Review
Recently I listened to an interesting TED talk suggested by our CEO:
https://www.ted.com/talks/yejin_choi_why_ai_is_incredibly_smart_and_shockingly_stupid
That was quite interesting insight in strength and weaknesses of AI LLM like ChatGPT.
- Large Language Models(LLM) training date is huge but reasoning is still quite weak.
- LLM sustainability is an open question
- LLM think what next word to say, not what to say.
- And that’s why they hallucinate.
- Even before watching this talk I got impression of LLM as super/very hardworking and at the same time very low intelligence system.
- AI Scientist in a talk tells that we should work on more intelligent models, idea is replace quantity with quality have a lot of downsides.
- So LLM is not something to fear, it’s something to stay and be useful. But we can do much better AI systems in future :)