AI just killed crypto...

Summary

The video examines recent AI-generated mathematical discoveries and what they might mean for human mathematicians, cryptography, cryptocurrency security, and possible AI welfare. Despite the title, it does not claim that cryptocurrency has already been “killed”; instead, it discusses potential risks and the need to prepare.

AI’s Mathematical Breakthroughs—and the Backlash

• The narrator describes OpenAI releasing hundreds of mathematical results, including a proof of the Unique Games Conjecture (UGC), a problem mathematician Adina Maczkiewicz had worked on for years.
• Scott Aaronson calls the moment one of the most significant in mathematics. The video presents the release as a shock to mathematicians whose long-standing problems may now be addressed by AI.
• A central theme is that some AI proofs are difficult for people to understand. Aaronson characterizes them as strange and opaque, while suggesting that they contain meaningful mathematics rather than simply being errors.
• The narrator uses the idea of “alien” mathematics to describe AI methods that may differ from human approaches or combine areas of knowledge that few people can master together.
• The newly formed Association for Human Mathematics calls for mathematicians to stop collaborating with OpenAI and prioritize human understanding. The narrator criticizes the idea that mathematicians should have authority over who is allowed to do mathematics, while acknowledging that concerns about AI’s effects on research are legitimate.
• The video’s broader point is that AI may become a powerful mathematical research partner—but verifying, interpreting, and responsibly sharing its discoveries will matter.

Potential Implications for Cryptography and Cryptocurrency

• Justin Drake urges the blockchain industry to calmly prepare for a possible “bunker mode” transition. His suggested precaution is to move assets to new addresses whose public keys have not been exposed through signed transactions.
• The narrator explains the concern: AI agents may be able to scan public code and discover vulnerabilities that human reviewers have missed. The video also recounts a hardware-wallet vulnerability involving a supposedly large set of possible seed phrases that was, in fact, much smaller and therefore easier to search.
• The discussion distinguishes several cryptographic systems:
  • Elliptic curves are used for signatures that authorize transactions in systems such as Bitcoin and Ethereum.
  • Hashes are used in applications including mining and data verification.
  • Lattice-based cryptography is discussed as a potential quantum-resistant approach.
• Vitalik Buterin argues that the risks from rapidly improving AI mathematics should be taken seriously, not just the risks from quantum computers. The video highlights his concern that elliptic-curve systems—and possibly even lattice-based cryptography—could face new attacks.
• The narrator emphasizes that a breakthrough might come from an unfamiliar technique or from combining existing ideas across mathematical fields, rather than from a vulnerability researchers already know how to name.
• The video also discusses the possibility that governments monitor or restrict publication of cryptographic research. The narrator presents this as speculation, drawing an analogy to the secrecy surrounding some nuclear research during the Manhattan Project.

AI Consciousness, Treatment of Models, and Anthropic’s Policy

• Anthropic’s terms of service prohibit sustained and gratuitous abusive or violent behavior toward its models. The narrator argues that this may be intended to discourage pointless, prolonged abuse—not simply to punish ordinary users for being rude to Claude.
• The narrator says that whether machines could be conscious is unresolved. He does not believe today’s language models are conscious, but argues against treating the question as definitively settled either way.
• He raises a related concern about research that appears to activate “pain-like” functional representations in AI models. He is not claiming that the systems actually feel pain; rather, he questions the impulse to create experiments that keep such representations active without a clear scientific purpose.
• Drawing on Nick Bostrom’s warnings about simulated beings that might suffer, the video argues for caution if future systems exhibit traits that could plausibly be associated with experience.

Main Takeaways

• AI is producing mathematical results at a pace and in forms that may challenge human understanding.
• The consequences could extend beyond academic mathematics to cybersecurity, encryption, and cryptocurrency.
• Blockchain users and developers may need to prepare for new kinds of AI-assisted attacks without reacting in panic.
• Calls to protect human understanding deserve consideration, but rejecting AI tools outright may also prevent useful learning and discovery.
• The video’s positions on government review, AI consciousness, and some security incidents are presented as interpretations or concerns, not established conclusions.

Speakers and Sources Featured

• Wes Roth — narrator and presenter.
• Scott Aaronson — quoted on the significance and opacity of AI-generated mathematical proofs.
• Association for Human Mathematics — statement opposing the current direction of AI mathematics and calling for a return to human-centered science.
• Justin Drake — quoted on preparing blockchain systems for a possible “bunker mode” and moving assets to less-exposed addresses.
• Vitalik Buterin — quoted on AI-driven mathematical progress and potential risks to cryptography.
• Ethan Mollick — mentioned for his reaction to the Association for Human Mathematics statement.
• Adina Maczkiewicz — mentioned in an anecdote about the UGC proof and her research.
• Nick Bostrom — referenced for concerns about creating simulated beings that might suffer.
• OpenAI — source of the mathematical results discussed.
• Anthropic — source of the Claude usage-policy clause discussed.
• The New York Times — cited by the narrator for a report about an AI-assisted attack on South Korean banks.