Aerosmith removed from roller coaster preshow at Disney's Hollywood Studios

Lord Gonchar:

No one seems to mind.

Really? You have different friends than I have - everybody I know is fed up with "the algorithm."

Yes, this. My own journey is that I realized how much time I was spending not allowing myself to be bored, when I actually was bored in the first place, but not enough to really seek out something else. I wrote about this recently, but I don't think people appreciate how useful boredom is.

Re: Remix... I know we're already in it, and as I suggested, it's ****ing boring. I think people can and will see this more and more when they happen upon "good" art.

And have you heard that AI country song that actually charted? Even I recognize how terrible that is, leaning into every country cliche there is. People like it because they're too lazy to find something better, for now.

I mean, Rebecca Black's "Friday" existed years before AI, so there's always going to be crap that charts. In fact, I've argued that most pop is crap not long after I started noticing music. AI just makes it easier to make.

Not sure if I even have a point or am taking a side, but...yeah. Carry on.

I got nothing.

Whole communities already exist that share AI creations and enjoy consuming those creations. Pop music doesn't exist because people are lazy. Hating the algorithm doesn't mean it doesn't influence (or guide) your consumption on a daily basis. Generative tools to create are just that - tools for creation.

We're not even playing the same game at this point.

If you think ANY of this is going away? I dunno what to tell you guys.

eightdotthree:

Maybe I’ll be wrong. But the quality of these models are declining in my experience.

It only took one sentence! (kidding)

hambone:

You have different friends than I have

Clearly.

I missed the the day we all put our phones down and doomscrolling was declared dead.

Jeff:
I think people can and will see this more and more when they happen upon "good" art. People like it because they're too lazy to find something better, for now.

The amount of gatekeeping and condescension in that is icky.

Also, it's like questioning why a kid would want a Snickers instead of Broccoli. Like, really? You don't understand why pop music is popular?

With all of that said, I think I'd like to bow out now. Feel free to point and laugh as I leave the room and head outside where all of these things are actively happening.

That last sentence was definitely "icky."

Lord Gonchar:

If you think ANY of this is going away? I dunno what to tell you guys.

I don't think anyone is arguing that. I'm sure there are online enthusiast clubs for rubber dog ****, but it doesn't mean that rubber dog **** is art. I imagine that's the point that I and others are making.

The amount of gatekeeping and condescension in that is icky.

The irony of the Internet is that access to everything for everyone circumvented the gatekeeping that the industry used to enjoy, but the sheer volume of crap has made it harder, not easier, to find the good stuff. If there's one thing that I miss about terrestrial radio, it's that there was value in the curation (for most genres). Even in college radio, we would sift through 50 songs every week, and we'd find a few gems.

Jeff:

there was value in the curation

There still is. I get what you're saying here and couldn't agree more.

But in the same breath, the idea of an 'authority' or 'tastemaker'...

...well, that's just the algorithm, isn't it?

Jeff:
I'm sure there are online enthusiast clubs for rubber dog ****, but it doesn't mean that rubber dog **** is art. I imagine that's the point that I and others are making.

We're defining art? Yeah, then I missed the point entirely.*

Who exactly gets to make that call? Because we've only gotten as far as rubber dog**** and I already disagree with you.

---

*I really didn't miss the point. I added that for effect there. Way back wherever in this thread, I think the argument was that you had to actually do something for the art to count or some other nonsense and AI was too good of a tool, so the user wasn't actually creating art because they didn't do enough. But even with this caveat, the point remains. You have to be a pretty pompous tool to think you can define art for others.

"I can't define art, but I know it when I see it" - everyone who ever existed

"Art" I'm sure lies on a sliding scale, you're favorite. We don't have to agree about what it is or isn't.

The algorithm isn't a curation device. Algorithms by definition are not intended to challenge you or introduce you to new things, just keep you in your comfort zone. That's why racist old people like "social" networks, or create cozy echo chambers. The intent is engagement, not discovery. Zuck doesn't care about whether or not you find new things, so long as you keep doom scrolling. I'm sure Ek feels the same way, whilst paying the artists as little as possible.

So if AI drives algorithms, it's just a more finely tuned way of driving engagement.

Jeff:

The intent is engagement, not discovery. Zuck doesn't care about whether or not you find new things, so long as you keep doom scrolling.

That's usually the intent with curation too - at least in the sense you mentioned it.

You share what you believe your audience will think is the good stuff so people stay engaged.

(We're admittedly interweaving a few different ideas here, but...)

And if the idea is "I know better, people, listen to this 'good' music" - then I would counter with, that's exactly why DJs don't exist anymore and computers run stations with algorithm generated playlists.

I guess maybe the disccusion is shifting to how do people find "new" or "different" experiences (in this case art)?

I think in the recent past, most people just watched TV and listened to the radio - which honestly isn't much different than what the algorithm feeds them now. That same racist old person (holy stereotype, batman!) used to have to tune to Fox News (and there's my contribution!).

Now it comes to them via algorithm. If you think people didn't live in cozy echo chambers before the algorithm, I think you're giving too much credit. Hell, with the algorithm, their exposure to other ideas is still probably better than it used to be - that's why they're so pissed.

"If you like this, then you'll probably like this" isn't really a new or novel or a broken concept or a dirty idea. It's sort of the first step on the path of discovery.

That's all curation really is.

And that's all the algorithm is doing. It's just that it's too good at it.

Just like AI is too good at quickly and easily making pop songs.

Honestly, it might be that people just get mad when we make the machines too good at stuff...and then we get used to it.

Look, YMMV, but this facebook page is not evidence that the machines are too good at it, and no human would have suggested it to me.

Fair.

Your one example is a solid one.

(and again, I still haven't learned when we scoff at anecdotal evidence around here and when we gleefully accept it)

But surely you understand why a page covered in roller coasters and theme park photos and "information" that is liked by 36,000 people would be suggested to you.

It's the equivalent of your grandmother buying you this book because she knows you like this sort of thing and the guy at the bookstore recommended it.

The logic isn't wrong. It's just misguided.

Politely thank your algorithmic overlord and move on, just like you would with Grandma. (she's probably racist and watches Fox News anyway...like all old people)

Lord Gonchar:

I still haven't learned when we scoff at anecdotal evidence around here and when we gleefully accept it

We scoff when it doesn't fit our narrative and gleefully accept it when it does. Duh.

Lord Gonchar:

But surely you understand

I do. And it’s why my typical facebook visit lasts less than 30 seconds these days.

And I’m just snotposting here; I don’t know what the future holds and what’s good.

As you were, troops.

Lord Gonchar:

You share what you believe your audience will think is the good stuff so people stay engaged.

That's not at all what we did in college radio. College music enthusiasts (especially Gen-X) didn't care about engagement. We didn't need ratings or to be popular. Popular wasn't cool back then.

When people rarely traveled more than a few miles from where they were born, and all of their learning came from their community, yes, it limited their world view. I wouldn't say that it reinforced it though.

Now we have access to everything, an overwhelming amount of information. And instead of engaging in critical thinking and making their world larger, people tend to just lean into what they want to believe. Bringing it back to algorithms, and AI, we seem to confuse the technology with expertise. In fact, society seems to be actively rejecting expertise. The machines are not experts, especially when the machines are trained on the loudest voices, instead of the factual voices.

Applied to art, sure, the subjectivity of it all makes things a lot fuzzier. But it doesn't mean that the tech is better at making or recommending the best art.

I'm not even sure what we're debating at this point. LLM's are starting to cap in usefulness, in part because they suck at context and tend to revert to the worst of human behaviors. There are some purpose driven use cases (writing code, scientific modeling of all kinds) that continue to improve, but general AI is mostly responsible for noise, in greater volume than humans alone could achieve. That doesn't feel like an improvement.

Jeff:

I'm not even sure what we're debating at this point.

Me either. But that's the only reason I come here anymore. I have fun with y'all.

LLM's are starting to cap in usefulness...

****ty answer? The tool is only as good as the person using it.

Good answer? I don't have one.

but general AI is mostly responsible for noise, in greater volume than humans alone could achieve.

Yes!

And the nihilistic corner of Gonch LOVES this. In a perfect world it would make the internet useless beyond specific tasks/tools (like getting online to pay a bill or to order a new widget or stuff like that) and general online "discovery" would be like wading through the scariest and dumbest inner city back alleys at 2am. Thus making the internet and connectivity a tool and bringing the focus back to real world ****. It was objectively better when online was the window, not the other way around.

But realistically, we'll push the AI to some directive that it interprets in a way that kills us all as casually and quickly as any of us kill countless living species every time we mow the lawn.

So we have that going for us.

Software is kinda my job, so I'm closer to what AI can do than maybe a lot of people. I hate that I can't really talk about specifics relative to work, because the generalities aren't great for making a point about anything. But consider this: I read today that 95% of investors in AI ventures haven't had any return yet on their investment, and there's no expectation about when that might happen. It feels an awful lot like 2000. What exactly is the business model for all of this, when people want everything to be free? I don't know. In software, I imagine that most companies are buying licenses for agentic services (Claude, Cursor), because as I said, there's a solid use case that's well defined.

My hot take is that a lot of these companies with giant evaluations building out a ton of data centers will die, and the tech companies that do other stuff (Google, Microsoft, Amazon) will buy up the pieces. Meanwhile, the hardware will get cheap enough that companies who need a tuned AI solution to whatever they do will be able to run models on their own hardware. The reason that I think this is that a company like Rivian is building an insane amount of computing power for their vehicles with custom silicon, able to conduct billions of operations per second. Heck, our phones (well, some phones) can do on-device language translation in real time. Again, tuned use cases > general generative AI (and obviously > RRR).

And the tools can definitely fail people who are good at stuff. I've seen it in the agentic coding scenarios, but even trying to get a basic two-color logo out of the image generators is a terrible experience. It's easier with my weak-ass skills to just bust out Illustrator and do it myself.

Jeff:

And the tools can definitely fail people who are good at stuff.

I give my exams online. I let students use The Internet. I tell them not to use anything that claims to be intelligent, but I know some of them will anyway. Amusingly, the better models typically get a solid D on my exams, though that is partly by design.

Brian Noble:

...is partly by design.

I'd love to know more.

I was looking at a local place on Google and read some of the reviews. The manager responds to some and they're clearly using AI. Here are just two of them.

We're so happy to hear that you enjoyed the Best Bites tasting event and found our selection of dips to be a pleasant surprise. It's great to know that the roasted red pepper dip was a hit for you, and we appreciate your feedback on the beet dip as well.

We're delighted to hear that you enjoyed the atmosphere and service at {business}, as well as the garlic shrimp. However, we're sorry to learn that the wings and burgers did not meet your expectations. Your comments are invaluable to us, and we will certainly pass them along to our culinary team for review.

There is more slop but this gets to what is so inhuman. They surely didn't read that before sharing right? I'd imagine students are doing the same?

Honestly, if AI gave me a response that poor, I'd be on your guys' side.

That's SO bad. It's almost like a basic generated autoreply within the review system more than an actual ask of a competent LLM.

I don't think it's AI. I've been seeing those canned responses to bad apartment/restaurant reviews long before AI was a thing.

A review describing an apartment like it's the 7th level of hell followed by "We're sorry we didn't meet your expectations, please call us to discuss blah blah" .. meanwhile in the review itself: "They're never in the office or answer the phone"