> research and development. I think that is the question, how much legitimate R&D is really going on here vs trying to shove an LLM into some random hardware a

> research and development. I think that is the question, how much legitimate R&D is really going on here vs trying to shove an LLM into some random hardware and ship it? Or shove an LLM in some app trying to solve a problem that it isn't capable of. No doubt that creating these models are hard, having the data is hard. But how much of the AI startups is actually that vs just shoving the OpenAI API in something.