Dario's Accurate Predictions Shift Timeline Expectations

Mike Krieger has fundamentally changed his perspective on AI timelines since joining Anthropic. His experience watching predictions materialize has shifted his thinking from skepticism to conviction about rapid AI advancement.

"I've taken the timelines a lot more seriously now," Mike explains, recounting how Anthropic's co-founder Dario made predictions that initially seemed ambitious but consistently came true. When Dario predicted they'd reach 90% on coding benchmarks by 2025 (from 50% at the time), they've already hit 72% - tracking exactly as predicted.

This pattern of accurate forecasting has made Mike view AI development timelines with greater urgency. The experience of having two browser tabs open - one with the "AI 2027" paper and one with his product strategy - created a surreal moment where he wondered: "Wait, am I the character in the story?"

For product leaders, this perspective shift has significant implications:

• Capabilities that seem years away may arrive much sooner than expected
• Planning cycles should account for rapid capability jumps rather than linear improvement
• The gap between what models can theoretically do and what products currently enable represents immediate opportunity
• Product development bottlenecks are shifting from code generation to upstream activities like strategy and alignment

This timeline compression also influenced Mike's career decision: "Looking out at my kids, I'm like, alright, they're going to grow up in a world with AI. It's unavoidable. What is the thing where I can maximally apply my time to nudge things towards going well?"