Ben Todd: Even a 15% GDP Boost Underestimates AI's Accelerating Impact
Ben Todd, founder of 80,000 Hours, commented on simulations from an AI-driven economic growth model. Even accepting the model's 15% GDP boost by 2030, the simulated impacts keep accelerating, with 2033 far more dramatic than 2030. He argues the output looks mild partly because the simulation was cut off just as things started to spiral. His conclusion: this scenario should not be treated as an upper bound on AI-driven change, and a wider range of scenarios should be considered.
Confirmed
- Todd argues that even if the 15% GDP growth prediction for 2030 holds, AI impacts would continue to accelerate, with 2033 far more intense.
- He notes the model output was truncated before things got out of hand, possibly explaining its mild appearance.
- In a later addition, he suggests that if key turning points fail to materialize before 2030, they could still trigger within a year or two, making gradual acceleration across the 2030s a plausible scenario.
Unconfirmed
- Todd explicitly clarifies he is not predicting events will unfold this way and acknowledges possible errors in the analysis; the 15% GDP figure and 2033 upheaval are model scenarios, not certainties.
Why it matters
- The discussion warns policymakers and researchers that even mild-looking macro forecasts may embed underestimated acceleration; treating a single model scenario as an upper bound could narrow preparation for AI's full range of impacts.
2026-09-29 ~ 2026-09-30 · 5 related posts
Primary sources
- Ben Todd: even 15% GDP growth in 2030 means accelerating impacts well beyond — ben_j_todd · 2026-09-29
- Ben Todd clarifies: not a prediction, but a call for wider AI scenarios — ben_j_todd · 2026-09-29
- Ben Todd clarifies: analysis may be flawed, but scenario isn't an upper bound — ben_j_todd · 2026-09-29
- [source] Ben Todd: gradual AI acceleration through the 2030s is a plausible scenario — ben_j_todd · 2026-09-30
1 near-duplicate retellings: ben_j_todd