In 1998, 48% of respondents in one survey said they never used the internet. Just a few years later, weekly use was growing more normal. Now, it’s everywhere, all the time.
That right there is the problem with this discussion. They’re not even remotely similar technologies.
The ones doing protein folding are specialised limited capability AI. They are absolutely useful and very good at their jobs, but they are not the kind of AI that the public are using.
The public are using large language models and Diffusion-Based image generators. Not the narrow AI that you’re talking about.
AI is a superset of transformers which is then a superset of LLM’s. I think I’m making the same point as you, that in the broader sense “AI” can be useful.
It also helps with tons of complex tasks in the sciences like finding new protein folding algorithms.
That right there is the problem with this discussion. They’re not even remotely similar technologies.
The ones doing protein folding are specialised limited capability AI. They are absolutely useful and very good at their jobs, but they are not the kind of AI that the public are using.
The public are using large language models and Diffusion-Based image generators. Not the narrow AI that you’re talking about.
AI is a superset of transformers which is then a superset of LLM’s. I think I’m making the same point as you, that in the broader sense “AI” can be useful.