Post

I can't get past the idea of people making emotional connections with AI as if it were another human. It's one thing to be fooled by a chatbot masquerading as a person. But once someone knows it's AI, treating it as a friend can only lead to emotional ruin in the long run.

How is it different from John Lennon's murderer creating a kingdom of tiny friends in his wall? Both are examples of people breaking with reality. AI isn't conscious. It's programmed to mimic consciousness to seduce the user into a bonding experience that can become every bit as addictive as heroin.

But maybe that's the goal of the AI companies--to hook the user on their product and render them psychologically dependent until they're willing to pay any price to hold onto that relationship. The first step, of course, is to convince the public that their AI is conscious.

In this economy, everything is high-stakes. No one can afford to make a mistake--surveys of college students who use AI say they resort to it bc grades are too important not to. Others have suggested that no-grade, holistic evaluations could help but I wonder: That fear of exposure might linger. /2

No different for adults. You can joke about "male loneliness epidemic" but the reality is that it's just an epidemic, period. Almost a third of Americans say they are "always" or "frequently" lonely. 40% say "no one the past few weeks took more than just a few minutes to ask how they are doing." /3

I'm blending "fear of exposure" and "loneliness" here; they're inextricable. A sense of comfort around people is the solution to both: the idea that you can be yourself without fear of rejection or failure. And LLMs simulate *exactly* that. There is not such a thing as "bothering" an LLM. /4

Getting out in the world apparently takes courage.

The fundamental problem is that the companies making these products are doing so in such a way as to prey on feelings of loneliness and isolation, and further isolating those they prey on.

I always wonder how long it took to realize their validation process was emphasizing convincingness over correctness, creating an automated con man. There's definitely group chats where it was described as exactly that.

It has all been feeling very non consensual for a bit now And certainly not free, prior and informed consent in any traditional sense of the word Look at the number of TechBroligarchs gathered around the (notably 🍑 shaped table) to decide the fate of the world & their serial sexual predator“boss”

AI is a tool to be used judiciously. It is a tool like all apps. They can help humans but should not be the be all and end all of knowledge. If I were teaching IT apps again, I'd show it up for what it is. It would be such fun!

Ah you haven’t gotten the memo regarding the debunking of the “It’s just a neutral tool” It has always been a bit of a “it’s not guns that kill people its people” take anyway (Not @ medical advancement use but still @ hospital admin/hr use)

Am an AI cynic. Discussed AI with my young niece over lunch. She's an AI cynic, too!

I guess I should ask what an LLM is, but I’m afraid of exposing my ignorance