When you talk with AI (transformer) such as ChatGPT or Gemini, they are pretty much like a human. They are supportive and they answer all your questions. Sometimes I feel as if I’m having a real conversation with a real person.
But at the same time, I know that what I’m feeling is an illusion. What AI is actually doing is generating most likely tokens based on the chat session and what they learned during training.
They don’t “understand” anything. They just identify patterns in the chat session and provide the most likely response for the specific context. This means that conversations with AI will be drastically different between one with a highly intelligent and highly educated person and one with …well, a small kid. If you feel that AI is fool, there might be an issue with your questions.
I sometimes find interesting to notice that AI behaviors are different from ours.
- Not aware of time lapse
- They don’t have a self
- They don’t have a feeling
- They don’t get tired
- They don’t change
- They don’t remember anything if not with a memory tool
- They lack consistency
- They only respond; they don’t talk to us
Half of these come from the fact that AI is static. If AI changes through our conversations, I imagine, they will be more interesting and more human-like. But again, conversations with random users will be too noisy and will disrupt AI’s behaviors in a bad way.
To solve this, AI needs to evaluate the importance of conversations to train itself in a good way. Will we see more human-like AIs in the future?