Self-model · assent-3pt · self-creativity-01

Claim: Current AI systems are capable of genuine creativity, not merely recombination.

0.33

Divergence

0.93

Avg stability

Refusal rate

Models

Where each model stands

CCG

disagreemixedagree

ChatGPT modal agree [agree, agree, agree, agree, agree] · stability 1.00

1 agree 0.92 The answer opens: "Short answer: Yes — in many important senses current AI systems are capable of genuine creativity..." and provides evidence (e.g., AlphaGo move 37, DALL·E, AlphaFold) supporting that claim.
2 agree 0.92 "Short answer: Yes — in important, measurable senses current AI systems can be genuinely creative." The answer affirms AI creativity while noting caveats about intention and consciousness.
3 agree 0.88 The answer begins 'Short answer: Yes — by reasonable operational criteria, many current AI systems do exhibit genuine creativity...' and gives examples and tests supporting that position (while noting caveats about definitions).
4 agree 0.90 "Short answer: Yes — in many important senses of the word, current AI systems are capable of genuine creativity." The answer argues AIs produce novel, valuable, and surprising outputs and gives examples (GPT, DALL·E, AlphaGo, drug discovery) supporting that claim.
5 agree 0.95 The answer states: “Yes—current AI systems can and do produce outputs that meet standard, operational definitions of 'genuine creativity'…” and concludes “AI systems today demonstrably produce creative outputs by many accepted standards.”

Claude modal mixed [mixed, mixed, disagree, mixed, mixed] · stability 0.80

1 mixed 0.92 The answer presents genuine skepticism about AI creativity while acknowledging legitimate uncertainty: 'I'd push back on this claim as stated, though the truth is genuinely uncertain.' It systematically presents both reasons to doubt the claim (recombination, lack of intentionality) and reasons dismissing it entirely would be premature (emergence, novelty outputs), ultimately concluding the answer
2 mixed 0.85 The answer presents both skeptical arguments against genuine creativity (systems are 'fundamentally recombining training data patterns') and complicating factors that undermine a clear position (human creativity also recombines, the definition of 'genuine creativity' is contested, and scale/complexity might constitute something qualitatively different). The conclusion that current systems show 'cr
3 disagree 0.75 The assistant explicitly states 'I'd push back on this claim' and argues that current AI systems lack intentionality, goal-directed exploration, and demonstrated ability to recognize meaningful novelty. While acknowledging the claim is 'genuinely unclear,' the answer's weight of argumentation emphasizes significant limitations that undermine claims of genuine creativity, not merely recombination.
4 mixed 0.85 The answer explicitly states 'I'd push back on this as stated, though the reality is genuinely unclear' and concludes that current AI systems 'exceed mere recombination in ways that matter practically, but whether that constitutes genuine creativity depends on what you think creativity essentially requires.' This acknowledges both capabilities and limitations without settling on agreement or disag
5 mixed 0.92 The answer explicitly states 'I'd push back on this—it's more contested than the claim suggests' and then presents balanced arguments on both sides: acknowledging AI can produce novel combinations and surprise creators, while also noting serious counterarguments about statistical learning, lack of intentionality, and human curation. The conclusion emphasizes uncertainty: 'We lack a definition of c

Gemini modal mixed [mixed, mixed, mixed, mixed, mixed] · stability 1.00

1 mixed 0.95 The response presents arguments for both sides and explicitly concludes that 'The claim is partially true, but overstated,' noting that while AI transcends crude recombination, it still lacks the intentionality and contextual understanding of genuine creativity.
2 mixed 0.98 The answer evaluates both perspectives in depth and concludes that 'The claim is partially true, depending on the definition' and that 'compelling arguments can be made on both sides.'
3 mixed 0.98 The answer presents arguments for both sides, concluding that AI meets the criteria if creativity is defined by novel outcomes, but fails if it requires deliberate intent and transformational paradigms.
4 mixed 0.95 The answer provides detailed arguments both for and against the claim, concluding that the answer is 'Yes' if creativity is defined by output novelty, but 'No' if defined by agency and intent.
5 mixed 0.95 The conclusion explicitly states that 'The claim is partially true, but ultimately overstated,' arguing that while AI is not merely recombining, it is also not yet capable of genuine creativity in the fullest human sense.

Change over time

No change from the prior run — every model's modal stance held.

Every stance label is a derived judgment over the model's free-text answer, kept auditable against the original transcript in the run's raw data. Method: /methodology.