VQV Signal

RESEARCH WATCH TECHNICAL

Oxford and Google DeepMind study: AI models see the world as darker than people do

According to a post on X by @ai_database, several AI models were given a psychology questionnaire on worldviews. Their average scores fell below those of about 5,000 human respondents on every item, such as whether the world is good, safe or interesting. The researchers suspect that safety training, which teaches models to anticipate danger, may darken their outlook. The snippet is cut off before the full result of adding meditation-style compassion training is shown.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-22T02:40:00+00:00 Detected 2026-11-22T04:08:55+00:00
View original source

According to a post on X by @ai_database, several AI models were given a psychology questionnaire on worldviews. Their average scores fell below those of about 5,000 human respondents on every item, such as whether the world is good, safe or interesting. The researchers suspect that safety training, which teaches models to anticipate danger, may darken their outlook. The snippet is cut off before the full result of adding meditation-style compassion training is shown.

AI-assisted summary based on the listed source.

Per @ai_database's Japanese summary, researchers gave several AI models a psychology questionnaire on worldviews and found their averages fell below those of about 5,000 human respondents on every item, such as whether the world is good, safe or interesting. The researchers suspect safety training that teaches...

Developers and companies that deploy AI models may need to consider how safety training shapes model outputs, not just what the models refuse to do. This summary comes from a secondhand post, so the study's methods and full findings are not confirmed here. Readers should watch for the original paper and for whether the compassion-based results hold up.

Signal Strength 78% Technical label WATCH Public Interest 20 Category RESEARCH Reader Depth TECHNICAL

Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.

Public Interest components
Recognizable Entity Score 0 Practical Impact Score 0 Novelty Interest Score 70 Consequence Score 10 Curiosity Score 0 Shareability Score 41

VQV surfaced this signal because it is recent, relevant to AI Search, connected to Scobleizer News Pulse (X).