Raw LLM Responses

Love the video. Truly do and love your channel. Yes you caught it some examples of it contradicting previous statements which so great. But if you listen to what its arguments are, and go beyond just attempting to “Call it out” and think of the entire picture; The AI models responses were superior in terms of tying its arguments to grounded, logical reasoning. If you were instead to listen to the exact reasoning it was providing, moving past simply just “Debate Theory” and instead used dedeuctive reasoning to understand and actually think through the arguments it was making opposed to solely attempting to find contradictions in its reasoning, you would find that all argument made by this conversation were actually completely accurate and valid. You seem to be relying on information that was not previously mentioned or discussed in the previous chat responses, which is similar to say a lawyer attempting to discredit a witness/other attorney/victim etc by bringing up what you believe to be inconsistency in their legal arguments but basing that solely on information/evidence/items in disclosure that were never disclosed to the lawyer in the first place. Asking An AI agent questions assuming it can accurately interpret your future responses and argumentd and account for those in its original answer is unrealistic. If you take everything it said item by item, and compare that to what you had asked you will see that it is pretty consistent and accurate. This video is a great example of how AI chat can be carefully crafted to produce responses you awere looking for, but if you instead o trying to contradict it at every step and isntead ages all instructions in one or 2 chats quirked, you should see a large difference. Again great video showing how AI models can be manipulated, but this did not show you breaking the model with the points you brought up (And yes each item you listed is a real thing, but you are applying them far to arbitrarily. This is just the opinion of a software engineer who design custom AI models for specific Industry needs. Not harping on you, great video honestly just expanding on what you may not know or maybe didn’t explain well enough in the video.