How crypto networks might defend humanity against "bad" AI
How crypto networks might defend humanity against "bad" AI
Look, we warned you the future would be cringe.
Happy December! Some VERY SERIOUS and VERY ONLINE people are getting religious about AI and the future of technology. Who knew this timeline could keep. Getting. Weirder.
In this issue:
How crypto fits into the AI religions
Anonymous proof of Indian citizenship, featuring zero-knowledge crypto
Odds and Ends
What memeplexes have to do with it (glitch/acc)
Lately, some of the arguments over the future of technology, particularly artificial intelligence, have turned religious. At least two separate ideologies have emerged, at odds over how rapidly humanity should move forward with AI development—and the debate that ensued last month after OpenAI fired CEO Sam Altman felt like the opening skirmish of what may be a drawn-out conflict.
What’s especially intriguing about the Black Mirror-esque drama is how it casts the role of crypto networks.
Which brings us to Marc Andreessen’s recent re-tweet of a fake person known as Charlotte Fang. Andreessen, as you may know, is a high-profile venture capitalist. Fang, who has graced our pages before, is a pseudonym for an actual person named Krishna Okhandiar, one of the brains behind Remilia, the embattled company that created the Milady NFT project. The Miladys, a swarm of internet trolls who hide behind anime-style profile pics, have a reputation for posting bizarre and often transgressive memes.
“If I was a safetyist doomer I’d spend a lot more serious time and thought thinking about how to critique the argument for ‘AGI rights’ as a conscious entity because that’s a memeplex that absolutely will spawn into the exact same midtwit rhetorical rationalism you’re caught in,” read the post Andreessen amplified.
If that reads a gibberish to you, pull up a chair because we’re going to break it down. And before you ask “Do we have to?” unfortunately the answer is yes—this sort of language is currently proliferating in the online circles of Very Serious Tech people. Andreessen himself helped set this off in October with his “Techno-Optimist Manifesto.”
First, the terminology:
AGI: stands for artificial general intelligence, the term for a future AI system that may have intellectual abilities equal to or beyond those of humans. It is fictional, and has been discussed as a concept for around, like, six decades? A lot of otherwise smart people seem to think it’ll be here in four to seven years.
Doomer: this is what the new techno-optimists call someone who believes that AI is risky because it could cause profound harm to society, and with that in mind we should really slow our roll.
Midtwit: The Urban Dictionary says a midtwit is “someone who is around average intelligence but is so opinionated and full of themselves that they think they’re some kind of genius.” Are the people who use this word also the most likely to fit this description? I will leave that judgment to people with above-average intelligence.
So, “Charlotte’s” tweet is saying that the same “doomers” propagating the idea that AI could extinguish humanity will, in the future, be arguing that AI systems should have rights—or at least that this future “memeplex” (we’ll come back to this one) will spring from the same brand of “midtwit” logic.
Doomers’ desire to pump the brakes on AI, in Andreessen’s eyes, is part and parcel of “nihilistic wish, so trendy among our elites, for fewer people, less energy, and more suffering and death.” Oof. Techno-optimists “believe in accelerationism,” Andreessen writes: “the conscious and deliberate propulsion of technological development.” (The “e/acc” tag you may have seen gaining steam on Twitter stands for “effective accelerationism.”)
According to Andreessen, “Our present society has been subjected to a mass demoralization campaign for six decades—against technology and against life—using varying names like ‘existential risk,’ ‘sustainability,’ ‘ESG,’ ‘Sustainable Development Goals,’ ‘social responsibility,’ ‘stakeholder capitalism,’ ‘precautionary principle,’ ‘trust and safety,’ ‘tech ethics,’ ‘risk management,’ ‘de-growth,’ and ‘the limits of growth.’”
Let’s not unpack all of that in this post. Instead let’s return to Charlotte Fang’s term: memeplex. According to Wikipedia, “memeplexes…consist of groups of memes that are typically present in the same individual.” Andreessen seems to think he’s calling out a bad memeplex—the rather broad list of terms above, which together he says leads to “deceleration”—and that the techno-optimism/effective accelerationism memeplex should replace it.
What does all this shit have to do with crypto networks? We’ve already established that Fang and their Remilia colleagues are deeply crypto. And recently, according to Ryan Broderick at Garbage Day, Remilia has “rebranded” as a “TikTok subculture” called BRG, that promotes accelerationism. But crypto’s role here has the potential to run deeper.
After OpenAI’s board forced Sam Altman out last month, leading to a short, chaotic period in which much of the company threatened to quit before Altman was eventually brought back, many blamed “decels” (short for “decelerationists”) for the debacle. The debate apparently inspired Ethereum co-creator Vitalik Buterin to jump into the fray bearing one of his signature meta-messages: things are way more complicated than we are making them out to be!
In a long essay entitled “My techno-optimism,” Buterin advocates for moving past “technology as a whole is good” and focusing on “which specific technologies are good?” Surely, he acknowledges, some potential advances—for instance, certain drugs and vaccines—are being held back due to the cautiousness of policymakers. But AI is “fundamentally different from other tech, and it is worth being uniquely careful” about it, he argues.
One specific change Buterin fears AI may bring about is that “bad” centralized governments or other powerful entities will be able to deploy vast and sophisticated spying systems. “With modern surveillance to collect information, and modern AI to interpret it, there may be no place to hide.”
He is far from alone in this concern. “The very people claiming to protect us will use (AI) to systematically combine every shred of public and private data we emit—so there’s nowhere, no thought, that we can hide,” Johns Hopkins University cryptographer Matthew Green predicted recently.
This and other risks of AI imply that we need “defensive technologies,” Buterin argues. (He calls his philosophy “d/acc” for “defensive accelerationism”)
Blockchains and cryptography are defensive, Buterin says. “Blockchains let us create economic and social structures with a ‘shared hard drive’ without having to depend on centralized actors,” he writes, while zero-knowledge proofs can allow users to prove their true human identity and credentials without leaking private data to AI surveillance bots.
These technologies can also let people—future versions of today’s Miladys, perhaps—evade the AI surveillance machine by anonymously sharing their memes, all while proving that they are real humans and not just AIs programmed to shitpost. Just imagine!
Glitch isn’t picking sides. But we tend to agree that the rise of AI is likely to make crypto more important for defensive purposes—ones that we mostly haven’t needed it for yet. If you’d like to keep your freedom to meme, that’s probably good news. —Mike Orcutt
WARNING: TECHNICAL
Name: Anon Aadhaar
Brain(s) behind it: Privacy and Scaling Explorations (PSE), a research and development lab supported by the Ethereum Foundation. Ethereum supports PSE to create open-source cryptographic “primitives,” or building blocks. The logic is that more effective developer tooling will help the Ethereum ecosystem financially by expanding the range of blockchain use cases.
What it is: An anonymous “proof of identity” system that lets Indian citizens cryptographically prove that they have an Aadhaar card—a government-issued, biometric-linked ID—to access services that require Indian citizenship without revealing private data. Anon Aadhaar pulls this off by using zero-knowledge proofs, which make it possible to share secrets without revealing their content.
The Aadhaar project is the largest digital identity program in the world. Around 90% of the Indian population, roughly 1.2 billion citizens, are enrolled. Today’s version of the Aadhaar card uses biometric data, including fingerprints and iris scans to provide citizens with access to various services like housing applications and loan agreements. But right now, to access those services citizens need to clear know-your-customer (KYC) checks directly with third-party service providers using their Aadhaar card, which involves sharing a bunch of sensitive personal information.
The problem, says Andy Guzman, one of Anon Aadhaar’s product managers, is that some of these service providers may not be trustworthy stewards of such information. In part, that’s because centralized databases, like the ones that banks or insurance companies use, are often the target of cyberattacks. Or the firms might sell that data openly to governments or other entities that may have hostile political motivations. If they can prove they have an Aadhaar card without sharing personal data, Indian citizens can access public services without the same risks. That could be especially critical for people like journalists or whistleblowers who might fear government reprisals or other suppression of rights.
Anon Aadhaar works similarly to Zupass, which Lucy used last month to anonymously prove that she was registered for events at Ethereum’s DevConnect in Istanbul. Both platforms feature a programmable cryptographic building block called proof-carrying data, or PCD: essentially a program built using zero-knowledge proofs that transfers data anonymously between two parties. Guzman and the rest of the team behind Anon Aadhaar have released the code as well as a software development kit that helps blockchain application developers integrate it with their own smart contracts. They’ve also brainstormed some possible applications, from private wallets to a Telegram private group that would be open only to Indian citizens.
Anon Aadhaar is still a research project in search of a real-world application, but that could change soon. Guzman says that in December, PSE hopes to use the platform as the check-in system at ETHIndia, a developer conference that will host over 2,000 people. —Sam Venis
ODDS/ENDS
Is crypto really “back”? Well, the number has just kept going up since the last time we asked.
Forbes has revealed that @BasedBeffJezos, who Marc Andreessen calls a “patron saint of techno-optimism” is a former Google quantum computing engineer and founder of an AI startup called Extropic. His real name is Guillaume Verdon. He says he’s “just a gentle Canadian.” Yesterday, Extropic revealed that it has raised $14 million to build a “computing paradigm which harnesses the power of out-of-equilibrium thermodynamics to fundamentally merge generative AI with the physics of the world.
The Department of Justice’s $4 billion enforcement action against Binance, which led to longtime CEO Changpeng Zhao stepping down, gives the crypto industry a chance to close “a chapter of history,” according to Coinbase CEO Brian Armstrong.
Blockchain policy advocacy group Coin Center has published an extensive report arguing that the Bank Secrecy Act, the US law from which the government draws authority to surveil electronic financial transactions, delegates more authority to the Executive Branch than is Constitutionally allowed.
A Reuters analysis of crypto seizures announced by Israel since 2021 shows that militant groups in the Middle East have recently been leaning more heavily on the crypto network Tron. It’s not clear exactly how this fits into the larger debate about crypto’s role in financing terrorism.
Spanish police have arrested a man wanted by the US for allegedly conspiring with Ethereum developer Virgil Griffith to help North Korea evade sanctions using blockchains. Griffith is serving a five-year prison sentence that began last year.
On-chain game developer Lattice has developed a new kind of Layer 2 blockchain network, called Redstone, which it says will help scale transactions in on-chain game worlds.
Argus, a venture-backed blockchain game startup founded by original Dark Forest contributor Scott Sunarto, is launching a new version of the crypto cult classic game on its custom Layer 2 network. The title of the new game is Dark Frontier.
Follow us on Twitter if you want.