[Syonyk]

And this is why I hate AI so strongly (I... debated several less charitable wordings of this opening).

I read, because I want to know what people think. What people thought, decades, centuries ago. Writing, and reading, are about the only mechanism we know to really get inside someone else's brain. To see how, and what, they think, in what they've chosen to communicate. I used to write more for a technical audience - I've got a blog, it's got 9 years of content, a lot of words. Lately, I've been writing more for myself, but also expecting that, at some point, I'll share some of this with other people. I've read things I wrote months back, and I'd forgotten a lot of what I put in there. It's a time machine, to see myself as I was last summer.

Or, a time machine, to understand someone's thinking in the early 1900s. Or older. We have writings that are thousands of years old. The grief of loss is something that's as old as human history, and there's no shortage of older writings on the matter - Psalms has no shortage of things to say.

I don't want to read "Machine, please generate me a chunk of text as would be written by someone who lost their daughter ten years ago, when they were turning six." There's no human behind it. I'm sure you'd get something that pulled the proper strings, but it would be empty. It would be hollow. Because there is no human behind it, with their own sufferings, dreams, hopes, grief. I'm sure it would be polished and bland and SEO optimized, but it wouldn't be real.

But, hey, maybe someone could make a buck off generating it!

[nullc]

but AI is telling you what 'people' think-- not a person-- but a blended up mismash of people, like the move Dark City, then strapped into a torture chair to diminish wrongthought. But it's people. or peopleish. It was made by people at least. sorta.

Besides, after being filtered through perception and narration who is to say any story you hear from another person is real?

Shouldn't it be more a question of what it does for you? And if some amorphous nonphysical branding of Realness(tm) and Authenticity(tm) is what you value most, then ultimately someone is marketing to that. And paradoxically it means that the most "real" stuff won't tend to be real at all.

So I think ultimately the only thing real is going to be the stuff you experienced yourself or comes from acquaintances you know well enough to know when they're not being earnest. I hope won't be so bad, after all it was the condition of all humanity until not so many years ago.

[Syonyk]

There's a reason the phrase, "Designed by committee", is not a praise. It applies to AI, too.

I don't want a mash-up of generic human thought, as filtered through a machine. I want to hear what people think, as individuals.

[foxglacier]

But there's no reason it can't emulate that. What people themselves think is mostly just a mashup of the crap they hear from other people anyway. It's also usually filtered for wrongthink, at least until they get to trust you enough.

I had this realization of how fake most people are when I started to use AI to talk online with people from a real-life-group I belonged to. I got it to talk the way they did and they liked it more than the real me. I then realized they're probably all just applying their own meat-AI filter to their own words anyway and not actually exposing their real feelings. I also know people who carefully draft and craft all their text and email messages to convey the feelings they want to show, whether they're honest or not.

[djeastm]

You're right. The sad thing to me is there's not going to be much of a way to tell if the text we're reading, the picture we're looking at, or the person we're talking to even IS a real person unless they're physically in front of us.

It's a huge loss for interpersonal connectivity at a distance

[foxglacier]

I hear this a lot, but why is it a problem? For most people I encounter ephemerally online, as long as I can't tell the difference, it doesn't matter. If you were an AI, I wouldn't mind. I usually talk to such people to try to understand ideas better and if an AI could provide that, then no problem. If you want some sort of real-world interaction with them, there's a good chance that'll never happen with a real person either. I've had many online friends who I never met but I still felt like I knew them. The AI just has to be good enough never to burst the bubble.

Then there's the opposite problem that we already have to deal with - real people who are effectively AI and not capable of forming a mutual relationship with you, such as narcissists, people with dementia, and celebrities. People can spend years of their lives having a relationship where the other person only really exists in their own mind. Actually it's quite normal to build an internal imaginary model of the people we know which might not actually be correct.

[dyauspitr]

Just wait until AI can feel pain- you know it will happen eventually when they have physical sensors and enough compute.