⬆️ UPVOTED | AI3: THE CONVERGENCE OF AI & WEB3
Around this time a year ago, I was astonished (along with the rest of the world) to witness OpenAI’s ChatGPT explode onto the scene. The new AI chatbot, built on top of a powerful, general-purpose large language model (LLM), went from zero to 100 million monthly active users in just two short months — making it the fastest-growing consumer application in history. And in spite of all the debate going on at the time - essentially questioning the intrinsic value of a hallucinating, stochastic parrot - it soon became clear, as the model improved, that we were living through the early stages of a complete paradigm shift in our relationship to artificial intelligence. For me at least, after that realisation came a sobering thought: that such power - wielded in the hands of a single company (OpenAI), or indeed controlled by a handful of the incumbent tech giants, including Microsoft and Google - might pose a significant threat to our future sovereignty and wellbeing, both as individuals and at the societal level. And having chewed on that thought for a while, I concluded that these recent developments would likely, within a relatively short space of time, make the mercurial, centralising power of Web 2.0 (and its concomitant business model of ‘surveillance capitalism’) look like an hors d'oeuvre to the main event…
Since then, alas, the strength of my conviction in that conclusion has only grown. The bizarre course of events surrounding Sam Altman’s firing last November, and the subsequent blitzkrieg of a countercoup he stage managed from his San Francisco bunker, left me in very little doubt that the normal rules of engagement are not going to apply when it comes to AI. In other words, I have the distinct feeling we’re not in Kansas anymore, as Dorothy said to Toto. The entire incident spooked the bejesus out of me — based as much on what we didn’t (and still don’t) know about what transpired behind the scenes, as what was playing out in the social media spotlight.
I do know one thing for sure: the official narrative about what happened during that insane week last November - and more importantly why it happened - has more holes in it than a piece of Swiss cheese. My personal opinion is that it may have had something to do with significant breakthroughs made by OpenAI’s internal skunkworks team on the road to achieving AGI. And this was the reason for the extreme knee-jerk reaction from the board, who hadn’t been kept entirely up-to-speed on the breaking developments by Altman. In other words, they panicked. And it was this dynamic (in my opinion) that brought one of the hottest multibillion dollar companies in the world to the brink of collapse overnight — faster than Bill Gates can type Control-Alt-Delete when confronted by his operating system’s blue screen of death.
In any event, all of this led me to believe that the impact of such a profoundly transformative technology can only have net-negative consequences in its existing format — i.e. unless it is widely distributed, rather than being heavily concentrated in the hands of a few unscrupulous, power-hungry, Ivy League sociopaths.
Because you know what they say: power tends to corrupt — but absolute power corrupts absolutely. Or put another way: a person's sense of morality lessens as his or her power increases. And a rapid decline in morality and ethics coupled to an exponential explosion in the power of AI?…
Well, I would encourage you to perform your own calculus on that. The OpenAI board clearly did perform some kind of calculus last year, to the extent that it seemed momentarily prepared to sacrifice the success of the entire company. And I appreciate that some of you may not arrive at the lair of a Bond villain, as he strokes his Turkish Angora — as did I.
But here’s my take for what it’s worth: if there are a handful of existential crises we may face at the civilisational level (pick your poison: nuclear war, mass species extinction, ecological contamination from GMOs, extreme electromagnetic pulses knocking out the energy grid, etc.), then I would rank this issue as perhaps being somewhere in the top five on that list. Such is the future power of AI.
Why am I being such a drama queen? Because we have already seen the harmful consequences that the Web 2.0 business model has had on both individuals and our society at large. We already know what it looks like — Shoshana Zuboff, Jeff Orlowski and others have already done the work. The fact that young teenage girls want plastic surgery more than puppy dogs should tell you everything you need to know. But with the centralisation of AI applications running on similar rails to the status quo with Web 2.0 (i.e. under the control of a few centralised entities), I think we should assume (given the disruptive power of AI) that the ultimate ramifications will be several orders of magnitude worse…
So, from where I’m sitting, it’s not a pretty picture. All of the things that are already deeply concerning about the influence of the existing Web 2.0 chokepoint operators - the erosion of our privacy, the commodification of our data, the use of psychological manipulation techniques and sophisticated algorithms, the warping of reality through the deliberate curation and censorship of content, and so much more - will all carry over, on steroids, into this new domain, which I am calling AI 2.0.
But you’re a crypto bro, I hear you say, and this is a crypto newsletter — so why are you straying so far out of your lane? Well, I’m glad you asked… because it turns out that we may already have many of the basic building blocks to prevent the worst case scenario from playing out — so long as we’re quick footed. These building blocks are embedded in the blockchain-based primitives that define the contours of the existing, decentralised world of Web3.
It’s a nascent field — the intersection of AI and Web3. But there’s a growing amount of interest in the topic, as illustrated by a couple of recent blog posts from both De University of Ethereum (UETH) and Ethereum’s founder, Vitalik Buterin, himself.
UETH published a blog post in January, entitled, When Giants Collide: Exploring the Convergence of Crypto x AI — in which it declared:
The pronounced dominance of centralised ownership in the AI landscape underscores the urgency of addressing these issues. Large corporations’ extensive control over data not only shapes market dynamics but also influences the direction and ethics of AI development. This situation calls for a careful examination of how AI models are trained, the data they use, and the mechanisms in place to ensure equitable and secure use of AI technologies.
And in terms of specificity, UETH raises the following areas of pressing concern attendant with the unbridled centralisation of AI (aka AI 2.0):
Market Dominance and Innovation Stifling: The monopolistic control exerted by these corporations can significantly hinder competitive diversity. It creates a high barrier to entry for smaller entities, potentially stifling innovation and limiting the scope of technological advancements.
Bias and Ethical Fairness: Predominantly based in the United States, these companies' AI systems are often trained on data reflective of specific demographics. This raises the likelihood of perpetuating existing biases embedded in their data sets, leading to ethical questions around fairness and representativeness.
Data Security Risks: With vast amounts of sensitive information centralised in the hands of a few, the risk of data breaches becomes a major concern. A single breach could have widespread, potentially catastrophic, implications.
Transparency in AI Decision-Making: The control of AI training data and algorithms by a small cohort of corporations clouds transparency. It becomes challenging to discern how AI decisions are made, assess their fairness, and understand their impact.
Vitalik Buterin takes a slightly more reserved stance in his January 2024 blog post, entitled, The promise and challenges of crypto + AI applications; recognising how early we are in terms of the convergence of these two intersecting fields, and highlighting some of the obstacles that we need to overcome, such as the vulnerability of decentralised, open source architectures to adversarial machine learning attacks:
Many people over the years have asked me a similar question: what are the intersections between crypto and AI that I consider to be the most fruitful? It's a reasonable question: crypto and AI are the two main deep (software) technology trends of the past decade, and it just feels like there must be some kind of connection between the two. It's easy to come up with synergies at a superficial vibe level: crypto decentralisation can balance out AI centralisation, AI is opaque and crypto brings transparency, AI needs data and blockchains are good for storing and tracking data. But over the years, when people would ask me to dig a level deeper and talk about specific applications, my response has been a disappointing one: "yeah there's a few things but not that much".
In the last three years, with the rise of much more powerful AI in the form of modern LLMs, and the rise of much more powerful crypto in the form of not just blockchain scaling solutions but also ZKPs, FHE, (two-party and N-party) MPC, I am starting to see this change. There are indeed some promising applications of AI inside of blockchain ecosystems, or AI together with cryptography, though it is important to be careful about how the AI is applied. A particular challenge is: in cryptography, open source is the only way to make something truly secure, but in AI, a model (or even its training data) being open greatly increases its vulnerability to adversarial machine learning attacks.
There’s so much to unpack here, and I’m nowhere near technically gifted enough to even begin to try. For example, the acronym “FHE” in the excerpt above stands for Fully Homomorphic Encryption, which is regarded by many as one of the skeleton keys to unlock this decentralised AI future:
Fully homomorphic encryption has for a long time been considered one of the holy grails of cryptography. The promise of fully homomorphic encryption (FHE) is powerful: it is a type of encryption that allows a third party to perform computations on encrypted data, and get an encrypted result that they can hand back to whoever has the decryption key for the original data, without the third party being able to decrypt the data or the result themselves.
[Source: https://vitalik.eth.limo/general/2020/07/20/homomorphic.html]
One topical area I can point to in relation to this decentralised future version of AI underpinned by Web3 without getting too deep into the weeds is decentralised compute. You may have noticed that NVIDIA continues to print money owing to the massive demand for its GPUs — largely driven by the recent explosion in AI. And you may have also seen the story about Sam Altman trying to drum up support for a $7T+ project to bootstrap a future supply chain with sufficient bandwidth to supply all the chips needed to keep OpenAI and other AI-centric companies moving along at warp speed. For the foreseeable future, however, there is a serious supply shock of high-powered GPUs.
The Render Network (RNDR) is a good example of a blockchain project operating at this new nexus between Web3 and AI. Render is essentially a blockchain marketplace for idle GPU compute. Launched in 2017, with the primary objective of “empowering artists to scale GPU rendering work on-demand to high performance GPU Nodes around the world”, the platform now finds itself in the enviable position of having an architectural capability of meeting at least some of the overwhelming demand for GPU processing power within the AI domain. And there are many more examples of this kind of cross-pollination between crypto and AI. Things are hotting up in this space, and AI-related crypto projects are generating an increasing amount of interest.
In concluding, I thought it might be interesting to ask ChatGPT for its input on the topic. I didn’t know quite what to expect — it felt a bit like asking a turkey about its plans for Christmas. And it came with the usual boilerplate disclaimers at the end, of course — but I was encouraged to find that it didn’t shy away from the crux of the question:
OpenAI, as an organisation, does present some aspects that could be considered centralisation risks within the context of promoting decentralised approaches to AI development. Here are a few considerations:
Control over AI Models: OpenAI has historically developed and controlled some of the most advanced AI models, such as GPT (Generative Pre-trained Transformer) models. This concentration of power in a single entity raises concerns about centralisation, as decisions regarding the development, deployment, and governance of these models are primarily made by OpenAI.
Access to Data and Resources: OpenAI's access to vast amounts of data and computational resources is another aspect that could be seen as centralisation. The organisation's ability to train large-scale models relies on proprietary datasets and significant computational infrastructure, which may not be accessible to all researchers or developers, potentially limiting innovation and democratisation in the AI field.
Decision-making Authority: OpenAI's decisions regarding model releases, access policies, and partnerships can have significant implications for the broader AI community. While the organisation aims to balance openness with responsible AI development, the centralised nature of these decisions raises questions about inclusivity, transparency, and accountability.
ChatGPT also suggested some important areas of focus at the intersection of AI and Web3:
Data Ownership and Sharing: Web3 technologies, particularly blockchain, enable individuals to have more control over their data through cryptographic techniques like encryption and decentralised storage. This ownership model allows users to decide whether and how their data is used for AI training, fostering a more transparent and fair data economy.
Decentralised AI Marketplaces: Blockchain-based platforms can facilitate peer-to-peer transactions for AI models, datasets, and algorithms. These decentralised marketplaces eliminate the need for intermediaries, enabling direct transactions between AI developers and consumers while ensuring transparency, security, and fair compensation for contributors.
AI Model Governance: Smart contracts and decentralised autonomous organisations (DAOs) can be employed to govern AI models' behaviour and decision-making processes in a transparent and democratic manner. Community-driven governance mechanisms help mitigate biases, ensure accountability, and promote ethical AI practices.
Federated Learning: Federated learning, a decentralised approach to AI training, involves training models across multiple decentralised devices or nodes without centrally aggregating raw data. Blockchain technology can enhance federated learning by securely coordinating model updates, maintaining data privacy, and incentivising participation through token rewards or staking mechanisms.
Decentralised AI Oracles: Oracles play a crucial role in connecting smart contracts with real-world data for AI applications, such as prediction markets or decentralised finance (DeFi). Decentralised AI oracles leverage blockchain-based consensus mechanisms to verify and deliver trusted data inputs, enhancing the reliability and autonomy of AI systems.
Tokenised Incentives for AI Collaboration: Cryptocurrencies and tokens can incentivise collaboration and knowledge sharing within the AI community. Tokenised incentive mechanisms, such as token rewards, bounties, and staking, encourage researchers, developers, and data providers to contribute their expertise, data, and computational resources to AI projects, driving innovation in a decentralised manner.
Immutable AI Model Repositories: Decentralised storage platforms built on blockchain technology ensure the integrity and immutability of AI model repositories. By storing AI models and associated metadata on decentralised networks, developers can securely preserve version history, track contributions, and facilitate reproducibility, fostering open collaboration and knowledge sharing.
Decentralised AI Governance Frameworks: Community-driven governance frameworks powered by blockchain enable stakeholders to collectively steer the development and deployment of AI technologies. By leveraging decentralised decision-making processes, AI governance frameworks can accommodate diverse perspectives, address regulatory compliance, and promote inclusive and ethical AI development.
These are just some of the key areas where crypto and Web3 technologies can support a more decentralised approach to AI development. By integrating decentralised principles into AI systems, we can create more transparent, accountable, and inclusive AI ecosystems that empower individuals and communities while mitigating centralisation risks.
So, there you have it: there’s still hope. And the hope lies in mitigating the problems associated with the centralising forces of AI 2.0 by forging ahead with the decentralising potential of AI3 — whilst we still have time (and ChatGPT) on our side…
JOHN DEATON TAKES ON ELIZABETH WARREN
This story amused me greatly! John Deaton, the lawyer that filed a motion in 2021 to intervene on behalf of thousands of XRP holders in the SEC’s case against Ripple, has just thrown his hat in the ring to run against Senator Elizabeth Warren for the Massachusetts Senate seat…
In brief:
Deaton teased a potential senate bid back in December when he asked his 300,000 X followers if he should buy a home in Massachusetts and run for senate.
“I’m not suggesting I would win, but how I would love to confront her,” he added at the time.
Warren is a vocal critic of crypto, previously claiming that crypto is widely used by terrorist organisations in the Middle East for funding.
The claim has since been debunked by both blockchain analytics provider Elliptic and, just last week, the Treasury Department itself.
Link to full story: https://blockworks.co/news/elizabeth-warren-challenger-senate
UNISWAP PRICE SKYROCKETS AFTER FOUNDATION PROPOSES LONG-AWAITED STAKING REWARDS
A new Uniswap governance proposal is being regarded by the cryptosphere as having much wider implications for the rest of the space. Many commentators on X have suggested that it implies that Uniswap, which is at the forefront of DeFi, is no longer afraid of the regulatory implications of distributing protocol revenue to token holders. And some have even gone as far as to suggest that this likely means a Coinbase victory over the SEC is imminent…
In brief:
Uniswap (UNI) has surged in price following a proposal by the Uniswap Foundation to overhaul the protocol's governance system. The 16th-largest cryptocurrency by market cap is up by just over 50% in 24 hours, according to CoinGecko. Its price now stands at $11.20.
That jump makes it the best performing token today among the top 100 cryptocurrencies. The reason is due to a proposed "large-scale upgrade to Uniswap protocol governance to incentivise active, engaged, and thoughtful delegation," developers said.
In short, users who stake their tokens—that is, pledge them to the network—will receive rewards.
The Uniswap Foundation added that it believes that "UNI token holders will be incentivised to choose delegates whose votes and engagement with the protocol will lead to the Protocol's growth and success."
Link to full story: https://decrypt.co/218760/uniswap-price-skyrockets-proposal-uni-staking-rewards
TRUMP NO LONGER ANTI-BITCOIN?
Yet another indication that crypto is rapidly becoming an election issue this year in the United States: Donald Trump has softened his stance on Bitcoin…
In brief:
Former United States President Donald Trump has changed his tune on Bitcoin. From making anti-Bitcoin statements and calling it a scam during his tenure as president, Trump now says BTC is something he can live with and acknowledges its growing demand.
During a recent interview on Fox News, Trump was asked about his views on the rise of Chinese digital currency and whether the right way to counter it is through a decentralised currency network, such as Bitcoin.
Trump, in response, said the U.S. dollar is still the top currency for him but, at the same time, acknowledged the growing popularity and adoption of Bitcoin:
“I like the dollar, but many people are doing it [using Bitcoin], and frankly, it’s taken a life of its own. You probably have to do some regulation, as you know, but many people are embracing it. And more and more, I’m seeing people wanting to pay Bitcoin, and you’re seeing something that’s interesting. So I can live with it one way or the other.”
Link to full story: https://cointelegraph.com/news/trump-no-longer-anti-bitcoin
KRAKEN DECRIES ‘EXPANSIVE NEW THEORY’ BEHIND SEC LAWSUIT, CALLS FOR DISMISSAL
In brief:
Crypto exchange Kraken filed a motion in federal district court late Thursday, asking that the court dismiss the SEC’s case filed in November.
"The law is clear," a Kraken spokesperson told Decrypt. “None of the assets in the SEC’s complaint constitute ‘investment contracts,’ and Kraken is not a securities exchange."
The motion, filed in Northern California District Court, picks apart the SEC's most recent claims—which came nine months after the agency and Kraken reached a settlement over prior allegations related to crypto staking.
Link to full story: https://decrypt.co/218673/kraken-motion-to-dismiss-sec-case-securities
TOM EMMER CALLS BTC MINER SURVEY AN “ABUSE OF POWER”
In brief:
House Majority Whip Tom Emmer, R-Minn., has a bone to pick with the U.S. Office of Management and Budget over its recent move to authorize collecting data from bitcoin mining companies in the U.S.
Emmer said he wants answers from the OMB after the government agency allowed last month for the Energy Information Administration (EIA) — a statistical and analytical agency within the U.S. Department of Energy — to begin surveying certain crypto mining firms on their electricity consumption. Emmer laid out concerns over OMB's use of its emergency approval authority and said bitcoin miners don't currently present a threat to public safety, in a letter sent to OMB Director Shalanda Young this week.
"Bitcoin mining is not a threat to public safety. Period," Emmer said in a post on X on Thursday. "The OMB's abuse of its emergency powers to attack Bitcoin miners demands an explanation."
Emmer said that the agency has to prove to the OMB that "public harm is reasonably likely," among other criteria.
"With the assistance of the OMB, the EIA seems to be enforcing the Biden administration's regressive policy position against energy consumption and applying it subjectively to the digital asset industry," Emmer said in his letter.
Link to full story: https://www.theblock.co/post/278879/republican-rep-tom-emmer-calls-government-agencys-bitcoin-miner-survey-an-abuse-of-power
WALMART EXPANDS AVAILABILITY OF PUDGY TOYS TO 1,100 MORE STORES
An interesting crossover project in terms of NFTs and IP: Pudgy Penguins, an NFT collection, has crossed over into the mainstream, with Walmart doubling down on a line of cuddly toy merch being sold in its stores…
In brief:
Walmart, the world's biggest retailer, has expanded the availability of Pudgy Toys — the NFT collection Pudgy Penguins' toy line — to 1,100 more U.S. stores, bringing the total to 3,100 stores nationwide.
Walmart is introducing 30 new Pudgy Toys, including Lil Pudgys Igloo Collectibles and Action Figures, each priced between $2.99 and $11.97. Pudgy Toys debuted at Walmart last September, initially appearing in 2,000 U.S. stores.
"The expansion into additional Walmart locations, alongside exclusive toy offerings, underscores the enduring appeal of Pudgy Penguins characters," Luca Netz, CEO of Pudgy Penguins, said in a statement.
Each Pudgy toy comes with a QR code that provides access to Pudgy World — a multiplayer digital social platform where users can build their "Forever Pudgy" characters, play mini-games, and interact with other users. Pudgy World is based on the Ethereum Layer 2 network zkSync Era and is currently in open beta mode.
Link to full story: https://www.theblock.co/post/278347/walmart-pudgy-penguins-expansion
CRYPTO-CENTRIC SUPER PAC SWELLS ITS FUNDS
In brief:
As the 2024 election season gets underway, crypto companies and their executives seeking more influence in Washington have opened their chequebooks.
Coinbase, Ripple Labs and Andreessen Horowitz are some of the top donors in the space, giving a collective $68.5 million across three crypto-focused super political action committees (PACs) since their launch, latest filings with the Federal Election Commission show.
Affiliated super PACs Fairshake, Protect Progress and Defend American Jobs raked in a collective $84.8 million in 2023.
The three committees intend to back “leaders who support American crypto and blockchain innovation and responsible regulation in the forthcoming 2024 elections,” a spokesperson told Blockworks in December.
Link to full story: https://blockworks.co/news/crypto-super-pacs-update-coinbase-a16z
INTERESTING OP-ED ON REMOTE WORK AND WEB3
In brief:
If you were to listen to recent comments from the CEOs of many major companies, you could be forgiven for thinking that the era of remote work is over.
Remote work is increasingly characterized as an emergency measure which — though useful during the height of the pandemic — was doomed to fail in the long run. There seems to be a growing consensus from traditional firms that a return to office is necessary for securing high productivity.
However, we in Web3 hear a different consensus. Web3 talent has predominantly worked remotely — and for those working on these cutting-edge technologies, a return or move to in-person work is unthinkable.
Web3 talent is at the forefront of innovation within the technology industry — so Web3 professionals will increasingly lead expectations for the future of work and the tech workplace environments in the years to come.
More traditional firms need to be aware of this ongoing shift in expectations — or risk stifling innovation and not being able to tap into global talent pools.
Link to full op-ed: https://blockworks.co/news/web3-remote-work-important
That’s all for now folks — until next week! 🥸👍