Eli Lifland
Eli Lifland is a research lead at the AI Futures Project and a co-leader of the forecasting team Samotsvety. His early technical work includes TextAttack, a shared framework for testing language-model weaknesses. At Ought, he worked on the research assistant Elicit and coauthored RAFT, a benchmark for practical text classification with few examples. He later coauthored AI 2027, a scenario with predictive intent and two endings. With Brendan Halstead, he shares conceptual development credit for the AI Futures Model. His recent work includes a forecasting survey, changes to timeline methods, and AI 2040: Plan A. The last project is principally a policy proposal. Forecasts depend on assumptions, and their accuracy is unmeasured.
- Now at
- AI Futures Project and Samotsvety
- Works
- 9
- Citations
- 817
- h-index
- 6
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2020
Trying to break a text model
He helps create TextAttack with John Morris and other researchers.
-
Ought; 2021 benchmark
From research assistant to research test
At Ought, he works on Elicit and coauthors the RAFT text-classification benchmark.
-
April 2025
The future becomes a scenario
He joins four colleagues in writing the AI 2027 scenario.
-
2025–2026
The machinery behind the dates
He and Brendan Halstead share conceptual development credit for the AI Futures Model.
-
2026
A policy scenario joins the forecast
He coauthors AI 2040: Plan A, a proposed path for international coordination.
Sources and credits
Photo: Eli Lifland; image supplied by www.elilifland.com. License: Not stated by source. Source: elilifland.com. Cropped to a square and resized. Ask us to remove or replace this photo.