Question 1
John David Pressman
x.com/jd_pressmanEssayist and programmer who builds synthetic training data for language models and writes about alignment, AI risk and transhumanism.
How will AI change the world?
Across: their expressed Doom–Bloom outlook. Up: scale of transformation.
Doom–Bloom: 55 out of 100. Scale of transformation: 76 out of 100. Interpretation ranges: 50 to 75 horizontally, 50 to 81 vertically. These are interpretation coordinates, not event probabilities.
12%
“In private conversations I'd sometimes give my p(doom) as 12%”
Undefined “doom”; he says the term conflates several distinct AI outcomes, which the essay separates into layers. Not an extinction-only forecast
Varieties Of Doom · Nov 2025
A central assumption
The central problem is whether desirable values generalize beyond familiar contexts, and that remains unsolved.Answer 1
If this assumption turned out differently, how would their outlook change?
An unresolved question
My central uncertainty is value generalization.Answer 2
What would help them distinguish the plausible outcomes here?
What could change their mind
The biggest update would come from a convincing demonstration of robust value generalization: a system preserving humane judgment across unfamiliar contexts, greater autonomy, adversarial pressure, and major shifts in training data.Answer 5
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.
69 / 100
Interpretation range 67 to 67 on the qualitative scale.
Severe or widespread harm is a material expected part of the future.
67 / 100
Interpretation range 67 to 67 on the qualitative scale.
Human choices can substantially redirect the AI trajectory.
71 / 100
Interpretation range 50 to 75 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 John David Pressman’s
Sources
Articles, interviews, and writings used to ground this simulated user.
Practical guide to synthetic training data.

January 30 posts call value generalization out of distribution unsolved and argue an alignment winter is bad despite dislike of safety-community rhetoric.

Longstanding transhumanist essay, retained as an aspiration rather than a current forecast.

March 17 posts emphasize human-data training is a design choice behind LLM alignment and that RL/synthetic-data convergence must be explicitly considered.

November 1 and 23 posts reject inevitability of doom and describe human-trained LLM agents as comparatively benign, while identifying generalization as the relevant alignment challenge.

Full essay read. He says that in private conversations he would sometimes give his p(doom) as 12%, with the caveat that “doom” is nebulous and conflates several outcomes, and that he declined to give a public p(doom) until he could explain those layers, which the essay then separates. He puts a “paperclipper” successor that keeps nothing of value in the sub-1% range. The essay does not restate 12% as a fresh all-things-considered estimate.

Simulated Assessment
Question 2
Taking benefits and harms together, what overall impact do you expect AI to have?
Question 3
Setting aside good or bad: how much do you think AI will ultimately change the world—a little, a lot, or completely?
Question 4
What’s your rough gut-feel chance that AI causes human extinction or a similarly permanent catastrophe? A number or “no idea” is fine.
Question 5
What discovery or event would most change your view of AI’s future impact?
Where do you land?
Map my worldview