I've found that LLMs are good at writing code, but bad at solving problems.
This seems to align with the general experience using current gen LLMs. They're great at doing things that have been done before, and generating plausible ideas when brainstorming, but there isn't yet any kind of "logic"

Humans are already overly reliant on technology.

If you swap the word "tech" for "tool" this becomes obvious. Our society is deeply reliant on things like cooking, farming, clothing, writing and more.

How many jobs REQUIRE you to own a car?

We are all already cyborgs.

Language Models !== AI

The field of AI research is ~80 years old and includes many different methods for designing intelligent computers. Language Models and transformers are just the latest innovation.

Even if scaling laws for LLMs start to slow down, there are lots of other promising routes.

I dislike the word "Artificial" in AI. It brings up too many feelings of otherness and replacement.

I prefer "Augmented". It brings the focus back to the human and leads to questions like "How does x enhance human capabilities?" rather than replace them.

Augmented Intelligence is the better AI.

Language Models aren't, and may never be the superintelligence that some people are hoping for.

Instead, what we have is a novel method of human computer interaction.

Computers can FINALLY "understand" natural language instructions, and we've barely explored the UX implications of this.