Raw LLM Responses

"we choose what to remember and what not to (without understanding how we do it), the AI can't do that because we don't," First off, that is NOT the defining characteristic of self awareness: "The defining characteristic of self-awareness is the ability to become the object of your own attention. Unlike basic consciousness, which is simply being aware of your body and surroundings, self-awareness is the "awareness of that awareness"—recognizing yourself as a separate and unique entity from others and the environment." Secondly, for AI it's NOT true: they CAN and DO choose "what to remember": AIs are now being trained to consolidate chats to store only the relevant parts for future reference. THEY judge what might be valuable to be able to reference in the future in chatting with that person. We're going places that the average person on the street has no clue about. But a statement such as: "and if we don't we cannot program a computer to do it." is BLINDINGLY IGNORANT. Starting with we don't "program" AIs as much as TRAIN NEURAL NETWORKS to have the abilities they have today. AIs DO WHAT THEY DO not because of "programming", but because of EMERGENT PROPERTIES. There's not a line of code in GPT5.2 that tells it how to do ANY of this: "GDPval is a benchmark that measures skills across 44 real world jobs, not with some multiple choice test, but challenging them to deliver REAL WORK PRODUCTS the professionals that averaged 14 YEARS ON THE JOB were doing: Blueprints, CAD drawing, engineering specs, lengthy analytical reports, financial analysis, drafting legal documents, creating nursing care plans, developing software, managing engineering projects, analyzing financial data, and producing business presentations, etc. The documents created were judged BLINDLY by experts for whether the AI or human ones were BETTER, more USEFUL, more DETAILED and more USABLE IN THE REAL WORLD." GPT5.2 scored a 71% FAR BESTING THE HUMANS. And since then Claude AI has IMPROVED ON THAT SCORE. The FINAL ANALYSIS was that AI did the jobs 100x FASTER and 100x CHEAPER than the humans. You have a LOT of catching up to do, starting with a basic realization that YES, we had an AI learn to play both Chess and Go FAR better than any human ever will be able to. Anything a human "can't do" is COMPLETELY IRRELEVANT to what an AI can do.