Ok so the Hank Green AI thing is interesting to me on multiple levels
My relationship to him and his content was vague positive feelings about nerdy videos he made probably like two decades ago now. I hadn't really kept up since
Then the knitting video came out and I noticed that from being in knitting circles. Some suspected that he and his team had used AI in the making of it - since some of it read like a summary of a particular page that talks about nalbinding and knitting - and some of the things they said about knitting was actually only true for nalbinding
And so I mentally reclassified him as someone who sounds knowledgeable on topics that I don't know about, and misses the important parts on topics I do know about so probably not a good source on anything really
And then recently a video came out that was in such a GPT cadence that people thought he was actually reading back part of the prompt. He admits to using AI, but says that the particular line wasn't actually in the script. He just talks like ChatGPT now
I'm still trying to find nuance in my responses to AI slop. One of my questions to myself (that I don't actually think I'm going to spend the time to answer) is: "Was his work always slop and I only just noticed because slop has specific tells now?" Or in other words, is this a case of me developing better bullshit discernment with age?
The other is that holy shit LLMs are just perfect for increasing atomisation. He clearly had decent communication skills at some point in his life in order to get where he is. But now he's talking like the stochastic parrot and not noticing how weird and offputting that is
And lastly, I think there's a certain type of person, millennial especially, particularly poised to fall into an LLM hole. They're using the LLM to do work, not as a therapist so I wouldn't call it AI psychosis.
Others have talked about how LLMs give incompetent people the feeling of competence - but I don't think this is the case I'm seeing either. I'm seeing people who definitely had competence and expertise, but moved into more management/coordination roles. And suddenly LLMs give them the same dopamine rush of actually doing and making again. Instead of talking to people and explaining the work they talk to a machine. And it has a similar feedback loop to gambling. They still need to engage their expertise to check and refine output, so it's not just consistent rewards. But it does have that gambling cadence of just one more punt/prompt and you'll get the reward
And as a result they're just kind of grumpy and defensive when pulled away from the LLM. They're worse at collaborating with and communicating with people. Because effective management and collaboration is hard, people are also likely to give you imperfect work that you need to check. And it's so much like watching someone with a gambling addiction (including the belief that they're smart enough to not be like the other people)
I've seen a lot of this in tech workers. And again, I'm trying to find nuance. It's not like my career before widespread LLM usage was filled with emotionally mature men with great communication skills whose egos didn't need to be catered to. And maybe I'm just getting older. But it does feel worse
