Question 1
Alex Zhang
x.com/a1zhangAI researcher who builds benchmarks and studies how scaffolds and recursive model calls can get more out of existing language models.
How will AI change the world?
Across: their expressed Doom–Bloom outlook. Up: scale of transformation.
Doom–Bloom: 57 out of 100. Scale of transformation: 50 out of 100. Interpretation ranges: 49 to 62 horizontally, 0 to 100 vertically. These are interpretation coordinates, not event probabilities.
Not estimated yet
Not enough about catastrophic risk in their simulated answers to estimate it.
A central assumption
Better models and better harnesses are complementary: training expands what components can do, while scaffolding determines whether those abilities combine into reliable performance.Answer 1
If this assumption turned out differently, how would their outlook change?
An unresolved question
I don’t have a well-supported overall societal forecast.Answer 2
What would help them distinguish the plausible outcomes here?
What could change their mind
If learned decomposition and recursive inference let models solve genuinely novel, out-of-distribution tasks—not merely longer versions of familiar benchmarks—that would strengthen my expectation of rapid capability gains from better harnesses.Answer 4
What evidence would be enough, and in which direction would it move their view?
More details
Substantial benefits are expected, with important conditions or distribution limits.
62 / 100
Interpretation range 33 to 67 on the qualitative scale.
A tentative estimate from your answers; the wider range shows other plausible readings.
51 / 100
Interpretation range 0 to 100 on the qualitative scale.
These interpretations keep their stated conditions. Benefits and harms can both be substantial. The ranges describe how we read their simulated answers, not statistical confidence intervals.
Similar worldviews
Thought leaders whose simulated worldviews are closest to Alex Zhang’s
Sources
Articles, interviews, and writings used to ground this simulated user.
Coauthored proposal on underused capabilities and learned decomposition.

First-party index documents RLMs, KernelBench, VideoGameBench and 2026 work on harnesses.

Coauthored explanation of decomposing long contexts through a programming environment and recursive model calls, enabling processing beyond a single context window.

Argues existing neural models remain underused and inference scaffolds can unlock abilities; does not claim that scaling has ended.

Explains how harnesses can turn difficult out-of-distribution problems into compositions of familiar subproblems; better training remains important too.

Simulated Assessment
Question 2
Taking benefits and harms together, what overall impact do you expect AI to have on our society?
Question 3
What’s your gut-feel chance that AI causes human extinction or a similarly permanent catastrophe?
Question 4
What discovery or event would most change your view of AI’s future impact?
Where do you land?
Map my worldview