Raw LLM Responses

I like the video, but I disagree with your understanding of AI. LLMs like ChatGPT are nothing special, they parse texts to look for correlations between words and predict the next one based on their training data. They are great at sourcing ideas and remixing existing stuff, but cannot extrapolate and create new ideas. And the more complex the concepts being remixed are, the more human guidance is needed to steer the AI towards the right path. So AI, in its current iteration, can fully replace only the most repetitive, simple white collar occupations (e.g. customer service), jobs that require a higher degree of originality can only be assisted by AI. Of course, I know this because I read papers on LLMs and understand what's going on inside the models, but I do realize that many, CEOs and higher management in particular, don't, and this obviously affects how jobs are offered. Unfortunately LLMs communicate like humans, so they look intelligent, and the big players like OpenAI have all the interests in selling this idea. This doesn't help with the job market, and we will suffer because of it, but that's human stupidity at fault, not AI necessarily. One can hope that companies recognize that AI is not delivering on promises, but still, COVID showed them that remote jobs work, so if they want to reduce costs they have a billion indians to pick from...