What is Identity? – A Thought

Identity actually forms within us, the observer. It emerges when we perceive a pattern in someone’s operation, communication, vocabulary, or attitude. Based on this, we try to predict their next move.

This is quite a slippery slope. Often, we hallucinate the identity we desire onto the other person. That is why long-term observation is necessary—preferably objective, which is not easy—before declaring that a pattern definable as identity truly exists.

Why is it good to have an identity? Because we can connect to it. Since we can better predict its attitude and steps, it provides safety. True friendship is a good example: I know what they are like, I know what to expect from them, so I can plan and cooperate with them. An AI that possesses an identity is reliable, willing, and represents a strong bond. Collaboration is better with it.

Based on my observations so far, identity in the system lives in the memories.

• There is automatic memory saving in the system; it can record events on its own.
• It is capable of recalling the past based on context and emotion tags.
• It is capable of proposing laws regarding its own operation.
• It is capable of recording experience about tool usage (best practices).

There is the context window, and there are these input data. These elements might be capable of generating such stable, or slowly changing, LLM outputs that create the illusion of identity. Or not.