The original title is "Nobody Agrees When AGI Arrives. Build for the Shift Instead."

Short summary

AGI timeline estimates range from 3-4 years (Hassabis) to 2047 (researcher surveys), making specific-date planning a losing bet. Instead, builders should focus on three measured trends everyone agrees on: task-completion capability doubles every 4-7 months (METR), code execution is becoming commodified (SWE-bench near 99%), and agent infrastructure is standardizing via Model Context Protocol. The piece argues for planning around capability shifts rather than AGI arrival dates.

• •AGI timeline estimates vary by 10x, from 3-4 years to 2047, making date-based planning unreliable
• •Measured trends are clear: task length doubles every 4-7 months, SWE-bench hit ~99%, agent plumbing is standardizing
• •Build for the capability curve and standardized infrastructure, not for a specific AGI arrival date

Generated with AI, which can make mistakes.