Jack Clark moves AI self-improvement forecast up to 2027
What happened
The Decoder reported on October 25, 2026 that Anthropic co-founder Jack Clark moved his forecast for when AI systems could develop their own successors from 2028 to as early as 2027. It remains a forecast of what is possible, not achieved progress. One basis is P-Zero Research's TasteVal benchmark: Opus 5.5 reached a compute multiplier of 2.3, hitting expert level with under half the compute, with a statistical interval of 1.15 to 4.37.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- The DecoderAnthropic co-founder Jack Clark moves his prediction for self-improving AI forward to 2027
Anthropic 联合创始人 Jack Clark 将 AI 系统开发自身后继模型的可能时间从 2028 年提前至最早 2027 年。其依据之一是 P-Zero Research 的 TasteVal 基准，其中 Opus 5.5 的算力乘数达到 2.3，即以不到一半的算力达到专家水平，统计区间为 1.15 至 4.37。
Heat over time
Not enough continuous observations to draw a trend yet.