A new Moore's Law for AI agents
A new trend in AI development suggests that the capabilities of AI agents are growing exponentially. Researchers found that the length of coding tasks that AI can successfully complete is doubling every few months, with predictions indicating they could handle tasks equivalent to a month of work by 2029. This rapid advancement could lead to transformative changes in AI research and other domains.
- ▪AI agents can now autonomously complete coding tasks that take humans over fourteen hours.
- ▪The length of tasks that agents succeed at 50% of the time is growing exponentially, with predictions for 2027 indicating they could handle an 8-hour workday.
- ▪The trend of increasing task length has accelerated, with time horizons doubling every 4 months recently.
Opening excerpt (first ~120 words) tap to expand
Last updated March 2026A new Moore's Law for AI agents When ChatGPT came out in 2022, it could do 30 second coding tasks. Today, AI agents can autonomously do coding tasks that take humans over fourteen hours.Scroll to continueThe length of coding tasks frontier systems can complete is growing exponentially – doubling every 7 months.This trend was discovered by researchers at METR. They took the most capable agents from 2019 to 2026, and tested them on about 230 tasks: mostly coding tasks, with some on general reasoning. Then, they compared the agent's success rate to the length of each task – how long it takes human professionals to complete, ranging from under 30 seconds to over 8 hours.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at AI Digest.
Discussion
0 commentsMore from AI Digest
