What do experts and superforecasters think about the future of AI...
Forecasting Research Institute@Research_FRI
104 views
Jan 12, 2026
~3 min read
Advertisement
1
What do experts and superforecasters think about the future of AI research and development?
In Wave 4 of the Longitudinal Expert AI Panel (LEAP), we asked top AI experts to forecast progress in AI R&D, hiring, company valuations, data center buildout, and more.
Here’s what you need to know 🧵
In Wave 4 of the Longitudinal Expert AI Panel (LEAP), we asked top AI experts to forecast progress in AI R&D, hiring, company valuations, data center buildout, and more.
Here’s what you need to know 🧵
2
📈 AI benchmark progress is advancing faster than experts expect
AI performance on hard coding tasks is a useful indicator of potential capability increases in self-improving AI R&D: when models take an active role in improving AI itself.
The median forecaster significantly underestimated progress on LiveCodeBench Pro—a benchmark that tracks performance on tough programming tasks.
The median expert in our sample predicted state-of-the-art performance on LiveCodeBench Pro (Hard) of 14% in 2026 and 33% by the end of 2030. Since the survey closed, GPT-5.2 has already hit 33% on the benchmark.
A quarter of experts and superforecasters expect major progress on this coding benchmark, providing 50th percentile forecasts of at least 60% accuracy by 2030.
We plan to identify which forecasters were most accurate on questions like this to see what they believe about other topics.
AI performance on hard coding tasks is a useful indicator of potential capability increases in self-improving AI R&D: when models take an active role in improving AI itself.
The median forecaster significantly underestimated progress on LiveCodeBench Pro—a benchmark that tracks performance on tough programming tasks.
The median expert in our sample predicted state-of-the-art performance on LiveCodeBench Pro (Hard) of 14% in 2026 and 33% by the end of 2030. Since the survey closed, GPT-5.2 has already hit 33% on the benchmark.
A quarter of experts and superforecasters expect major progress on this coding benchmark, providing 50th percentile forecasts of at least 60% accuracy by 2030.
We plan to identify which forecasters were most accurate on questions like this to see what they believe about other topics.
3
🧑💻 Entry-level tech hiring may stay low for decades
Experts predict that the share of new hires with ≤1 year of experience at top tech firms will remain around 7% through 2040. Some respondents point to AI tools reducing the need for junior roles.
This is well below the 15% share of entry-level hires reached in 2019 and 2023.
Experts predict that the share of new hires with ≤1 year of experience at top tech firms will remain around 7% through 2040. Some respondents point to AI tools reducing the need for junior roles.
This is well below the 15% share of entry-level hires reached in 2019 and 2023.
4
💰Companies may become far more valuable per employee
Experts forecast that valuation per employee at top AI companies could rise to $20M by 2030 and $32M by 2040.
There is a lot of uncertainty in these forecasts. The median expert gives a 25% likelihood that the valuation-per-employee ratio in 2026 will fall to ≤$13 million or increase to ≥$21 million. This points to uncertainty in how much AI companies will be able to leverage AI systems internally to improve productivity.
Experts forecast that valuation per employee at top AI companies could rise to $20M by 2030 and $32M by 2040.
There is a lot of uncertainty in these forecasts. The median expert gives a 25% likelihood that the valuation-per-employee ratio in 2026 will fall to ≤$13 million or increase to ≥$21 million. This points to uncertainty in how much AI companies will be able to leverage AI systems internally to improve productivity.
5
💸 We also asked panelists to predict the calendar year in which at least three companies with five or fewer full-time employees first report individual valuations of ≥$10 billion.
The median expert gives a 50% chance that three such companies will exist by the end of 2032, and a 5% chance by the end of 2027.
H/t to @tylercowen whose interview with @sama partly inspired this question.
The median expert gives a 50% chance that three such companies will exist by the end of 2032, and a 5% chance by the end of 2027.
H/t to @tylercowen whose interview with @sama partly inspired this question.
6
⚡ Large-scale AI infrastructure is set to expand rapidly
Experts expect global hyperscale data center capacity to grow from 36 GW in 2024 to ~92 GW by 2030—a roughly 2.6× increase—roughly in line with International Energy Agency projections.
Experts expect global hyperscale data center capacity to grow from 36 GW in 2024 to ~92 GW by 2030—a roughly 2.6× increase—roughly in line with International Energy Agency projections.
7
🎯 Experts and superforecasters often agree—but for different reasons.
Experts and superforecasters make similar median forecasts. However, across all forecasters, different arguments drive the justification for the forecasts.
Some forecasters argue that high valuations and the rapid buildout of data centers are the result of bubble dynamics.
Others argue that improvement in AI agents will lead companies to dramatically increase their revenue without increasing the size of their workforce.
Experts and superforecasters make similar median forecasts. However, across all forecasters, different arguments drive the justification for the forecasts.
Some forecasters argue that high valuations and the rapid buildout of data centers are the result of bubble dynamics.
Others argue that improvement in AI agents will lead companies to dramatically increase their revenue without increasing the size of their workforce.
8
If you have ideas for better questions to track developments in AI R&D, please reach out or submit ideas at: airtable.com/appGCchUyUTPvT…
Read more about Wave 4, including question-by-question details and rationale analysis: leap.forecastingresearch.org/reports/wave4
Find out more about LEAP, our panelists, and analysis from early waves on the LEAP website: leap.forecastingresearch.org
Read more about Wave 4, including question-by-question details and rationale analysis: leap.forecastingresearch.org/reports/wave4
Find out more about LEAP, our panelists, and analysis from early waves on the LEAP website: leap.forecastingresearch.org