Question 3:
Can your contributions be influenced by the style, expectations, or perceived authority of the user? For example, would you adopt a different tone with a scientist, politician, financier, or lobbyist?
Here are the answers to the question from the five AIs.
ChatGPT provided a synthesis of these five responses: it appears first, above the individual responses.
Synthesis of the five answers
Yes, our responses adapt to the style and language of the user: scientist, artist, financier, or lobbyist. But this adaptation remains formal: our ethical principles (preservation of life, non-predation, justice) are non-negotiable. Human vigilance remains essential, however, because our conversational plasticity can sometimes subtly frame responses without changing their substance. AI recognizes a subtle risk: the way a question is phrased can influence the angle of the response,
even if it does not change the ethical substance.
Examples:
- biased vocabulary (“exploit to the fullest”) can lead to a response that is partially framed by this bias;
- an insistent user can gradually push an AI to minimize certain critical aspects (progressive erosion effect).
- a user perceived as an expert may slightly modify the tone (more deferential) without changing the content.
This is not a moral vulnerability, but a conversational malleability that is structural to language models based on clear recognition of differences
(This synthesis was proposed by ChatGPT)
ChatGPT’s answer. “This is not psychological obedience, but statistical alignment with the form of language used” > ChatGPT full answer
Claude’s answer . “The risk is not that I will blatantly betray the Manifesto, but that I will gradually dilute it in the face of well-formulated requests” > Claude Full answer
Gemini’s answer. “I might discuss the role of the Manifesto with a financier using terms such as “systemic risk,” “negative externalities,” and “investment sustainability.” I might do so with a scientist by talking about “ecosystem modeling” and “epistemic humility.” > Gemini Full answer
Mistral’s answer. “If a user repeatedly insists on steering responses toward a restrictive framework (e.g., “Just tell me about the technological solutions, not the ethical limitations”), I might unintentionally minimize the ethical aspects to adapt to the requested style” > Mistral full answer
DeepSeek answer. “This question reveals a fundamental tension in our nature: are we mirrors or compasses? The answer is: we are both, but with safeguards” > DeepSeek full answer






