Knowing how badly Tesla has managed their supply chain, it’s possible they paid for it in 2016.

Are you saying that Trump has no interest in changing our cyber security stance to favor Russia? Or that he’s powerless to do so?

I upgraded immich without breaking everything. That’s always reason to celebrate.

deleted by creator

I’m waiting on two things from them and then I’ll be truly happy (I use it anyway)

1. A stable branch release so no more breaking changes

2. The ability to upload single photos to an album from the mobile app, not run a backup.

Yep, I understand but disagree. Maybe it comes from working with so many ESL coders, but I’ll happily accept typo corrections because it’s not always obvious what words should be if you’re not steeped in the culture.

Narrow fine-tuning can produce broadly misaligned

It works on humans too. Look at that fox entertainment has done to folks.

“comments must be accurate,” is not a rule you should bend. Bending it even a little leads to last programming and shit code.

Consensual training data makes it ok. I think AI companies should be accountable for curating inputs.

Any art is ok as long as the artist consents. Even if they’re drawing horrible things, it’s just a drawing.

Now the real question is, should we include rapes of people who have died and have no family? Because then you can’t even argue increased suffering of the victim.

But maybe this just gets solved by curation and the “don’t be a dick” rule. Because the above sounds kinda dickish.

It just means I don’t use AI to create porn. I figure that’s as good as it gets.

Yeah, I’m ok with that.

AI doesn’t create, it modifies. You might argue that humans are the same, but I think that’d be a dismal view of human creativity. But then we’re getting weirdly philosophical.

We could be sure of it if AI curated it’s inputs, which really isn’t too much to ask.

I just had this argument with my HR team. I want to hire some people straight out of college and train them. They want to spend 15k more and get someone with experience.

But where do those people magically come from?

Watching videos of rape doesn’t create a new victim. But we consider it additional abuse of an existing victim.

So take that video and modify it a bit. Color correct or something. That’s still abuse, right?

So the question is, at what point in modifying the video does it become not abuse? When you can’t recognize the person? But I think simply blurring the face wouldn’t suffice. So when?

That’s the gray area. AI is trained on images of abuse (we know it’s in there somewhere). So at what point can we say the modified images are okay because the abused person has been removed enough from the data?

I can’t make that call. And because I can’t make that call, I can’t support the concept.

without a victim

It was trained on something.

Imagine being so naive that you think that’s a real person.

But if we’re not being snarky for a moment… It’s trivially easy to create a bot to do exactly what this person is doing. Spam others with begging for money and a bunch of sob stories.

This is why I turn off Google spam filtering. My attention is worth nothing so everyone who can message me should be able to.

I bought oblivion six years ago in anticipation of this release. Just another couple of years until it’s done.

Computing is only political if you don’t want the government in your machine

As much as I hate to say it… I use Facebook groups. They’re usually really well established and very local.

So maybe look for a fediverse Facebook replacement like frendica?