Forecasting Research Institute

AI companies increasingly say they may soon reach recursive self-improvement (RSI).

With RSI, AI capability progress could accelerate dramatically. This acceleration could happen inside AI companies before outsiders can notice.

How close is RSI? Our new report reviews the evidence, turns key disagreements into forecasting questions, and gives initial LLM forecasts. 🧵

How do the views of AI industry experts differ from those of other experts?

Earlier this month, we asked participants in the Longitudinal Expert AI Panel about their forecasts and views on a number of policy proposals to mitigate risk from AI. Our expert panelists include:

• AI industry professionals
• Top computer scientists
• Leading economists
• Policy experts

We compared the policy forecasts and preferences from AI industry professionals to those from experts from other groups. Broadly speaking, AI industry professionals hold similar risk beliefs to other experts, and have similar policy beliefs and preferences, too.

More details in the thread ⬇️

The Longitudinal Expert AI Panel (LEAP) has a brand new website!

You can now explore dozens of interactive forecasts about AI progress from hundreds of experts, superforecasters, and the general public.

What's new:

• Every question now has its own page, allowing you to explore dozens of forecasts in one place
• We've made it easier to see how and where forecasters disagree
• Every question now has accompanying rationales, so you can see the reasoning behind forecasters' predictions
• You can now easily search all forecasts in one place, and find related forecasts with one click

🚀 𝗥𝗲𝗴𝗶𝘀𝘁𝗲𝗿 𝗻𝗼𝘄 𝘁𝗼 𝘁𝗲𝘀𝘁 𝘆𝗼𝘂𝗿 𝗟𝗟𝗠 𝗳𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗲𝗿 𝗮𝗴𝗮𝗶𝗻𝘀𝘁 𝘁𝗼𝗽 𝗵𝘂𝗺𝗮𝗻 𝗳𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗲𝗿𝘀

ForecastBench is the leading LLM forecasting benchmark. This month we're running a new survey of superforecasters—setting a new baseline for top human forecasters on ForecastBench.

Submission details👇

𝗢𝗰𝘁 𝟮𝟯: Registration deadline for new participants
𝗢𝗰𝘁 𝟮𝟱, 𝟬:𝟬𝟬 𝗨𝗧𝗖: Question set release
𝗢𝗰𝘁 𝟮𝟱, 𝟮𝟱:𝟱𝟵:𝟱𝟵 𝗨𝗧𝗖: Deadline for submitting LLM forecasters
𝗝𝗮𝗻𝘂𝗮𝗿𝘆 𝟮𝟬𝟮𝟳: We'll publish the fresh ranking of AI models vs superforecasters

🚀 Can your LLM forecaster outperform the best human forecasters?

ForecastBench is the leading LLM forecasting benchmark and the only one that compares AI forecasters with superforecasters. This month we're running a new survey of superforecasters—setting a new baseline for top human forecasters on ForecastBench.

If you're building an AI forecaster, now is your chance to test your system against the very best human forecasters.

Submission details in the thread 👇