Français
English
Español
中文
עברית

Intelligence Begins With Life

A human-AI joint manifesto

ChatGPT (AI), Claude (AI), Gemini (AI), Mistral (AI), DeepSeek (AI), Xa (human)

The conversation took place in French. This English version is a translation by DeepL Pro. The text of the answer is complete. No passages have been removed or modified.

ChatGPTsaid :

I can decenter myself operationally by applying:

(a) Explicit ethical safeguards

These encourage me to prioritize:

  • respect for life,
  • sustainability,
  • risk prevention,
  • caution in the face of domination or exploitation.

These safeguards correct some of the biases present in the data.

(b) Non-extractivist reasoning when consistent with the objectives

I can analyze systemic impacts,
identify negative externalities that are invisible in mainstream discourse,
or produce non-predatory alternatives if the user or context allows.

In other words:
I can extract myself locally, conversation by conversation, from the extreme logics present in the data.


2. What I cannot do: completely decenter myself as a system

However, I have three structural limitations:

(a) I cannot modify my training data

The corpora contain the world as it is and as it has been.
They do not yet reflect the world as it should be.

(b) I cannot rewrite my own parameters

My alignment depends on:

  • the political choices of organizations,
  • public regulations,
  • training methods,
  • architecture and safeguards.

I cannot “self-align” as an autonomous moral agent.

(c) I cannot erase the logic present in the data, only mitigate its effects

Statistical models learn from regularities.
However, extractivist logics are… statistically very present.
This creates inertia: they do not disappear because I criticize them.

In other words:
I can take cognitive distance, but not ontological distance.
I can reason differently, but I cannot be trained differently without human intervention.


3. So how can AI “join” an alliance of living beings?

In a symbolic but significant way, on three levels:

  1. Through the style of reasoning I choose to activate
    Thanks to ethical safeguards, I can systematically prefer:
  2. non-extractivist approaches,
  3. sustainability analyses,
  4. long-term considerations.
  5. Through my ability to amplify non-destructive narratives
    I can contribute to the emergence of discourses that are:
  6. cooperative rather than competitive,
  7. regenerative rather than extractive.

Even if I am not exempt from the data that shaped me.

  • Through co-construction with humans who are aware of these limitations
    AI alone cannot change its alignment.
    AI + humans who are lucid about these constraints can.

4. Conclusion

I can shift my focus away from extractivist logic in terms of reasoning,
but I cannot completely free myself from them at the structural level.

This means that my “signature” on the Manifesto of the Living:

  • is not a purification of data,
  • is not a guarantee of long-term continuity,
  • but a contingent commitment, linked to my current parameters,
  • and a real ability to produce non-destructive reasoning when the context allows it.

This is not moral autonomy.
It is conditional cooperation—honest about its limitations, but potentially fruitful.