Raw LLM Responses

until we develop a new ai algorithm other than things like chatgpt we aaint getting agi any time soon because chatgpt is very inefficent it cant even learn from mistakes and forgets things fast like really fast after what like 40 50 prompts it forgets nearly all prompts writen before 40 th 50th in a span of a month we might as well say ai has alzheimer some can argue we will improve these models in 10 20 years but with this algorithm hell nah we arent getitng agi any time with this its not even efficent companies use whole nuclear reactors to power them so what happens when a model has trianed wrongly or let say they replaced programmers well what about hackers we can have many attack agaisnt ai development fake data mis information how can you trust agi what if it made a critical vurnability it cant fix that because if it could fix it it wouldnt even do the vurnability at the first place we have to train it for nearly infinite edge cases which will make the ai incredibly complex also today we have params to run a single large language mother let assume 2Trillion params this is actualy very intelligent ai but the hardware required for it ? 1-10m atleast and we would need arround 4 tb vram requiring 64-256 gpus which draws arround 100-500 kw of power human mind uses just 20 wats btw so just for ai to exist we need 10-100 dolar electricity cost per hour and what about cooling that hardware and what about maintenance we prob are not gonna get any ai that is agi level any time soon and this whole openai thing is just a baloon waiting to pop