“In an effort to avert an election contest following a year of distractions and disappointing performance, we hope you join us in encouraging the Board to pursue a viable compromise with Trian Fund Management, L.P. and Nelson Peltz,” Ancora wrote in the letter. “Mr. Peltz (or a qualified designee) would make a fantastic addition to Disney’s Board.”

Ancora also suggested that much of Disney’s difficulties in recent years — including streaming losses and and several box office flops — could be pinned on the company board.

“A degree of shareholder-driven change is certainly warranted in Disney’s boardroom following an extended period of absentminded governance, ineffective succession planning, polarizing actions and sustained value destruction,” Ancora said Tuesday. “While it has been argued that challenges largely stem from the tenure of Bob Chapek, the Board was in the driver’s seat before, during and after that time.”

Disney fired back at Trian last week, suggesting that the move was fueled by a personal grudge against Disney CEO Bob Iger held by Peltz ally and former Marvel boss Ike Perlmutter. Trian oversees about $3 billion in Disney stock, but the overwhelming bulk of the shares is owned by Perlmutter, whom Disney laid off earlier this year. Trian is seeking more than two seats on Disney’s board, which is populated by directors seen as loyal to Iger.

Ancora’s announcement Tuesday didn’t disclose the size of its stake in Disney. Ancora owned more than 60,000 shares of Disney as of September, according to FactSet. That would be equivalent to an approximately $6 million stake as of Tuesday.

Disney had a market cap of about $160 billion as of Tuesday, its shares closing down by more than 1%. The stock is up more than 4% this year, underperforming the broader S&P 500.

According to US financial media outlet Business Insider, the combined market capitalization of these seven tech stocks soared by 60 percent, or $4.1 trillion, to $11 trillion this year. That's almost three times the size of the German economy. By the end of 2022, the German economy will be worth slightly more than $4 trillion, according to the World Bank.

Amazon is the most popular stock on Wall Street, with 91% of analysts rating its stock a "buy" or equivalent. The average buy rating for Standard & Poor's 500 stocks is about 55%.

Is the Current Extreme Heat for Generative AI a Bubble? Vasi Philomin, global vice president of generative AI at Amazon Cloud Technologies, responded to Punch (www.thepaper.cn), "I don't think we can deny that the internet bubble ever existed, but it did change everyone's life, and we can't imagine life without the internet now . I believe generative AI will change every profession, every business in every industry. It will take time."

Philomine asked the question that the reporter wanted to ask: Are we now in a Hype Cycle? And he answered bluntly: "Yes. If you look at the chart of the technology maturity curve (also known as the 'Hype Cycle') published by the consulting firm Gartner, which I think you should be familiar with, we probably haven't peaked yet, and there's definitely more hype to come. But in the meantime, a lot of people are quietly building useful things behind the scenes, and they're implementing real business use cases behind the scenes."

When all the hype dies down and the technology maturity curve starts to level out and take off again, the companies that will survive will be those who want to think in business terms and enable all their customers to build applications for real, and that's where they want to go, according to Philomine.

According to the latest data released by Gartner, the global infrastructure services market grew by 29.7% in 2022, totaling $120.3 billion. Amazon continues to lead the global IaaS (Infrastructure as a Service) market with $48.1 billion in revenue and 40% market share, while Microsoft comes in second with 21.5% share. In fact, because of Microsoft's strategic investment in OpenAI years ago and its active involvement in generative AI, the challenge posed by the second place (Microsoft) to the first place (Amazon) has been a hot topic.

"As the excitement around generative AI has been focused on Microsoft and Google, many investors question whether AWS is at a disadvantage," according to Wall Street analyst Rob Sanderson. "We think more investors will soon recognize that AI will drive meaningful expansion of demand for the cloud, and AWS will be a major player."

But not all analysts are bullish on Amazon's cloud growth, Itau BBA analyst Thiago Alves Kapulskis wrote in a research note following Amazon's first-quarter earnings report, adding that competition between Amazon and Microsoft's cloud services is a concern. "With so much uncertainty surrounding Amazon and significant differences between Azure (Microsoft Cloud) and AWS, we believe the market will favor the former and its parent company over the latter." The analyst wrote.

Philomine didn't shy away from the question during the interview. In response to a reporter's reference to Microsoft's latest quarterly earnings release, he mentioned that a core issue lies in the source of revenue, and that there are differences in the two companies' business models from that perspective - Microsoft's core business is likely to be productivity-boosting office products, whereas AWS is a B2B (business-to business) company that provides cloud services to businesses.

"Our focus is on the enterprise, and we provide not productivity apps, but tools that our customers can apply to their own business workflows." Philomine said the business models are not the same, "If people use AWS services, it's pay-per-call. It's a very traditional AWS approach where you only pay for the services you use. It's not the same as the office suite or the Microsoft way."

Apple GPT is an AI tool developed by Apple based on its self-developed Ajax framework, and is currently undergoing a small-scale internal test within Apple; Huawei, on the other hand, demonstrated its end-side AI big model at the application level through the intelligent assistant "Xiaoyi" of HarmonyOS 4, which was released on the afternoon of August 4 at the HDC 2023 developer conference. Huawei, on the other hand, demonstrated the end-side capabilities of AI big models at the application level by releasing the intelligent assistant "Xiao Yi" of HarmonyOS 4 at the HDC 2023 developer conference on August 4 afternoon.

Previously, Meta, OpenAI, Qualcomm, Google, Tencent and Baidu have launched or will soon launch applications or technical frameworks that support end-side AI models, and are rapidly outlining the technical and market prospects of this new direction.

HarmonyOS 4: What are the end-side AI big model capabilities

On August 4, Huawei officially released HarmonyOS 4. Huawei HarmonyOS is a full-scene distributed operating system for the Internet of Everything, supporting a wide range of terminal devices such as cell phones, tablets, smart wearables, and smart screens.

In the fourth version of this famous IoT OS from Huawei, the ability to run end-side AI big models (i.e., running big models on the phone side) becomes the key core.

"Today we have entered the era of big models, and Huawei Pangu big models will help the Hongmeng ecosystem." Yu Chengdong, Huawei's Managing Director, CEO of Terminal BG, and CEO of Smart Car Solutions BU, said, "With Pangu's underlying capabilities, Huawei will bring a new transformation of AI experience for users with smart terminal interactions, high-level productivity efficiency, and personalized services."

Text generation based on AI big models is part of HarmonyOS 4. With the growing development of the big model, Xiaoyi receives a new upgrade that enhances the capabilities in three areas: interaction, productivity, and personalized service.

Text generation and summarization capability, which is a standard capability for all the big model apps launched, HarmonyOS 4 is no exception. With Xiaoyi, content and text in images can be recognized, text content can be read aloud and more services can be accessed.

According to Ziad Asghar, senior vice president of product management and head of AI at Qualcomm, big models will rapidly reshape human-computer interaction.

The interaction changes in HarmonyOS 4 don't look significant. Based on voice interaction, it expands on multiple forms of input such as text, images and documents. For example, by communicating with the AI in a natural way with everyday speech, Little Art can automatically complete specified tasks. This has long been achievable in the iPhone through Siri.

However, as the first intelligent assistant with AI big model capability, Xiaoyi has a deeper understanding of natural semantics.

For example, Siri, the AI assistant of iPhone, requires clear and precise semantics of voice commands input by C-support users, while HarmonyOS 4 can understand voice commands with relatively less obvious meanings, and thus smartphones equipped with HarmonyOS 4 can complete tasks more accurately and quickly.

In addition, it can also access services faster than before due to the large-model AI generalization capability that Xiaoyi has and the fact that the phone's intelligent assistant connects to a wide range of services and special scenarios.

For example, if an event poster image is received and the C-end user gives instructions to Little Art, the AI can automatically recognize the address on the image and give a navigation button or save the poster phone number as a contact method.

If browsing an English news article with thousands of words, Xiaoyi can directly speed read the article, do the translation and complete the summary, and can also answer questions related to the news article.

HarmonyOS 4 also has generative capabilities. For example, it can automatically generate all kinds of business e-mail content or generate images; it can also use the personalized photos saved on the terminal, and Xiao Yi can call the AI drawing function to generate various styles of pictures.

As C-side users and Xiaoyi continue to interact, Xiaoyi will continue to improve its AI capabilities. These interaction data will be kept on the end-side to protect user privacy.

Personalization is a feature of smartphone applications. Xiaoyi has the memory ability to understand the "owner" more and more with the use of a long time, to give more intimate advice: for example, travel, before the trip to provide complete and rich travel information, and after the landing to provide local information, and according to the user's habits, to achieve personalized recommendations.

According to Huawei, small art of these new capabilities, will be opened in late August public test experience.

This is not Huawei's first attempt to access AI big models to mobile terminals. In March this year, Huawei released the P60 cell phone, which comes with a smart search function. This function is based on multimodal large model technology, which realizes the operation of large models on the cell phone side by doing miniaturization of the model on the cell phone side.

Apple moves slowly, Qualcomm is enthusiastic

Smart terminal vendors aiming at end-side AI big model capabilities are not only Huawei, but also Apple.

Apple is secretly developing GPT software called Apple GPT, based on Apple's self-developed Ajax framework development.

However, at present, Apple GPT lacks more details, and it is difficult for the outside world to know what the highlights of its technology or application capabilities are. In this regard, Apple made it clear that it has not yet decided on the future direction of Apple GPT development.

Previously, in the second fiscal quarter earnings call, Apple CEO Cook admitted that Al has great potential, but how to use AI technology still requires in-depth thinking. Cook said that Apple has integrated AI technology into its products and services, and will continue this process in the future.

The world's first intelligent assistant for the consumer-grade market, large-scale landing, the application of NLP technology, iPhone Siri, is 12 years old this year. With the new upgrade of Huawei Xiaoyi, millet Xiaoyi, Baidu Xiaodu a number of "juniors" compared to Siri is not only not intelligent, but also seems a bit "retarded".

The outside world that Apple landed GPT ability carrier is likely to be the presentation of the "twilight" state of Siri, but this has not been confirmed by Apple.

Nevertheless, the signs of Apple's entry into the end side of the AI model are very many. For example, in January this year, Apple launched a new program for Apple Books to add digital narration function, in order to automatically generate high-quality AI narration audio from written text; in the iOS17 update, Apple used the Transformer language model to improve the prediction and spelling correction function of the input method.

Also, the new AirPods Pro uses machine learning (ML) to implement an adaptive audio mode that automatically adjusts the volume by recognizing the external environment; iPadOS 17 uses a machine learning model to enable recognition of PDF fields; and Vision Pro uses the machine learning technology "encoder-decoder" Neural Networks to build digital bodies for users.

Apple how to layout AI track, this is a system of big problems, it is difficult to say in hundreds of words to understand. But the Siri development team of technical engineers to change the Siri "stupid" seems to lack of real motivation, behind the reflection of the reality of "organizational dysfunction and lack of ambition", is likely to affect the efficiency of Apple on the end side of the ground AI big model.

But Apple's inefficiency can not change the scale of deployment of end-side AI models in smart terminals, which is an indisputable fact. Because Qualcomm, Meta, OpenAI, Google, Amazon, Tencent and Baidu are all committed to realizing the lightweight deployment of AI models in mobile terminals.

The OpenAI team is launching a mobile ChatGPT app covering iOS/Android in the fourth week of July; Meta will work with Qualcomm to make its open source big model Llama 2 capable of running on flagship smartphones and PCs from 2024; Qualcomm says it is migrating more generative AI use cases to the edge side, and now AI models with more than one billion parameters are capable of running on run on cell phones with similar levels of performance and accuracy to the cloud.

Google, Tencent, Baidu and others have tightly integrated model compression technology with mobile model deployment frameworks/tools.

Among them, Qualcomm is enthusiastic about it. Qualcomm CEO Cristiano Amon emphasized in a statement that Qualcomm's ability to run AI models on cell phones rather than cloud servers provides Qualcomm with the opportunity to achieve an "inflection point" that will drive future growth.

"In short, we are uniquely positioned to help shape and capitalize on the upcoming Gen AI opportunity on devices." Amon said.

It remains uncertain when the overall smartphone market decline will stop; however, this wave of many B-side players ushered in by the end-side AI mega-modeling track is where new hope lies to reshape the application and market landscape of this dwindling consumer electronics category.

Prior to the acquisition, Lightyear Beyond had total net cash of approximately $285 million.

Light Years Beyond is an AGI (General Artificial Intelligence) company founded by Huewen Wang, co-founder of Meituan. In February, after ChatGPT ignited the big model market, the retired Huewen Wang announced that he was stepping out to groove into the big model.

Light Years Away is one of the most talked about big model startups in the Chinese AI market. But over the weekend, the soul of Lightyear Beyond, Huiwen Wang, was revealed to be hospitalized due to illness, and thus withdrew from the day-to-day management of Lightyear Beyond. This not only surprised many practitioners, but also raised concerns about the future of Lightyear Beyond.

Just less than a week later, Lightyear Beyond has found its latest home. This is perhaps the best ending for both Wang Huiwen and Lightyear Beyond.

And for Meituan, it will leap into a very competitive force in the wave of big domestic models after bringing a gradually assembled team of mature AI talent under its wing.

1. Meituan wholly acquired, Wang Huiwen exited

Prior to this, there was very little information disclosed outside of Lightyear, except for the "heroic post" from Wang Huiwen.

The announcement of Meituan disclosed the equity structure of Lightyear.

Qimai is an investment holding company wholly owned by Wang Xing, chairman and CEO of Meituan, with a 0.43% stake, Sequoia China with a 2.44% stake, and other investors with a total stake of 20.41%.

The identity of the other investors was not disclosed in the announcement. However, on June 5, Lightyear Beyond was revealed to have received the latest round of financing of $230 million, in which Source Capital led the investment of $80 million, Five Sources Capital invested $40 million, Suhua invested $40 million in his personal capacity, and Tencent bet $50 million, but the amount of financing was denied by Wang Huiwen.

According to information from the Meituan announcement, the acquisition agreement contains two main parts.

The first is an offshore share redemption agreement. Meituan's cash acquisition is primarily for payment to investors outside of Lightyear. Specifically, it includes zero consideration to AI Age, $5 million to Qimai, $28 million to Sequoia China, and $200.673 million to other investors.

The second is an inbound equity transfer agreement. According to the agreement, Huiwen Wang will sell 100% of the equity interest without any liens. The main asset outside of the inbound Lightyear is its equity interest in First Class Technology, which it acquired in April 2023.

For this, Meituan is required to pay a nominal amount of RMB 1 to Wang Huiwen and assume the debt outside of Lightyear, totaling RMB 366.92 million.

After the acquisition, Meituan owns 100% of Lightyear Beyond and Wang Huiwen no longer retains any rights.

1. Reasons for Meituan's acquisition

In the announcement, Meituan did not mention the personal physical reasons of Wang Huiwen, which was speculated.

On the afternoon of June 25, after the media exposed that Wang Huiwen was hospitalized due to illness, Lai Binqiang, the former HRD of Meituan, said in his circle of friends that he was accompanying Wang Huiwen in the hospital and said that please rest assured that he would need to follow the medical advice for the next period of time to recuperate and recover, and that everything would be fine.

Lai Binqiang is one of the founding team members of Meituan, is also the first HRD (human resources director) of Meituan, or Wang Xing's high school classmates, he and Wang Huiwen, Wang Xing three personal friendship is very deep. Later that day, Meituan also announced on the Hong Kong Stock Exchange that Wang Huiwen had applied to resign as a director of Meituan for "personal health reasons". Previously, Wang Huiwen had been transferred from an executive director to a non-executive director of Meituan.

On "instant", Wang Huiwen's dynamic stays in a job posting released on May 5, which mentions that the company is recruiting crawlers, data, front-end, back-end, product interns.

As for the reason for the acquisition, Meituan firstly affirmed in the announcement that Beyond Lightyear's position in the field of AGI in China and its management and technical team have high level experience in developing deep learning frameworks.

By acquiring leading AGI technology and talent, Meituan has the opportunity to strengthen its competitiveness in the fast-growing artificial intelligence industry and better fulfill its mission of "helping people eat better and live better".

The acquisition of Light Years Away is not Meituan's first attempt at a large model. Previously, Leopard Change reported that Meituan's big model was almost simultaneous with Wang Xing's investment in Wang Huiwen's company.

The S-team, the highest decision-making body headed by Wang Xing, is extremely concerned about the big model within Meituan. Wang Xing would ask the head of the algorithm team about the progress of the big model about every week or two. The company has been expanding its algorithm team and has started planning for a separate "platform department" to help the group's big model get off the ground in a concrete commercial form. on the afternoon of May 18, the group held a big model technology sharing session with almost all of the internal decision-making body, the S-team, participating.

Large manufacturers have an inherent advantage in doing big models. Linear Capital senior director Bai Zerin told "Azi Lightyear", "The big factories have their own data and scenarios, and for startups in the base layer of the big model, in addition to the engineering landing capability, they need to accumulate enough high-quality data and applicable scenarios to compete with the big factories."

After the completion of this acquisition, it is still unknown how the big model team and light years away will divide the work again within Meituan, but at this moment for Meituan, the road of big model is just like a tiger with wings.

1. Light years beyond the starting gun of the big model venture

Wang Huiwen's big model adventure journey began with a dinner party on February 10, 2023.

At that time, he had a dinner date with Zhifei Li, the founder of Go out Ask, and two partners of Zenith, Yusen Dai and Yuan Liu.

At the dinner table, talking about this wave of AI, Wang Huiwen directly invited Li Zhifei to join hands in the big model. Li Zhifei gave Wang Huifei an account about the basic investment of AI big model: $20 million for computing power, $20 million for people, and $10 million for data - later the widely circulated "$50 million" was calculated in this way. This is how the much-talked-about "$50 million" was calculated.

When talking about the rise, Wang Huiwen picked up the phone, published that "AI declaration" - "my AI declaration: 50 million dollars, with capital to join the group, do not care about the position, salary and title, seeking team. "

Two days later, Wang Huiwen released the idea of creating a large model company beyond light years: at 9:10 a.m. on February 13, Wang Huiwen posted a recruitment post on the AI Hero List immediately, saying that he wanted to "build China OpenAI".

According to Time Finance, Baidu has recently adjusted the organizational structure of Intelligent Driving Group (IDG), and its Intelligent Transportation Division (ACE) has been subsumed into Intelligent Cloud Group (ACG). After the adjustment, IDG's business segments have shrunk from the original three segments of autonomous driving, intelligent vehicles and intelligent transportation to only the first two.

An informed source close to Baidu Cloud confirmed the authenticity of the above adjustment to Box Lunch Finance, and said that all the existing 1,000-plus people of ACE will be merged into ACG.

The main reason for this adjustment is that the intelligent transportation business, which provides vehicle-road collaboration solutions, is not developing as expected. In January this year, news was spread externally that ACE's full performance was hit with the lowest grade, "basically everyone didn't get any bonus, and ACE's revenue of several billion yuan last year was not up to standard overall." The above-mentioned people familiar with the matter said that

ChatGPT before the fire, heavy bet on self-driving Baidu, in order to support ACE, once the business involving intelligent transportation under ACG and to ACE, "ACE can sell things, the cloud can not sell." said the above-mentioned people familiar with the matter.

Right now, ACE, which is under IDG, is slow to make a big breakthrough, so Baidu has a new plan to integrate it into its cloud business. In the view of Wu Kewei, CEO of the vehicle-road collaboration startup, this is not a good thing, "so that two teams in one company don't have to do the same thing."

A month before Baidu shrunk its self-driving business, Ali made a major adjustment to its Dharma Institute's self-driving team of about 300 people in mid-May, in which the self-driving lab was merged into the Cainiao Group, along with staff optimization. There were rumors that the scale of the layoffs was as high as 70 percent, a ratio that Ali promptly denied.

Baidu, Ali outside, other technology majors in the field of autonomous driving is also gradually losing the will to explore: in January this year, the United States Mission unmanned vehicle department rumored layoff plans, the outflow of a WeChat screenshot shows that it is suspected that the United States Mission is located in Beijing Shunyi unmanned vehicle department for layoffs, the authenticity of this news has not been officially confirmed; and Ali with the same in 2018 around the field of autonomous driving Tencent, although The company has not been involved in the layoffs, but has long been invisible from the public eye in recent years, and has almost never revealed any relevant developments to the outside world.

Radical industry investor Jiang Yi even directly defined autonomous driving as a pseudo proposition. "It has prerequisites, that is, all people within a traffic environment, have to use autonomous driving to work. From this basic point of view, autonomous driving can only be driven forward by the rule makers of public utilities, and it is simply unrealistic to rely on the power of one company alone."

ChatGPT detonated the new wind of AI big model, when Nvidia founder Huang Renxun shouted that "AI's iPhone moment" is coming, all products are worth redoing with big model consensus, soon reached within the major technology manufacturers.

Even Musk, who has been preaching self-driving, has recently stated that the impact of general artificial intelligence is much greater than self-driving.

The new productivity changes brought about by the big AI models have likewise stirred the ambitions of the domestic tech moguls. Zhang Yong, who made his first public appearance after stepping down as Ali's CEO, shouted to the public, "I believe more than ever that the development of core technologies such as cloud computing, big data and artificial intelligence will bring huge changes to society with great strategic significance, which is an important reason for my decision to devote myself to the Ali Cloud Intelligence Group."

Robin Li has also continued to externally proclaim the new blueprint of the big model reshaping the digital world.

Compared to the sci-fi-like autonomous driving, in Jiang Yi's opinion, "the concept of big model is a bit like the concept of whether there are aliens in the universe, and the former is expected to be realized with great probability."

It's not just the tech majors that are involved in the AI big model windfall. According to the data of "China AI Big Model Map Research Report", as of May 28, at least 79 basic big models above the scale of 1 billion domestic parameters have been released.

This fast-hyped battle for big models has begun to show the first signs of an approaching bubble. However, in the view of Qiji Chuangtan CEO Lu Qi, the current AI big model era is still in the early stage, and "after high growth, the bubble burst will be an inevitable phenomenon."

01

The tech majors, which are moving from expansion to contraction, are having trouble juggling multiple windfalls and investing a steady stream of R&D funds as they once did.

From 2022 onwards, cost reduction and efficiency gains will be a common choice for the major technology companies, with shrinking or shutting down loss-making businesses bearing the brunt.

Under the strategic guidance of Ma Huateng's "many businesses should be cut, don't stay", Tencent has cut more than 10 long-term loss-making products with low strategic value in the past year, including Watchpoint App, e-commerce platform Xiaoguo Pianping, digital collectibles collection platform Phantom Core, Tencent Wi-Fi Manager, Tencent Map PC version, Sogou Search App, etc.

Half a month after Ma Huateng's speech, Robin Li also said in an open letter that the future development of business "can not only look at revenue, to look at profits, but also to look at the input-output ratio".

Compared with the management costs and efforts paid by the company, non-core business spread too much, is being seen as a not cost-effective business. When the new AI big model wind comes, once the core business of self-driving also had to stand aside.

Moreover, general AI is even a game that may burn more money than autopilot. In the face of the same need for heavy investment, the same need to adhere to the long-termism of autonomous driving and big models, technology makers and then want to fish and bear paw both, obviously a less realistic thing.

Open AI, which is in full swing, has burned up more than $1 billion in training costs during its eight years of existence, and is still in a high loss state that requires blood transfusion from the giant Microsoft.

With the increase in the number of users, OpenAI's losses have also risen. According to media reports, OpenAI's losses will double to about $540 million in 2022. In order to develop sufficiently advanced general-purpose AI while keeping the company afloat, OpenAI CEO Sam Altman said OpenAI may need to try to raise as much as $100 billion in the next few years.

Another point worth noting is that, despite the same commercialization challenges, the effect of AI big models on big tech companies is undoubtedly more direct and considerable compared to autonomous driving. As Jiang Yi said, the probability of achieving a big model is much greater than that of autonomous driving.

02

The troubled self-driving industry is also losing its appeal to the tech majors.

Around 2015, almost all the world's major technology companies released plans for autonomous driving, with declarations that "humanity will soon enter the era of fully autonomous driving". Google has Waymo, Baidu has Apollo, Amazon, Ali, Jingdong and Meituan have their eyes on unmanned delivery vehicles, and Uber, Lyft, Drip and other online car companies are looking at the unlimited future brought by Robotaxi (self-driving cabs).

By 2018, global technology majors have even reached a new consensus in the field of autonomous driving, that is, to skip the L2/L3 stage and directly develop L4 and above higher-order autonomous driving capabilities without drivers.

But the harsh reality soon made the tech majors recognize the difficulty of landing autonomous driving.

Waymo, once the global leader in autonomous driving, suffered a plunge in valuation due to delays in commercialization that did not meet expectations.

In 2018, Morgan Stanley valued Waymo at $175 billion, making it the third-largest unlisted company in the world after BytePulse and Space X. By 2020, after Waymo launched a new round of $2.25 billion in funding, Waymo's valuation at this time was only $30 billion, a plunge of more than 80%.

Waymo's former CEO John Krafcik confessed in an interview with the Financial Times, "Self-driving cars are just too abrasive ...... this guy is harder than putting a rocket into orbit, because we have to keep repeating tests in terms of safety to get peace of mind. "

In Waymo valuation plummeted at the same time, foreign online car giants Uber, Lyft, has sold the self-driving team; Amazon was shut down in October last year the size of about 400 people Scout unmanned delivery vehicle project.

"Once on the open road, involving a series of issues such as the complexity of the right-of-way, domestic self-driving companies often fall into the 'scale trap', the larger the scale, the more investment, resulting in higher losses." A new energy vehicle enterprises autonomous driving business leader told Box Lunch Finance, Baidu adjust ACE, Ali autonomous driving business before and after the two principals choose not to go on the road in the direction of clean robot start-ups, the reasons behind which are inseparable from the "road problems.

Entering 2023, the winter in the driverless field is still spreading globally: Embark, a self-driving truck company with a market value of $5 billion, announced the end of operations and fired most of its employees; Locomation, a self-driving truck, announced layoffs of nearly 70 percent, and its products were delayed in commercial operations; Nuro, an unmanned delivery company valued at $8.6 billion, announced massive layoffs and called off plans for mass production and commercial landing of unmanned vehicles.

In May, Tucson Future, known as the "No. 1 stock in autonomous driving," officially said it was "retaining cash and maintaining its business" with a new round of layoffs, dropping its U.S. company headcount from 550 to 220. The layoffs come shortly after Tucson Future received a delisting warning from Nasdaq for failing to file its 2022 annual report.

Compared to a market cap of $12 billion at its peak, Tucson Future is now down to just over $500 million.

The latest self-driving trucking company to head into bankruptcy is the domestic Prime Smart Truck. The startup, which was branched out from Pony Smart, was forced into a liquidation and write-off process after only 19 months of operation.

03

The big AI model has become a shot in the arm for tech majors to get out of the shadow of autopilot.

Under the consensus that all applications are worth redoing with big models, Baidu released Wenxin Yiyin and Wenxin Qianfan big model service platform; Ali released Tongyi Qianqiu and Qianqiu industry big model platform; Tencent launched industry big model solutions, hoping to build a selective store in the era of big models; Byte also followed by Volcano Ark, eyeing to "sell shovels The business of "selling shovels".

The other side of the excitement of the tech majors is the obstruction of AI big models in the commercialization process.

Despite having more than 800 million monthly active users, OpenAI has still failed to explore a commercialization path of scale on the C side, and is still trapped in the trap of losing more and more money the larger the scale is.

In order to explore a new commercialization path, according to the media, OpenAI is planning to launch a large model app store similar to App Store, allowing customers to customize large models on the shelves.

Before OpenAI explored the industry big model, domestic technology companies have long moved their competitive focus to this area, but the new challenge is that it is not easy to find a suitable scenario for industry big model to be landed and promoted on a large scale.

OpenAI CEO Alterman has said in an interview, "There is no product-market fit in ChatGPT's plug-in application."

It is because of the scarcity of scenarios, recently Zhu Xiaohu, the general manager of GSR Ventures, and Fu Sheng, the CEO of Cheetah Mobile, started a debate between themselves on "how big is the value of big model entrepreneurship" and "how many opportunities are there for big model entrepreneurship".

Because model, computing power and data, the three pillars to support the operation of large models are concentrated in large factories, Zhu Xiaohu believes that the large generic model is no longer an opportunity for startups, and the moat of making applications directly on the large model is too low.

More and more people in the investment circle are cautious about the AI big model windfall, and Wei Zhe, chairman of Jiayou Capital, even said that the current round of the "100 model war", no more than 2 domestic final winners.

In Lu Qi's view, the current AI big model era is still in the early stages, after the rapid growth, the bubble burst will be inevitable phenomenon. "However, after the bubble bursts, a new dawn will come. At that time, a new generation of more powerful companies will rise from the broken bubble and occupy the industrial high ground."

Now, it's up to which tech powerhouse can hold on until the moment the dawn really arrives.

In a 136-page lawsuit filed in federal court in the District of Columbia, the SEC detailed the details of the violations by Coinan and Changpeng Zhao. intent to evade the law, and the public should be careful not to invest any of their hard-earned assets in these illegal platforms."

Commingling of Customer Funds, Hundreds of Millions of Dollars in Laundered Transactions

Similar to the allegations against former Cryptocurrency competitor FTX and its founder Sam Bankman-Fried, the SEC alleges that CZ and Cryptocurrency exercised control over customer assets by allowing them to commingle customer assets or transfer them at will, including to an entity known as Sigma Chain, which CZ owns and controls.

The filing further alleges that BAM Trading and BAM Management US Holdings, Inc. (BAM Management) misled investors about non-existent trading controls on the Binance.US platform, and that Sigma Chain engaged in manipulative trading by artificially inflating the platform's trading volume for only the period January 1, 2022 through June 23, 2022. Between January 1, 2022 and June 23, 2022 alone, Sigma Chain accounts made clean trades in 48 of the 51 newly listed crypto assets.

In addition, the complaint alleges that the defendants commingled and sent billions of dollars of investor assets to another company, Merit Peak Limited, which is controlled by Coinan founder Changpeng Zhao and appears to have been used to buy and sell cryptocurrencies.

CZ and its holding company allegedly control 100% of the various entities of Binance and Binance US, and the following is a chart of the equity structure of the Coinan entities:

The SEC also alleges that at least $145 million was transferred from Binance.US to Sigma Chain's account by 2021. Another $45 million was deposited into the account from a Nevada trust related to Binance.US, and Sigma Chain also used $11 million from the account to purchase a yacht.

Also of note, similar to the murky structure of FTX and FTX.US, the SEC says that Coin has access to Binance.US's wallet and assets, as well as the U.S. entity's custody vehicle and private keys. Until at least December 2022, Cryptocurrency is also the designated custodian of crypto assets deposited, held, traded and/or accrued on the Binance.US platform. Zhao Changpeng signed off on a number of decisions related to trading services on the Binance.US platform, including the customer account opening process, front-end access development, and the creation of reserves to cover ACH deposits.

The SEC asked the court to order an immediate "verified" accounting of Binance's and Binance.US's finances, the freezing and return of assets, the preservation of documents, and the appointment of a receiver for Binance's assets.

Binance operated as an unregistered exchange

The SEC stated, "With respect to the Binance.US platform, Binance acted as an exchange with BAM Trading, BAM Trading acted as a broker, and Binance and BAM Trading each acted as a clearing agency and were not registered."

CZ is also accused of violating the Exchange Act "as the controller of Binance and BAM Trading. Under the leadership and control of Changpeng Zhao, Binance and BAM Trading unlawfully provided three basic securities market functions - exchange, broker-dealer, and clearing agency - on the Binance platform without registering with the SEC," the filing states. were acutely aware that U.S. law required registration for these functions, but they chose not to register, so they could evade critical regulatory oversight designed to protect investors and markets."

According to the filing, Binance's chief compliance officer had told a colleague in 2018, "Dude, we're an unlicensed stock exchange in the U.S."

Circumventing regulation to solicit U.S. customers

Part of the SEC's lawsuit refers to the so-called "Taiji" documents, which Forbes originally reported in 2020, as what appears to be a plan for Cryptocurrency to officially exit the U.S. market while still maintaining a presence through an affiliate. Cryptocurrency initially sued Forbes over the report in November 2020, before dropping the suit in February 2021.

The lawsuit cites internal Cryptocurrency employees who said they had discussed ways to allow U.S. customers to trade on binance.com. The document quotes the CCO of Cryptocurrency as saying, "On the surface we don't have U.S. users, but in reality, we should be engaging them in other creative ways."

The SEC says, "In effect, Coin implemented much of the Taiji scheme, with Zhao and Binance maintaining substantial involvement and control over the U.S. entity, and behind the scenes, Zhao directed Binance to allow and conceal continued access to Binance.com for many high-value U.S. customers. "

The complaint alleges that from June 2018 through July 2021, Binance.US operated its own over-the-counter trading desk with $11.6 billion in revenue, most of which came from transaction fees, and for two years its only counterparty was Alameda Research, founded by FTX founder Sam Bankman-Fried.

Selling unregistered securities, at least 12 tokens are securities

The SEC considers the pledged investment products available on BNB, BUSD and Binance.US to count as unregistered securities. It also said BAM Trading's pledge program is an investment contract and falls under its definition of a security.

The filing alleges that Binance and BAM Trading illegally engaged in the offering and sale of unregistered cryptoasset securities, including Binance's own BNB and BUSD, as well as Binance's earnings programs BNB Vault and Simple Earn, and so-called "pledged" investment programs available on the Binance.US platform The SEC has named 12 tokens in the lawsuit as "pledged" investments.

The SEC refers to 12 tokens as securities in the lawsuit, namely BNB, BUSD, Solana (SOL), Cardano (ADA), Polygon (MATIC), Filecoin (FIL), Cosmos Hub (ATOM), The Sandbox (SAND), Decentraland ( MANA), Algorand (ALGO), Axie Infinity (AXS), and Coti (COTI).

The SEC says the list is not exhaustive: "Since the launch of the Binance platform, Defendants have been able to trade cryptoassets on these platforms, which are offered and sold as investment contracts and therefore as securities."

Notably, the SEC did not classify Ether as a security in the lawsuit. As previously reported by BizTrust, SEC Chairman Gary Gensler declined to specify whether ETH was a security during a nearly five-hour hearing in April.

What is the SEC's purpose?

Binance was founded in 2017 and has quickly grown to become a headliner in the crypto space. As of last month, more than 40 percent of crypto transactions were made through the Cryptocurrency platform, according to data provider CCData. The case is wide-ranging and has the potential to reshape the landscape of power and wealth in the crypto space.

The SEC and CFTC have been engaged in a turf battle to determine who becomes the primary regulator of crypto trading, with the CFTC filing charges in March that accused Coinan and its founder Changpeng Zhao of offering unregistered crypto derivatives in violation of federal law, and many of the allegations in the SEC's lawsuit are similar to the CFTC's complaint, which for the SEC is another major bet that a U.S. court will agree that it has jurisdiction over the crypto industry. For the SEC, this lawsuit is another major bet that a U.S. court will agree that it has jurisdiction over the crypto industry.

In the lawsuit, the SEC seeks to prevent Binance, Binance.US and "their respective agents" from continuing to violate federal law by requiring the defendants to pay "all ill-gotten gains" plus pre-judgment interest. Some optimists believe that the charges are civil and that Binance could end up paying a significant fine.

The SEC, in its filing, said it wants to bar CZ from serving as an officer or director of any issuer of securities and prevent Binance, Binance.US and CZ from participating or trading in any securities, including cryptoasset securities. This also includes preventing the defendants from acting as unregistered brokers, clearing agencies or trading with any crypto-asset securities, the lawsuit says.

Both the CFTC and SEC have accused Coinan of failing to comply with U.S. anti-money laundering laws, and the U.S. Treasury Department, which oversees much of the U.S. anti-money laundering regime, has yet to respond, and much of the evidence contained in the civil lawsuit could be used to support separate enforcement actions by the Treasury and Justice Departments if Coinan is suspected of violating sanctions.

It is not yet known if CoinA will survive the crisis, and it has said it hopes the U.S. Congress will step in and pass bipartisan legislation to create a workable regulatory regime for digital assets to prevent the crypto industry from becoming "a victim of bureaucratic overreach.

Aaron Kaplan, co-founder of Prometheum, a digital asset securities trading platform registered with the SEC and FINRA, commented in a tweet, "This meaningful action by the SEC marks the beginning of a shift to a regulated cryptocurrency market infrastructure in the US, which should ultimately help the industry move forward. I expect this to be a net gain for U.S. investors and should allow innovation to flourish."

One of the AIGC startups known to Planet Digital has accumulated more than 100,000 users since it explored the To B market. Investors laughingly say, "This has only happened in the social and gaming industries in the past."

Today, business owners are leaving behind the SaaS pie and embracing AIGC for no other reason: because there is an immediate need for layoffs.

To B has great potential

The enterprise market has always been considered a slow market because of the "collective decision". Behind the slow burn, what attracts investors most is the speed of expansion of SaaS products, for example.

However, recently, AIGC venture capital investment in the enterprise services industry has clearly overshadowed the SaaS track. One tool-based AIGC startup had this to say to Planet Digital: "We got 100,000 users in two months and now have more than 600 agents nationwide." This is a SaaS transformed AIGC company, and after this founder communicated with several customer companies, he found that the acceptance of To B type AIGC products is good.

"Office category is definitely a bigger scene now, office is the one that involves people's greatest energy, and ordinary people actually want to save time and improve efficiency here the most," an investor said to Digital Planet.

Objectively speaking, the combination of AIGC and office has many scenarios, in addition to the underlying speech recognition, text classification, machine translation, etc., there are also enterprises using related technologies to detect the emotional state of employees. In the technical scenario, AIGC can pre-process large data sets in advance to reduce data errors.

As mentioned in the chatbot products pushed by some major domestic manufacturers, AIGC in To B can now complete automatic replies, automatic typesetting, etc.

One company told Planet Digital that "we already have more than 100,000 users in the queue", which shows how hot AIGC is in the To B side.

SaaS can only converge

Among the multiple companies we spoke to, companies that focus on content marketing copywriting are also being sought after by To B customers.

"The track is very crowded, and while growth is still good, there are very many parties involved", the executive replied. However, in the knowledge of Planet Digital, the business has already started to differentiate itself. "We have made our own Promote based on GPT, we have trained more than 20 models, and now we have covered more than 90 scenarios in apps like Little Red Book." In the executive's view, the differentiation between this company and its peers is that it has an exclusive model that works better in writing marketing copy than other big models in the market.

There is another company that has accumulated a lot of official writing data in the past, and since this kind of data is not easy to get, the official writing model derived from the combination of AIGC is quite popular in the market in this company. We learned from a practitioner that the current attitude towards official writing at the government level is not exclusive, and several government-level clients have been approached by this company.

In addition to the above mentioned, in fact, in the To B field, there are other scenarios that have quite a lot of potential: such as generating PPT with one click, generating brain maps with one click, generating official websites or digital people with one click, and many other needs.

What is more exaggerated is that some AIGC products have in fact stepped into the forbidden domain of SaaS companies::

For example, in customer support and service scenarios, AIGC can now help as a first responder. It generates fast, accurate solutions that reduce the stress on customer service teams, and AI that taps into knowledge bases, FAQ and tutorial systems;

In the area of business process automation, report creation, data analysis, and document review can already be done with AIGC; even in business decisions, it can be used to predict market trends, customer retention, and potential revenue.

In addition to development, marketing and customer service, sales people can also experience automatic generation of sales proposals, quotes and contracts in the related products. Some foreign trade industry executives have shared to, "In this regard, there is no longer any difference between what robots do and people, and I even think they do it better."

A founder who had rolled within SaaS poured out his bitterness: "In 2019 we did SaaS, the industry was particularly rolled, so much so that we could only do vertical and focused to have room to survive, and finally we had to cut other unrelated functions and only do segmentation, which was really too rolled at that time."

However, in this year, after the first trial of AIGC, the situation has been improved.

Although the product types are diverse and the scenarios are different. However, in our series of discussions, in the question of the user's willingness to buy executives gave the same answer: layoffs. "Because everyone is under particular pressure now, I have many customers on my side asking if there are tools to cut 10 people to five," mentioned one industry practitioner.

This means that, at least recently, the growth logic of enterprise services will be affected. One practitioner put it this way: "Previously, we couldn't push outward even if we begged, and China's SaaS market is 1/8 of foreign countries, but this time it's different."

In the past, the product focus of the enterprise service industry was on data collection, process management and process optimization: "In essence, enterprises are not very interested in these," with two main reasons for this result: one is that domestic enterprises do not have a high awareness of digitalization compared to foreign countries; the other is that domestic enterprises generally do not have money. One investor added that he believes that "in China, the previous software had people over-invested in it, so much so that it took thousands of clicks and hundreds of operations to do something, and AI has taken that province-wide and the work has been extremely simplified."

In terms of efficiency improvement, the effect of AI is obvious and gives a more intuitive result from a human perspective: that is, the workload remains the same and the salary given to the employee becomes a gain for the boss.

All of the above changes took place over a period of two months. It is worth noting that the concept of AIGA came out of the blue when the corporate world was not yet familiar with AIGC.

AIGA means Artificial Intelligence Generated Action, which in the industry is understood to be a re-upgrade of automation. Under the AIGA framework, AI can help analyze project requirements, balance team skills and assign task priorities.

In CRM scenarios, AIGA can automatically generate plans to interact with customers based on their behavior and needs. In the past, some domestic CRM products with LBS recommendations for sales leads have emerged, but the sales actions of these products still rely on real people to carry out. In AIGA, following up with potential customers and pushing information that is more targeted to customers may be replaced by AI.

To be more radical, if AIGC replaces the low-level workforce, then AIGA may replace the mid-level workforce. If we are more radical, the emergence of pure AI enterprise service companies in the future is not unlikely. Translated with www.DeepL.com/Translator (free version)

If you are someone who is interested in understanding what forms a good relationship and how to connect people together more closely, then you have come to the right place. Our voluntary group is dedicated to studying the science of intimacy and exploring the secrets of successful relationships. Through our research, discussions, and study reports, we have discovered some fascinating insights and strategies that can help couples build and maintain healthy, fulfilling relationships.

In this article, we will share some of the key findings from our research on intimacy and relationships. We will explore the factors that contribute to successful relationships, the challenges that couples face, and the strategies that can help them overcome these challenges. We will also discuss how you can get involved in our community and receive a free NFT by subscribing to our account.

What is Intimacy?

Intimacy is a term that is often used to describe the emotional and physical connection between two people. It involves sharing personal information, thoughts, feelings, and experiences with another person in a way that is open, honest, and vulnerable. Intimacy can take many forms, including emotional intimacy, physical intimacy, and intellectual intimacy.

Intimacy is a crucial component of healthy relationships. It helps to create a sense of closeness and trust between partners, and it can enhance the quality of communication and the overall satisfaction of the relationship. However, intimacy can also be challenging to achieve and maintain. It requires a willingness to be vulnerable, to communicate openly and honestly, and to be accepting of one's partner.

Factors that Contribute to Successful Relationships

Our research has identified several factors that contribute to successful relationships. These include:

1. Communication:

Effective communication is a crucial component of any healthy relationship. It involves listening actively, expressing oneself clearly, and being respectful of one's partner's opinions and feelings. Good communication can help to build trust, increase intimacy, and resolve conflicts in a constructive way.

2. Empathy:

Empathy is the ability to understand and share the feelings of another person. It is a crucial component of emotional intimacy and can help partners to feel understood and validated. Empathy involves listening attentively, validating one's partner's feelings, and showing compassion and support.

3. Trust:

Trust is the foundation of any healthy relationship. It involves having confidence in one's partner's integrity, honesty, and reliability. Trust can be built through open and honest communication, consistency in behavior, and mutual respect.

4. Shared Values:

Shared values and interests can help to create a sense of connection and purpose in a relationship. They can provide a basis for shared experiences, conversation, and growth. Shared values can also help partners to navigate challenges and conflicts in a way that is consistent with their shared goals and beliefs.

Challenges Faced by Couples

While successful relationships can be deeply fulfilling, they can also be challenging to maintain. Some of the most common challenges that couples face include:

1. Communication Issues:

Communication problems are one of the most common challenges that couples face. This can include difficulty expressing oneself clearly, not listening actively, and misunderstanding one's partner's intentions or feelings.

2. Differences in Values or Interests:

Differences in values or interests can lead to conflicts and misunderstandings in a relationship. It is important for couples to be aware of these differences and to find ways to respect and appreciate each other's perspectives.

3. Infidelity:

Infidelity can be a significant challenge to a relationship. It can lead to trust issues, hurt feelings, and a breakdown in intimacy. Recovering from infidelity can be a long and difficult process, requiring honesty, forgiveness, and a commitment to rebuilding the relationship.

4. Life Transitions:

Life transitions, such as moving, starting a new job, or having children, can be stressful and challenging for couples. These transitions can disrupt established routines and create new stressors in the relationship. It is important for couples to communicate openly and to support each other through these transitions.

Strategies for Building Successful Relationships

Despite the challenges that couples may face, there are strategies that can help to build and maintain successful relationships. These include:

1. Practicing Effective Communication:

Effective communication is key to building and maintaining healthy relationships. This involves actively listening to one's partner, expressing oneself clearly, and being respectful of each other's feelings and opinions. Couples can practice effective communication by setting aside dedicated time to talk, using "I" statements instead of "you" statements, and avoiding criticism and defensiveness.

2. Building Trust:

Trust is essential to a healthy relationship. Couples can build trust by being reliable, consistent, and honest with each other. They can also communicate openly about their expectations and boundaries, and work together to establish a sense of shared responsibility and mutual respect.

3. Prioritizing Intimacy:

Intimacy is a crucial component of any healthy relationship. Couples can prioritize intimacy by setting aside dedicated time for each other, engaging in shared activities, and expressing their love and affection regularly. This can help to build a sense of connection and trust, and can enhance the overall quality of the relationship.

4. Seeking Support:

Seeking support from friends, family, or a therapist can be helpful for couples facing challenges in their relationship. A trained therapist can provide a safe and neutral space for couples to communicate, explore their feelings, and work together to find solutions to their challenges.

Join Our Community: How to Get Involved

If you are interested in learning more about the science of intimacy and building successful relationships, we invite you to join our community. Our voluntary group is dedicated to studying the science of intimacy and exploring the secrets of successful relationships. We post regular study reports and insights on our account, sharing our knowledge and discoveries with our followers.

As a special offer, we are also giving away free NFTs to our subscribers. NFTs, or non-fungible tokens, are digital assets that represent ownership of a unique item, such as a piece of art or a collectible. By subscribing to our account, you can mint your own free NFT, which will give you exclusive access to our future study reports and events.

https://opensea.io/assets/ethereum/0x0F9413c13B344b3f4bdCD398b9ef5c1367D47665/0

So what are you waiting for? Join our community and discover the secrets of successful relationships. By subscribing to our account, you can mint your own free NFT and gain access to our exclusive content and insights. We look forward to connecting with you and sharing our knowledge and discoveries with you.

Content production, especially creative work, has always been considered the exclusive province of humans and the embodiment of intelligence. In the 2019 book "The Complete Biography of Artificial Intelligence" by Michael Wooldridge, Dean of the School of Computing at Oxford University, "writing interesting stories" is listed as one of the "far from achieved" tasks of AI.

AIGC (AI Generated Content) not only achieves "human-like" performance in many areas of writing, drawing, and composing, but also demonstrates extraordinary creative potential based on big data learning. The benchmark GPT-4 model was officially released to further improve the accuracy and compliance of generated content. A new paradigm of human-computer collaboration in digital content production is emerging, allowing creators and more ordinary people to cross the limits of "technique" and "effectiveness" and to be creative with their content.

There are also concerns about whether AI will make creators "unemployed" en masse, or even let "creation" itself go into decline, just as artworks in the era of mechanical reproduction may lose their "spirit". The same way. In other words, the popularity of AIGC gives us an opportunity to re-examine the questions of what "creation" is and whether it is uniquely human.

In this paper, we will analyze the current situation, key breakthroughs and challenges of AIGC in changing digital content creation, and try to explore the above questions.

AIGC is becoming the infrastructure of Internet content production Digital content is moving into an upgrade cycle of strong demand, video and creativity, and AIGC is at the right time. Online life has become the norm, on the one hand, user-created content has greatly liberated productivity, for example, short videos have turned videos that originally required long production cycles and high attention to investment into "industrial products" and "FMCG products" that can be produced continuously; on the other hand On the other hand, creativity as the core is still scarce, and new models are needed to assist creators to continuously generate, iterate and verify ideas. All these factors require new tools and methods that are more cost effective.

AIGC is increasingly involved in the creative generation of digital content, releasing value in a collaborative human-machine way and becoming the future content production infrastructure of the Internet.

In terms of scope, AIGC is gradually and deeply integrated into the production of text, code, music, pictures, videos and 3D in various media forms, and can serve as news, essay and novel writers, music composers and arrangers, painters of diverse styles, editors and post-processing engineers of long and short videos, 3D modelers and other diverse assistant roles, and complete the creation of specified thematic content under the guidance of humans, and editing and style migration work under human guidance.

In terms of results, AIGC is initially satisfactory in the areas of natural language-based text, speech and image generation, especially for knowledge-based short and medium texts, illustrations and other highly stylized image creation, where the creation results can match those of creators with intermediate experience; it is in the exploratory stage in areas of high media complexity such as video and 3D. Although there is still much room for improvement in AIGC's handling of extreme cases, control of details, and accuracy of the finished product, the potential it contains is promising.

In terms of approach, multimodal processing across text, image, video, and 3D in AIGC is hot. Andrew Ng sees multimodality as the most important trend in AI in 2021, and AI models have made significant advances in discovering relationships between text and images, such as OPEN AI's CLIP that matches images and text, and Dall-E that generates images corresponding to input text; DeepMind's Perceiver IO that can classify text, images, videos, and point clouds. Typical applications include Text-to-Speech (TTS), Text-to-Image (TTI), and in a broader sense, AI translation and image stylization can also be seen as a mapping between two different "modalities".

Key breakthrough: Natural language technology liberates creative power The liberation of AIGC for creators is reflected in the fact that "if you can talk, you can create", without the need to know the principles, learn code, or professional tools such as Photoshop. The creator describes the elements or even ideas in his mind to the AI in natural language (the term is "prompt"), and the AI generates the corresponding results. This is the next leap in human-computer interaction from punched paper tapes, to programming languages, to graphical interfaces.

Natural language is the root information and link between different digital content types, such as the word "cat" is the picture of Garfield, the musical "Cat" and countless content indexes, these different content types can be called "multimodal".

In this wave of AIGC, the biggest underlying evolution is in the leap of AI's ability to "understand" and "use" natural language, which cannot be separated from Google's Transformer released in 2017, which opened the Large Language Model (LLM). Large Language Model (LLM) era. With this powerful feature extractor, subsequent language models such as GPT and BERT have advanced by leaps and bounds, not only in terms of quality and efficiency, but also in terms of big data pre-training + small data fine-tuning, freeing them from the reliance on a large number of manual tuning parameters, resulting in significant breakthroughs in handwriting, speech and image recognition, language understanding, and increasingly accurate and natural content.

But large models imply extremely high barriers to research and use, such as GPT-3 with 175 billion parameters, which requires both large computing clusters and is not available to general users. midjourney, deployed on the Discord forum and available as a chatbot, became the first user-friendly AIGC application in 2022, bringing an AI painting boom, and one designer used it to generate images that even won an offline competition.

The low threshold of using simple text to communicate and the search engine like usage immediately ignited the enthusiasm of ordinary users to use AI. This was followed by a series of Text-to-Image products based on Diffusion Models, such as Stable Diffusion, which brought AI painting from the design community to the masses. The open source Stable Diffusion, which can run on just one computer, has been downloaded by more than 200,000 developers and has accumulated more than 10 million daily users as of October 2022, while the consumer-oriented DreamStudio has gained more than 1.5 million users and generated more than 170 million images. Its stunning artistic style, as well as the copyright and legal issues involved in images, have also sparked much controversy.

Before the shock of Diffusion wore off, ChatGPT came out and really "spoke" to humans, understanding a wide range of needs, writing answers, short essays and poems, writing code, mathematical and logical calculations, and more. In addition, human feedback reinforcement learning (RLHF) technology allows ChatGPT to continuously learn from human suggestions and evaluations of responses and move in the right direction, thus achieving excellent results with less than 1% of the parameters of GPT3. Although ChatGPT still has some flaws, such as references to non-existent papers and books and poor quality answers to questions that lack data, it is still a milestone in the history of artificial intelligence, and two months after its launch surpassed 100 million users, making it the fastest growing consumer app in history in terms of users.

The Next Challenge: Toward a 3D Internet of Presence After text, graphics and video, the important direction of digital technology evolution is to move from "online" to "present", and AIGC will be the cornerstone of building 3D Internet. People will build a simulation world in virtual space and "superimpose" virtual enhancement in the real world to realize a real sense of presence. With the breakthrough of XR, game engine, cloud game and other interaction, simulation and transmission technologies, the information transmission is getting closer to lossless, the digital simulation ability is indistinguishable from the real one, and human interaction and experience will reach a new stage.

At present, AIGC is still in the exploration stage in the field of 3D model, a path is based on diffusion model in two steps: first generate pictures from text, and then generate 3D data containing depth. Google and Nvidia in this field is more leading, has released their own text-generated 3D AI model. However, from the generation effect, there is still a distance from the average quality of 3D content produced manually now; the generation speed also fails to be satisfactory.

In October 2022, Google was the first to release DreamFusion, but its shortcomings are also significant. Firstly, the diffusion model only takes effect for 64x64 images, resulting in low quality of generated 3D; secondly, the scene rendering model not only requires massive samples, but also is computationally time-consuming and laborious, resulting in slower generation speed. Then, NVIDIA released Magic3D, which took about 40 minutes to generate a 3D model with texture in the face of the prompt "a blue poison dart frog sitting on a water lily". Compared to Google, Magic3D generates faster and better results, and can retain the same theme during successive generation, or migrate the style to the 3D model.

The second path is to use AI to "composite" photos of the same object from different perspectives to generate 3D directly, and NVIDIA is demonstrating at NeurIPS in December 2022 a generative AI model, GET3D (short for Generate Explicit Textured 3D), that can synthesize 3D models on-the-fly based on the 2D image categories it is trained on, such as buildings, cars, animals, etc. Generate Explicit Textured 3D), which can synthesize 3D models on-the-fly based on the 2D image categories it has trained on, such as buildings, cars, and animals. GET3D is trained on NVIDIA A100 GPUs and uses about 1 million photos taken from different angles to generate about 20 objects per second. It can generate about 20 objects per second. Combined with another of the team's technologies, the AI-generated models are able to distinguish between object geometry, lighting information and material information, allowing for significantly enhanced editability.

Possible path: Combining with procedural generation techniques in games Nevertheless, AIGC's ability on the 3D side is still a long way from creating 3D Internet. And the more mature programmed content generation (PCG, Procedural Content Generation) technology in the game, may be a big help for AIGC to step through the deep water.

From the technical path, AI generated 3D is difficult to follow the old way of "vigorous miracle", that is, to improve the effect by feeding the AI a large amount of input alone. First of all, the amount of information is different, a picture and a 3D model compared to the difference in a dimension, reflected in the storage is a different level of data; secondly, the picture and 3D storage and display principle is different, if 2D is a pixel array in the display of the objective display, 3D is real-time, fast, massive matrix operations, just like the model in 1 second dozens of times " photographing" in one second. In order to accurately calculate each pixel point and "render" it on the display, we need to consider at least (1) the geometric characteristics of the model, which are usually represented by thousands of triangular surfaces (2) the material characteristics, the color of the model itself, whether it is a strongly reflective metal or diffuse reflective fabric (3) the light, whether the light source is point-like, the color and intensity. Finally, there is relatively little data available for native 3D models, with only a small accumulation in games, film and television, and digital twins, far less than images that have existed for thousands of years and can exist in a non-digital form, such as ImageNet, which contains over 14 million images.

The matter of using computers to help creators has been explored by the gaming community for over forty years. Game content generated with algorithms first appeared in the game Rogue (Toy and Wichman) in 1981, with a random map, different for each game. 3D era, procedural generation techniques were heavily used in art production because of its huge time and labor costs, taking the game "Wilderness Dart Hunter 2" released in 2018 as an example, which involved more than six hundred artists and took eight years to complete a virtual scene of about 60 square kilometers.

Procedural generation is somewhere between pure manual and AIGC in terms of effectiveness and controllability. For example, No Man's Sky, an indie game released in 2016 that focuses on cosmic exploration, uses PCG to construct a series of generation rules and parameters that claim to create 184 billion different planets, each with a different form of environment and creatures.

The Slope wallet team invited the Slow Fog security team to analyze and follow up with them, and after continuous follow-up and analysis, the data provided by Solana foundation showed that nearly 60% of the stolen users used Phantom wallet, about 30% of the addresses used Slope wallet, the rest used Trust Wallet, etc., and both iOS and Android versions of the app had corresponding victims. We started to focus on analyzing the possible risk points of the wallet application.

Analysis process

While analyzing Slope Wallet (Android, Version: 2.2.2), we found that Slope Wallet (Android, Version: 2.2.2) uses the service Sentry, which is a widely used service, and Sentry runs under the o7e.slope. Sentry runs under the o7e.slope.finance domain and sends sensitive data such as helpers and private keys to https://o7e.slope.finance/api/4/envelope/ when creating a wallet.

Continuing the analysis of Slope Wallet, we found that Version: >=2.2.0 package Sentry service sends helper words to "o7e.slope.finance", while Version: 2.1.3 does not find the behavior of collecting helper words.

Slope Wallet history version download.

https://apkpure.com/cn/slope-wallet/com.wd.wallet/versions

Slope Wallet (Android, >= Version: 2.2.0) was released on and after 2022.06.24, so it affects users who use Slope Wallet (Android, >= Version: 2.2.0) on and after 2022.06.24, but according to some victims feedback did not know about Slope Wallet, and did not use Slope Wallet

Then, according to Solana Foundation statistics, about 30% of the victim's addresses may have been captured by Slope Wallet (Version: >=2.2.0) Sentry's service and sent to Slope Wallet's https://o7e.slope.finance/api/4 /envelope/ server.

But the other 60% of the stolen users were using Phantom wallets, so how were these victims stolen?

Analysis of the Phantom (Version:22.07.11_65) wallet revealed that Phantom (Android, Version:22.07.11_65) also uses the Sentry service to collect information from users, but no obvious collection of helpers or private keys was found. (The security risk of historical versions of Phantom Wallet is still being analyzed by the Slow Fog security team)

Some questionable points

The Slow Fog security team is still collecting more information to analyze the reasons why another 60% of the stolen users were hacked, so if you have any ideas, please feel free to discuss them together and hopefully do your part for the Solana ecosystem. The following are some of the questionable points in the analysis process.

1. is Sentry's service collecting users' wallet mnemonics a common security problem?

2. Phantom uses Sentry, so will Phantom wallets be affected?

3. what are the reasons for the other 60% of users being hacked?

4. As Sentry is a very widely used service, could Sentry have been officially compromised? Could this have led to a targeted attack on the virtual currency ecosystem?

Reference information

Known attacker address.

Htp9MGP8Tig923ZFY7Qf2zzbMUmYneFRAhSp7vSg4wxV

CEzN7mqP9xoxn2HdyW6fjEJ73t7qaX9Rp2zyS6hb3iEu

5WwBYgQG6BdErM2nNNyUmQXfcUnB68b6kesxBywh1J3n

GeEccGJ9BEzVbVor1njkBCCiqXJbXVeDHaXDCrBDbmuy

Victim's address.

https://dune.com/awesome/solana-hack

Solana foundation statistics.

https://www.odaily.news/newsflash/294440

https://solanafoundation.typeform.com/to/Rxm8STIT?typeform-source=t.co

https://docs.google.com/spreadsheets/d/1hej7MnhI2T9IeyXpnESmMcIHwgxGucSGUxQ5FqHB9-8/edit#gid=1372125637 (requires access request)

A report released Friday by the Bank for International Settlements presents the results of a survey of 81 central banks conducted in the fall of 2021. The 81 banks included 25 developed country central banks and 56 developing country central banks, representing nearly 76% of the world's population and 94% of global economic output. The survey explored the extent to which banks are involved in CBDC-related work, as well as their motivations and intentions for issuing CBDCs. Fifty-six of these respondent banks participated in the 2020 survey, with 41 of them participating for the fourth time. This allows BIS to assess how their views and involvement in CBDC issuance have changed over time.

The proportion of central banks actively engaged in some form of CBDC-related work has grown to 90% in the past year. The survey data shows that central banks are particularly interested in retail CBDCs: all central banks doing CBDC-related work are either focusing on both wholesale and retail, or on retail-only CBDCs. an increasing number of central banks are in the advanced stages of exploring CBDCs. Compared to last year, the percentage of central banks that are currently developing CBDCs or conducting pilots has almost doubled, from 14% to 26%. In addition, 62% of banks are in the process of experimentation or proof of concept. Retail central banking efforts are at a more advanced stage than those of wholesale central banking. Nearly one-fifth of central banks are developing or testing retail-style CDBs, twice the share of central banks building or piloting wholesale CDBs.

The report concludes that the majority of central banks surveyed are exploring CBDCs, and more than half are developing or conducting specific experiments. Retail CBDC-related efforts, in particular, have moved into a more advanced stage. Two-thirds of central banks are considering issuing retail CBDCs in the future, joining the Bahamas, China, the Eastern Caribbean and Nigeria, countries that have already issued CBDCs.

The shape of the future retail CBDC ecosystem is likely to become very clear as CBDC issuance is gradually moving from research to actual implementation globally. Many central banks are exploring interoperability with existing payment systems and are considering bringing the private sector into the fold. Public and private sector cooperation, coupled with interoperability, will help create an ecosystem in which CBDCs coexist with other means of payment.

Currently, there is a general consensus among central banks that the use of stablecoins is limited to niche groups or specific use cases. However, a significant portion of the population believes that stablecoins have the potential to become a widely used payment method. Surveys show how the emergence of stablecoins and other cryptocurrencies has accelerated global efforts related to the issuance of CBDCs.

The issue of cross-border payment efficiency has come up several times over the past year, and has played a large role in central banks' work on issuing wholesale CBDCs in particular. A multi-year G20 initiative is underway to make cross-border payments faster, cheaper, more transparent and more convenient. An industry source said, "CBDCs can play an important role in this regard, especially in terms of shortening the current transaction chain and providing longer operating hours." CPMI will work with BISIH, the IMF and the World Bank to present a report to the G20 in July 2022 that will further analyze options for interconnecting countries' CBDCs to improve cross-border payments.