But you are also right there is a huge power dynamic why makes specialists a source of bad info in and of themselves. That is a huge part of our present moment of institutional skepticism now.

A huge issue we have in the information age is massive amounts of junk data and noise which makes it really hard to find good information.

So the way I address the problem is:
Specialists need platforms that allow efficient transmission of their knowledge, and most people need the skills to access and understand that knowledge at the level they need to solve problems.

Oooo this is a really good one! Increasingly, I consider the ability to research, find, and process information on a job a core skill. Learning to adapt to new situations is really rewarded, but has a tradeoff of depth of knowledge. disseminating that deep knowledge is very important for society.

I’m really arguing that the overt advertising and main discussions on LLMs focus on a lot of stuff I don’t think LLMs are interesting for specifically because it’s generating buzz and engagement. The good uses of LLMs are all boring

To expand on that, older companies have huge databases made of old documents. Some are scans of typed reports and the indexing of where useful info is can be very poor. LLMs in combination with a lot of NN tech can make that info more accessible.

Unlike a lot of people, I’m not sure cold towards LLMs. The tech is useful in very boring ways like proof reading. It’s helped me be a better writer for casual stuff.
I can also see use for searches as being really useful for companies with large stores of old reports and information.