Experts Question Technical Feasibility of AI Superintelligence Claims
A commentary piece examines the ongoing industry debate over artificial superintelligence (ASI), sparked by Anthropic researcher Jacob Coxon's resignation warning that Anthropic and OpenAI were pursuing self-improving AI systems that could endanger humanity. Anthropic executives, including alignment science head Evan Hubinger and CEO Dario Amodei, publicly acknowledged sharing these concerns, while the piece argues that the more fundamental question of whether superintelligence is technically achievable has been largely overlooked in these discussions.
GoKawiil's interpretation of the reporting above, not reported fact.
The framing suggests that public debate has focused heavily on the dangers and ethics of a hypothetical superintelligence while skipping over whether recursive self-improvement toward such a system is even engineeringly plausible. This gap could mean that policy, funding, and safety discussions are being shaped by speculative scenarios rather than grounded technical assessments, according to the piece's author. It also hints that some actors may have incentives—commercial or regulatory—to keep the conversation centered on existential risk rather than feasibility.
- Anthropic researcher Jacob Coxon resigned citing fears that Anthropic and OpenAI are racing toward dangerous self-improving AI.
- Anthropic's alignment science head and CEO Dario Amodei publicly agreed that AI could pose existential risks to humanity.
- The commentary argues the key unanswered question is whether superintelligence is technically achievable at all, not just whether it would be dangerous.
Nick Bostrom's Superintelligence (book) — This is the foundational text behind the exact debate the article describes, laying out the recursive self-improvement and existential-risk arguments that AI researchers are now publicly wrestling with. If you want to understand the intellectual roots of why people like Dario Amodei take these risks seriously, this book is essential background reading.
See Nick Bostrom's Superintelligence (book) on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: spectrum.ieee.org — Cal Newport, 2026-11-19
Published there as: “The AI Superintelligence Debate Has Serious Flaws”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.