[mateo1]

Yep, and it's disconcerting how much trust people put into this system. I've got a masters in engineering and some random guy would try to argue with me about something as simple as calculating energy consumption from wattage because "chatgpt said something similar, but your answer is a little off".

Articles, social media posts and photos are now often AI generated. There is going to be a certain percentage of today's students that will graduate "knowing" blatantly wrong things an LLM told them. These things will eventually make their way to books and society.

One thing the scientific community never got around to accepting themselves and admitting to the public is how much of the learning is trust-based rather than fact-based. Nobody can prove everything from the ground up, you need to put your trust into your predecessors. Eventually you'll have the capacity to doubt and check a percentage of what they said, but not everything. When this circle breaks and you don't know if your lecturer is for real or just parroting LLMs we'll be in trouble. The "financialized" ultra-competitive environment of higher education doesn't help with that either.

[JohnFen]

> One thing the scientific community never got around to accepting themselves and admitting to the public is how much of the learning is trust-based rather than fact-based.

In the years I spent working with research scientists, I observed the exact opposite of this. They were acutely aware of this and it was a common topic of discussion. It's important to them because they need to be as aware of and to call out as many assumptions as possible.

[SuperNinKenDo]

Then perhaps only the latter part is true.

Having spent time in academia myself though, my reading is that while everyone professes an epistemology that accepts its own limits, in practice everybody behaves as though they'd grasped absolute truth through divine revelation.

[gandalfff]

For what it’s worth, I’ve found using Perplexity Pro (with Claude 3 Opus) to exceed the accuracy of asking a non-expert armed with Google Search and a few minutes of time.