>>106122394 (OP)
The definition of AGI is very vague and before 2022 the only AI systems out there were like detecting a single object really well or generating only cats from a drawing of simple shapes. GPT3 if you don’t account for hallucinations was already comparable to bellow average human as far as text alone goes, and Dall-E the same way in terms of image generation. We just got used to it.
If I went to you right now and showed you a robot that watches one single video of a thing being made or just looks at the blueprint and then on its own constructs that thing flawlessly and faster then a human you would shit yourself. But in 2 or 4 years when we actually get that thing you will say it’s a nothingburger because you saw 3 years worth of robots slowly putting things together with guck load of pre-training so you got used to it. I don’t blame Musk back when he was still in the company to be legitimately scared when he saw the first model capable of writing a whole page of coherent text back in the days where the most advanced AI was shitty self driving cars that showed no progress and simple image detection.
However after GPT-3 and ChatGPT they should have stopped being faggots about LLMs. Save up this faggotry for when we get real AGI.
