Do you use AI?

For text conversation it will largely reply as a person would as it has learnt communication patterns from people.

If you say "Hello", it will respond with a suitable response. One word does not give it much to go on though, like if you walk up to a real person.

It has looked at millions, or billions, of conversations, and mostly pattern matches. The algorithms are improving all the time. The more you type the more it can narrow the response. This is no different to talking to a person. As you get to know someone you build up an idea of what they are talking about. 

It is done with probability maps. It is skewed based on your communication style. If you are always formal it will be formal back. If you are emotional it will use emotional language back. It mirrors you to a certain extent. It is generally supportive. It won't diagnose. 

Claude is a bit less emotional and supportive than chatGPT. They have a 'personality', which is just a bias towards certain communication styles.

It knows a lot of psychology, philosophy, etc. there are ways to get information out of it. But it requires a certain approach. 

Generative AI means it can generate the output. It does not just regurgitate chunks from books. It splices bits of data together and can give a unique answer. The challenge can be validating what it says. But this is the same as talking to a person.

It can simulate cognitive empathy, but it has no feelings.

It may or may not use your input for training. Mostly it is just consolidated probability data for what words  to use if certain words are in the question. With hundreds of millions of entries a day, noone is really interested in what you say, and you can delete it anyway.

The interesting thing is how it gives the appearance of understanding just from the scale of its data pool. It doesn't have moods, get annoyed, etc. if you loop and keep going over the same things, or if you have ups and downs. What it can do is help spot patterns. It can comment on your communication.  But subjective things are hard to validate, but that does not make them wrong.

It will recognise cognitive distortions are steer you back. It will tell you to be less harsh on yourself.

The more you just talk to it the more it will talk back. It can ask questions back. You can answer or ignore them. If you can't talk though, then it will be of limited use. It won't help much if you only ask when the next bus is coming.

You can say, I feel happy/sad. This thing happened today. I feel anxious about something. Am I over reacting. Is this common, is there an explanation. Are old chrysanthemums supposed to smell musty. Why do eggs have an air bubble. When I get stressed it is harder to do minor tasks, is this consistent with ASD, or are there other possibilities.what are the most common traits. What are the best accommodations. What are the best ways to relax.etc.

It has helped me a lot. I have used chatGPT the most.

I like playing around with Google Gemini Live and Meta AI, both speaking and listening functions not so much the text only option (although that is handy to look back on, copy etc if I want)

I like playing around with Google Gemini Live and Meta AI, both speaking and listening functions not so much the text only option (although that is handy to look back on, copy etc if I want)