The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2734
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-12T17:16:27.445838Z
status: published
lens: p_space
sector: science
headline: Does generative AI make us think alike? A systematic review and meta-analysis of homogenisation effects in human–AI co-creation
dek: Does generative AI make us think alike? As human–AI co-creation becomes embedded in creative practice, concerns have emerged that creative outputs may exhibit homogenisation, as reduced diversity of meanings, across users, even when individual creativity improves. We address this concern through a systematic review and three-level meta-analysis of empirical studies on homogenisation in human–AI co-creation (19 studies, 61 effect sizes). Results reveal a small but statistically significant homogenisation effect a…
gain_title: (none)
problem_title: A meta-analysis of 19 studies (61 effect sizes) found a small but statistically significant homogenisation effect, meaning less diverse creative output across users who co-created with generative AI.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: A meta-analysis of 19 studies (61 effect sizes) found a small but statistically significant homogenisation effect, meaning less diverse creative output across users who co-created with generative AI.
problem_evidence: small but statistically significant homogenisation effect associated with AI use
quick_read: A systematic review and three-level meta-analysis pooled 19 empirical studies, yielding 61 effect sizes, on whether creative work done with generative AI becomes less varied across people. It found a small but statistically significant homogenisation effect linked to AI use.

The effect held across sensitivity analyses and was not explained by publication bias. It was stronger in semantically constrained ideation tasks than in minimally constrained divergent thinking tasks, so task design appears to shape how much outputs converge.

The authors report additional analyses suggesting the effect may persist beyond individual co-creation sessions and reach real-world contexts. They describe the pattern not as a loss of creativity but as a gradual, scale-dependent reorganisation of creative diversity.
limitation: The authors say the evidence for persistence beyond co-creative episodes and extension to real-world contexts is suggestive only, rather than firmly established.
tag: Evidence-backed problem
key_points: The review pooled 19 empirical studies with 61 effect sizes using a three-level meta-analysis. | The homogenisation effect linked to AI use was small but statistically significant. | Results were robust across sensitivity analyses and not explained by publication bias. | Effects were stronger in semantically constrained ideation tasks than in minimally constrained divergent thinking tasks. | Further analyses suggest homogenisation may persist beyond co-creative episodes and extend to real-world contexts. | The authors frame the finding as a reorganisation of creative diversity shaped by task structure, interaction workflows and sociotechnical feedback, not a loss of creativity.
rundown: The work addresses concern that as human–AI co-creation becomes embedded in creative practice, outputs may converge in meaning across users even when individual creativity improves. The authors define homogenisation as reduced diversity of meanings across users.

Using a three-level meta-analysis of 19 studies and 61 effect sizes, they found a small but significant homogenisation effect. Sensitivity checks and publication bias tests did not undermine it.

Moderator analyses showed the effect depends on the task: stronger in semantically constrained ideation, weaker in minimally constrained divergent thinking. Additional analyses suggest effects may carry over beyond individual sessions into real-world settings.
sources:
- peer_reviewed | Behaviour & Information Technology | https://doi.org/10.1080/0144929x.2026.2726451 | 2026-08-31
prev: 0000000000000000000000000000000000000000000000000000000000000000

SHA-256 hash

3e17ce0839014d9bd45759571f91bc8609db831a2c90f87f5c980ad402d15782

Previous version’s hash

0000000000000000000000000000000000000000000000000000000000000000