@adamgreenfield I might be misremembering, but I feel like you once expressed the sentiment that people on the fediverse were Being Too Much about this technology. That the nerds here ought to let people figure out that this stuff was bad on their own and stop being so preachy about it. I apologize if I got that wrong.

Regardless, it should be clear that people are NOT going to decide that this technology is bad — or (for that matter) that they have any responsibility to address climate change or defeat fascism — without a major reckoning about the widening chasm between the way they live their lives and their stated values. Even liberal ideas about “human rights.”

Using any hyperscaled LLM is climate denialism and lines the pockets of fascism’s biggest backers. I don’t know to get people to see this who don’t want to see it. I don’t know what tools to reach for other than virtue. I don’t see how we get out of this without something that looks like a religious movement.

@adamgreenfield Well, I looked up some basic information on pipe organs and found that there are only two in the world with 64 foot pipes, both built since the end of the 19th century, i.e. not at all Gothic. That seems...ya know...relevant.

The fundamental problem with LLM's isn't their writing style, it's that they cannot distinguish truth from falsity, so they lie convincingly and most humans (including the "smart" ones) will simply accept it. And the better they get at mimicking human writing, the more convincing their lies become. I'm sure this will lead to wonderful outcomes for all involved. 🤦‍♂️

@adamgreenfield I'm finding many interesting bits coming out of the hallucination machine. but all the interesting bits live in the human side: what we read into these sentences, what emotions and connections arise in ourselves, what incredible dialogue we're prepared to have with ourselves. i think in all its harms and all its unfulfilled promises, it's a profoundly human technology.

@adamgreenfield incompetence has always been the flimsiest of the reasons not to use LLMs longterm imo. technologies get better, and the best funded companies in the world have been working to make it so. it’s not surprising to me that they’d be able to assemble a compelling piece of language, it’s something they’re pretty well suited for as a technology