·
·
3 min read
Capriole Investments founder Charles Edwards declared on October 21 that artificial intelligence has surpassed the best human intelligence in specific fields, marking a point where human cognition is no longer the apex predator. Edwards made this assessment on X following a month in which AI models achieved perfect scores in elite mathematics competitions, a milestone he described as an evolutionary turning point.
Perfect Scores and Research Output
The claim is grounded in recent performance at the 2026 International Mathematical Olympiad (IMO). Models developed by Huawei and Xiaohongshu achieved a perfect score of 42 out of 42. According to official IMO results, only seven of the 666 human contestants managed to achieve the same perfect score. This parity between machine and elite human performance in a traditionally human-dominated domain forms the core of Edwards’ argument.
Beyond the IMO, OpenAI provided an unreleased model with approximately 4,000 open mathematical problems in early October. The model returned 722 research manuscripts, which were grouped into 372 families of results. Stefano Gogioso, an Oxford lecturer, summarized the implication of these results by stating that mathematical rigour is now free. While OpenAI cautioned that not every proof generated by the model has been independently verified, Gogioso noted that many of the solved problems were already close to a human solution.
Market Impact and Cryptographic Concerns
The implications of this mathematical advancement extend into the labor market and digital security. IMF Managing Director Kristalina Georgieva has warned that young job seekers face the greatest risk from AI. Stanford researchers found that employment for 22- to 25-year-olds in highly AI-exposed jobs is currently 19% below its expected trend, a gap that stood at 15% a year earlier. This decline is attributed to fewer new hires rather than mass firings. Additionally, US employers cited AI in 120,136 planned job cuts through September, representing about 21% of all 2026 cuts, according to Challenger, Gray & Christmas.
In the cryptocurrency sector, Ethereum Foundation researcher Justin Drake argued that mathematical superintelligence has arrived. He urged large holders to prepare a “bunker mode” by moving funds to never-used addresses. The concern is that AI-driven mathematical capabilities could weaken Bitcoin’s wallet cryptography without the need for a quantum computer, as every Bitcoin address that has sent funds exposes its public key on-chain. However, Ethereum co-founder Vitalik Buterin advised against rushed migrations, and Coinbase cryptographer Yehuda Lindell dismissed the warning as alarm without evidence. Currently, there is no sign that the underlying cryptography has been broken.
Despite these advancements, AI performance remains inconsistent in broader professional contexts. Scale AI’s Remote Labor Index, which tests agents on real freelance projects, shows that the top agent delivers professional-quality work on about 21% of tasks. This figure has improved from 2.5% when the index launched a year ago. Former Meta chief AI scientist Yann LeCun continues to argue that human-level AI remains years away, highlighting the jagged nature of current AI capabilities.
Frequently Asked Questions
What score did AI models achieve at the 2026 IMO?
Models from Huawei and Xiaohongshu scored a perfect 42 out of 42 at the 2026 International Mathematical Olympiad.
Why did Justin Drake suggest moving Bitcoin funds?
Justin Drake urged large holders to move funds to never-used addresses as a precaution against potential weaknesses in Bitcoin’s wallet cryptography caused by AI-driven mathematics.
