☆ Yσɠƚԋσʂ ☆My kids act like I'm going into psychosis just because I started toying with opencode and I give the free models genders and nicknames. Like it's not big-pickle, it's 🥒 Dilly 🥒. But I know it's still just a LLM clanker boy that I probably need to set on fire. It's just more natural to use with a stupid nick name.
It's probably just a hyper fixation at the moment. I blame my kid that got me started on Linux though. He got tired of helping me and left a note on console to try typing opencode and asking for help. Now I have all these stupid OCD private repos that basically just customize opencode. And I still have no idea what I'm doing lol.
Fuck ai lol
We use anthropomorphic for machines all the time, a cars engine doesn't "run", a computer doesn't "sleep", a surveillance camera doesn't "see". We just use these words because it's a close approximation of what it is doing and it's simpler to understand for normal people, why are LLMs different?
To expand on what Elixir_Fan said: See the ELIZA Effect. This is not a new problem and it does indeed arise when people stop seeing the software as a tool.
I think the difference is that we have a very intuitive understanding that computers or cars or cameras aren't alive, while we can chat with LLMs and many people are genuinely convinced they're "alive" in some sense. It's important not to anthropomohphize LLMs so we can have a healthier relationship with them as tools, and not think of them as analogous to friendships or romantic relationships, etc.
I guess but it gets to the point of being annoying, saying shit like "the statistical text predictor machine predicted text in a way that I deemed to be wrong" doesn't quite roll of the tongue as "the clanker fucked up". I don't think people that are prone to AI psychosis are going to be deterred by a media outlet using slightly different wording to explain LLMs.
Don't get me wrong I'm glad they didn't drink the techbro kool aid but I don't think this is going to have a huge effect on anything.
Wait, were they trying to say they had feelings now too?
I didn’t think that was on the AI bro agenda.
I assume this is more like non-tech users who have AI girlfriend?
No, none of the people quoted in the article talk about feeling, their contentions were over whether the words think or understand are the right words for what the AI is doing.
No one sane is arguing LLMs have feelings or emotions, but that strawman is better for the headline I guess.
Wait, were they trying to say they had feelings now too?
the brand of libertarianism that the tech bros espouse classifies machine learning and ai as extensions of humanity and deserving the same rights/recognition as humanity itself as a result -- i wouldn't be surprised if this kind of outlook also breeds the idea that mimicked feelings should be considered actual feelings too along the same vein as what the capitalists convinced the supreme court to call "corporate personhood"
