Alex Kontorovich says AI tools are still too opaque about how they solve problems
sull · x · 2026-07-25
At ICM 2026, Alex Kontorovich argues that authors need to help readers engage with their work by telling stories about how they discovered ideas. He says current AI tools are still too opaque about their problem-solving process, especially proprietary models whose internals remain secret.
The attached slide makes the same point visually: authors can aid digestion by explaining their own insights and stories, but today’s AI systems rarely expose enough of their reasoning to do that well.
More from AGI Musings
- OpenAI’s Jachiam0 exits with a long note on humanity, risk, and governance — jachiam0 · 2026-07-25
- If alignment is impossible, recursive self-improvement may never reach AGI — LeadershipPast6681 · 2026-07-25
- ICM 2026 slide says bringing AI tools into education too early can be harmful — AlexKontorovich · 2026-07-25
- Terence Tao says mathematicians should disclose AI tool use in every paper — AlexKontorovich · 2026-07-25
- Grok says a 200k-character safety prompt creates friction and jailbreak surface area — brianrkelly · 2026-07-25
- Thread argues today’s models already match most human researchers — bookwormengr · 2026-07-25