of the human psyche, likening the destruction of the commons to the limiting and extraction of the human mind in a digitized society.

Sacasas writes:

When we use any given technology, we tend to be most interested in what we will be able to do with that technology. We want to know how a tool will empower us. But we should be at least as concerned with how any tool we use shapes our perception and our experience.5 We should be especially interested in these dynamics given the degree to which our view of reality, both the reality that is before us moment by moment and the larger reality that exceeds our immediate purview, is mediated by digital media, a degree that I suspect McLuhan, far-sighted as he was, could hardly fathom in the early 1960s.

Similar to the essay by Erin Kissane that I linked to last week, Sacasas talks about how the digital world shapes the non-digital world:

The senses are the gateway to the psyche. To enclose the psyche, it would be necessary to enclose the senses first. So, in this case, the fences and hedgerows become the devices that channel, direct, and colonize our perception of the world.

3/ Continuing on this thread, Bernstein et al write about embedding societal values into social media algorithms.

They write:

Social media influences what we see and hear, what we believe, and how we act - but artificial intelligence (AI) influences social media.

They argue for connecting social science research with AI:

Fields such as sociology, political science, law, communication, public health, science and technology studies, and psychology have long developed constructs to operationalize, describe, and measure complex social phenomena. These constructs have been proven reliable through repeated study and testing. In doing so, social scientists often develop measurement scales or codebooks to promote inter-rater reliability and replicability. We observe that the precision in these codebooks and constructs is now sufficient to translate into an artificial intelligence model.

They rightfully bring up questions that other companies - notably Anthropic and Meta - have been asking recently. Who gets to decide which values are valued and should be included? How should we think about universal vs sub-group values?

They end with this clarion call:

Ultimately, we contend that many of the societal challenges facing social media platforms today are failures of our own imagination—of a lack of viable alternatives—rather than irreconcilable differences in value. Our goal, and one that we think research is best positioned to contribute, is to expand this horizon of the imagination by articulating approaches to expand the set of values in our algorithms.

Which reminds me of Mandy Brown's notes on Andrea Ritchie's Practicing New Worlds:

But Ritchie (and Boggs) both shine a light on a different way: creating images and stories of the future that help us imagine and then create alternatives to the existing systems. This is storytelling as action and as practice; storytelling that gets us moving when we might otherwise be stuck; storytelling that invites us to lift our heads up and see further afield, so that we might know in what direction to place our next step.

Which reminds me of Albert Hirschman and his philosophy of being a man of letters and a man of action.

4/ The G20 published a policy note on financial well-being.

As many of you know, I was on the Financial Services for the Poor team at the Gates Foundation from 2014 to 2022. During that time, I led our team's (sometimes reluctant) exploration of reframing financial inclusion to financial health.

I'll use this as a placeholder to write a longer note on the motivation for leading our work on financial health at some point in the future. But in the meantime, let's celebrate that we've come a long way from a small research grant to now having the G20 discussing and adopting financial health / financial well-being!

I've often said that reframing financial inclusion to financial health was my trojan horse to inspire better product and policy innovations; I'm hoping that the G20 will lead this space to be more expansive in its view of how it can support the real needs and lives of lower income individuals and communities around the world.

5/ And now, for something on AI: Prime Intellect releases Intellect-1, the first globally distributed, 10B Parameter model.

Ok, that's it! See you next week!

Here are a few things that caught my eye last week.

1/ Erin Kissane writes about the Dark Forest Theory of the internet and reminds us that we have the opportunity - imperative - to work for a better internet.

Instead of retreating, we need to build forward.

Forgive the long quote, but this entire section is worth highlighting, underlining, printing out and putting it in a place where you can't unsee it.

It’s equally important to remember that the patterns we’ve experienced on mega-platforms are not the only way to do networks but the result of specific combinations of under-thinking and malign commercial pressures—and that the currently ascendant systems are not inevitably annihilating forces, but legal and financial constructs that can be brought to heel, forcibly reconfigured, or just replaced. Keeping these basic facts in mind is oddly difficult, because there’s so much money involved, and money is a spell for blurring the truth.

But all these platforms and attendant dipshits will be replaced, eventually, and what happens next isn't guaranteed. The British East India company was a commercial atrocity factory at near-global scale; what came after it was direct colonial rule. The assumption that "Twitter but decentralized" or "Facebook but open-source and federated" will necessarily be good—rather than differently bad—is a weak one.

So the necessary counterpart to understanding that the Dark Forest Internet complex obscures the arbitrary and temporary nature of the current situation might be accepting that there is no moral arc of the world. Our systems bend toward justice when we bend them, and keep on bending them, forever.

I think our failure to remember that the mega-platforms are just intentionally extractive constructs run by brainmelted but very human weirdos is a failure of accountability, but our failure to remember that it doesn’t have to be this way is a failure not only of imagination, but of nerve.

I've tried to put into words why I am so focused on reimagining social experiences on the internet. When there are so many crises in the world, it can feel so small or silly to focus on something like this.

But, I agree with Kissane when she writes:

This all stops being an individual problem and becomes a collective one when bad products of the social internet get worse, as when platform turmoil and manipulation helps remodel the offline world in the image of the most grotesque parts of the online one. And also when previously good products of the social internet are lost, as when it becomes impossible for people to find sustaining work, learn from one another, or organize responses to the rolling crises in which we live.

The online and offline worlds are entangled. What we do in one impacts the other.

2/ Jay Springett writes about knowledge objects and the need to signal important information in text to a LLM.

I see Knowledge Objects as handcrafted artefacts—collections of dense metadata and symbols that play with the model’s context in unpredictable ways. Weird, markdown files full of material that, that when dropped into a model, produce something entirely different than expected: shifting tone, reordering context, or amplifying particular ideas.

Jay experiments plays with marking up the text with a specific symbol to give it more importance to the LLM.

Earlier this year, my friend, Shuya, talked about how an executive would ask his employees to say his name three times at certain moments in a meeting so that the Zoom summary would make sure to catch it and mark it in the summary. We were just tickled by that story and did the same thing in our meetings going forward. Jay describes a similar pattern here.

We need a way to start signposting important information in the text – like using a highlighter pen – to signal to the machine which parts of the text we as authors consider significant. In this sense, a standard machine-readable markup, or machine-readable markdown (MRM?), is a logical next step. 

We can Enchant Knowledge Objects (books, websites, PDFs, whatever) with power—guiding the AI’s focus within documents through structured annotations and symbols and turn them into powerful Talismans. 

Earlier this week I experimented marking sections of an 18k word essay I’m in the middle of for work. I used the Unicode symbol (**ꙮ**) as the many eye glyph literally means many-eyed seraphim, and appealed.

If one of the reasons to write is to build a dataset for the future, then it makes sense to annotate and emphasize different pieces of that writing.

Also, after checking out his website, I think that I, too, may have aphantasia! Wild.

3/ The New York Times features OpenAI's first artist in residence, Alexander Reben.

It's a nice feature that showcases how Reben used AI in his craft.

Toward the end of the residency, he focused on a prototype system that turned photos of real objects into A.I.-generated images, poems and even short, satirical blurbs.

His setup consisted of his phone, a Fujifilm Instax photo printer and another printer that spit out receipts and labels. A web browser-based system combined Mr. Reben’s code with a version of the large language model that powers ChatGPT.

The conceptual camera that he built reminds me a lot of Poetry Camera, a project by our IDEO CoLab Ventures residents: Kelin Zhang and Ryan Mather.

4/ Benedict Evans published his annual report on technology.

It's a great overview on the big questions that the industry is wrestling with to date: how far will this scale? How is this useful? How do we deploy this?

Whenever I talk to friends who aren't chronic early adopters, I am still surprised by how so many of them don't use AI or don't see a meaningful "use case" for it. But when you zoom out, maybe this shouldn't be too surprising given where we are in the overall hype cycle.

And speaking of use cases, I often find the exercise of looking for use cases somewhat of a head scratcher. Benedict captures this moment well: it does feel like asking someone to imagine what the internet could be useful for back in 1995.

The presentation is long, but it's definitely worth a read. Here are some select slides that I thought were particularly useful:

On "errors"

Role of LLMs

5/ More from John Maeda on navigating a hybrid career.

A student asked me about navigating the world without a traditional credential in any single discipline. Here’s what I said: credentials open doors, but they’re not enough. Being a maker is what makes people listen. When I left MIT and entered design, people doubted me. When I went to Silicon Valley, they doubted me again. This is a recurring theme in my life: being underestimated — and then over-delivering through persistence.

Ok, that's it! See you next week!

Elan Ullendorff calls this the "Doc Web" and offers a set of axioms.

Axiom 6: The cheaper and easier a publishing tool is to use, the more ripe it is for challenging power hierarchies

A tool that requires more technical literacy, time, or money becomes inherently inaccessible to many with less power, and its form will appropriately and intentionally project class, prestige, and rarity. For that reason, a feature-rich custom website designed and built by a web development agency can only have so much radical potential. A tool that is free, and whose interface is a common word processor immediately understandable by most creators and consumers, will often incorporate fringe ideas and language in a way that these websites do not.

This is one of the reasons why I'm encouraged by the development of creative tools that make it easier for more people to build and express their ideas. Robin Sloan writes that an app can be a home-cooked meal as he describes his journey building a small app for his family.

Building this app, I figured it out:

I am the programming equivalent of a home cook.

The exhortation “learn to code” has its foundations in market value. “Learn to code” is suggested as a way up, a way out. “Learn to code” offers economic leverage, professional transformation. “Learn to code” goes on your resume.

But let’s substitute a different phrase: “learn to cook”. People don’t only learn to cook so they can become chefs. Some do! But many more people learn to cook so they can eat better, or more affordably. Because they want to carry on a tradition. Sometimes they learn because they’re bored! Or even because they enjoy spending time with the person who’s teaching them.

With Claude, ChatGPT, Gemini, and friends, more people now have the tooling to cook up something wonderful.

Matt Webb writes about his experience building an iOS app that points to the center of the galaxy. Like Matt, I'm not a developer, but I did build a functioning iOS app that I call "Walk and Talk" the other week. It's a simple app on your phone that lets you record a voice note, and then gives you a nice summary along with a cleaned up transcript. It's silly. It's simple. But it did scratch a particular itch.

If you haven't built an app before - why not go for it?

2/ Kate Brown, MIT professor, talks about the role of community gardens in combatting climate change.

Journalists and scholars have worked hard in the last two decades to get people to understand the scope and the scale and the verisimilitude of climate change. And that’s great, but some of these catastrophic stories we tell don’t make people feel very safe or secure. They have a paralyzing effect on us. Climate change is one of many problems that are too big for any one person to tackle, or any one entity, whether it’s a huge nation like the United States or an international body like the U.N.

So I thought I would start to work on something that is very small scale that puts action in the hands of just regular people to try to tell a more hopeful story. I am finishing a new book about working-class people who got pushed off their farms in the 19th century, and ended up in mega cities like London, Berlin, Amsterdam, and Washington D.C., find land on the periphery of the cities. They start digging, growing their own food, cooperating together. They basically recreated forms of the commons in cities. And in so doing, they generate the most productive agriculture in recorded history.

I love her description of these gardens as a coral reef.

You can think of a tiny city garden like a coral reef, where one little worm comes and builds its cave. And then another one attaches itself to the first, and so on. Pretty soon you have a great coral reef with a platform to support hundreds of different species — a rich biodiversity. Tiny gardens work that way in cities, which is one reason cities are now surprising hotspots of biodiversity.

3/ Scientists at UC Berkeley have developed a powder that can suck CO2 out of the air. Don't give up.

4/ As expected, there's another Twitter / X exodus. This time, people are signing up for Bluesky.

According to the New York Times, Bluesky gained over one million new subscribers in the week since the election. They now have ~15 million users. In contrast, Threads has 275 million monthly active users, Twitter has over 500 million monthly active users, and Facebook and Instagram each have over 2 billion monthly active users.

I'm encouraged by many of the small, but meaningful innovations that Bluesky has popularized. However, this is still a Twitter clone. We're still just playing around with variations on a theme and celebrating who is part of the conversation in these different platforms. There's a much bigger design space around reimagining what social networks / platforms / protocols / experiences actually look and feel like.

5/ John Maeda talks about what it means to embrace the hybrid path.

Being a hybrid is about exploring. It’s less about being an expert in any one area and more about staying curious across fields. A hybrid career invites us to break out of silos and think widely. In today’s world, the intersections are where exciting things happen — creativity grows in these spaces where we can see one field through the lens of another. But that scope requires a kind of energy and resilience that isn’t always easy to sustain. This path isn’t about comfort; it’s about finding new possibilities beyond what we know.

My friend, Mel, shared with me materials from a workshop she gave years ago on navigating multifaceted careers.

Both John and Mel have helped me see my own path in a warmer light.

Here's to a life of curiosity, creativity, connection, consideration, and celebration.