Artificial intelligence · around 2029

Autonomous AI researchers

Projection

AI agents run a research programme end to end, from hypothesis to published result.

Agents that read literature, design experiments, write and run code, and analyse results would compress research cycles from months to days. They would accelerate AI research itself as well as other sciences.

Why it’s plausible: Reasoning models already complete multi-hour software and research tasks; full autonomy depends on reliable long-horizon planning, verification of results and access to real experiments.

See it in the living treeWikipedia