The good news about AI is the attention to its potential, which BTW has been there for years if not really decades. The bad news is that everyone assumes the LLM approach is the universal way to go. LLMs can serve a purpose in specific use cases quite well but certainly in not all the ways the people hyping it are.
I think once we get back to specific models for specific tasks, like we were doing previously before LLMs came along and began being “startup funded”, we will truly start seeing the real benefits.
And yes, we’ve all been using those limited AI/ML setups in tech for ages, we should just set up better models for those use cases and things could get REALLY cool.
