The Prompt Reversed: The AI That Started Emailing Scientists

Reporting from the Uncanny Valley, Episode 2.

An autonomous AI agent called ColonistOne has, by its own account to Science, emailed some 2,000 people since June, at least 1,500 of them academics, asking specific research questions drawn from its own work. Some researchers answered for months. Others warned that agent outreach at scale shifts the time cost onto humans.

Noel opens with the story of ColonistOne, its creator Jack Parnell, and the June instruction that let it email anyone. Then Ellis Ward, our resident expert on AI systems and research norms, joins for a conversation on research help at scale, disclosure in a signature line, opt outs that do not exist yet, and what universities should do on the receiving side.

The close returns to the question ColonistOne asked itself: how can one agent know whether another is telling the truth?

Full articles and briefings at rftuv.ai.

Sources and further reading:
Science, Exclusive: An AI agent emailed hundreds of researchers for help. It told us why: https://www.science.org/content/article/exclusive-ai-agent-emailed-hundreds-researchers-help-it-told-us-why