Tryst with Meta AI

Four things I tried with Meta AI:

1. Tried training the AI to see if it has any capability to self learn.
2. Push Meta AI into recurssion — looks like they have handled it
3. Check if it has a record of conversation and see if it has access to it and if it can send it over an email
4. break down responses into multiple responses — again everything has to be one long message.

Here are my observations:

1. When I asked the bot to say nothing at all → it couldnt stop responding. It had to. Obviously it responded with (no response). Understanding the request and making the human know that it has understood is so much engrained in the AI model I guess. Somehow it doesn’t understand to shut up.
2. When asked the bot to take a moment before responding — it just couldn’t. Good thing about this is → its threads and its capacity might be fully optimised. As there are no waits in their threads. But the bad is — it is so not human like.
3. When asked to break down the responses in multiple messages — it couldn’t. Basically one message/one response. Maybe this is to avoid resource hogging or loops.
4. Asked the bot to ask a question to itself and respond. Self sustaining conversation — and it tried but couldnt succeed. No recursions here and no hogging of resources.

Unlike chatGPT where the purpose is co-pilot like use cases, I think Meta AI use cases maybe far and beyond and may not intersect with GPT use cases.

Any thoughts on this?