When will they make AI which is a good debater?

I find it too easy to beat AI in almost any debate. It always ends up agreeing with me or just making blatant contradictions which I then use to disprove it.

AI seems really bad at defending its arguments, and this is mainly because it picks arguments and facts from internet search, and most of those are bad and easy to disprove in relation to topic.

Also, AI seems to be incapable of not making generalizations. I dont know why this is, but it never gives precise argument with proper premises, unless you ask it to every time.

It will just write wall of copy pasted text, which is usually easily disproved with few sentences.

Now, maybe these free models are inferior, but I hope they actually make free model which is good at debating. There are hundreds of AI, but most are similar because they copy paste same content.

Now, I tried debating Grok 3, easy to defeat. ChatGPT is even easier, sometimes it doesnt even understand the argument until I explain it further. ChatGPT is much more censored too.

Deepseek is generally trash AI, it censors so many topics and it spies on you, not worth even having.

Perplexity is more like search engine than AI.

Liberty AI is like Grok 3, except that its not completely free. You eventually run out of starting credits.

Gemini is good for research, but its debating abilities are trash.

The only way AI can be made into a good debater is if you specifically instruct it on how to make arguments. It literally doesnt know what argument is otherwise.

I experience the same issues so I just started debating against a mirror. You feel absolutely insane but it's an effective way to both practice arguments and to become a generally better public speaker. 

My main problem with AI is that it doesnt know what premises and conclusion is. It will just list 2 random claims as two premises, unless I specifically tell it to make deductive argument with premises and conclusion.

Also, it seems incapable of understanding what truism argument is.

Even with deductive argument, it makes premises which are unproved or just false.

I have to take AI step by step so that it makes "A or not A" truism argument, otherwise it just doesnt know. So in this area, AI is still very bad. Maybe its because most internet is filled with trash arguments and AI just picks those.

debate it on things it is forced to agree with and it won't concede. tell it gypsies are genetically inferior and then don't let up on the debate

I debated it on child's consent and it conceded. Depending on how you say it, it will concede. Maybe dont say inferior, but say similar thing which isnt blatant racism, like gypsies have less success in life, and then debate that cause could be genetic.

   He's Gooooooooone!