• Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more

Merged Artificial Intelligence

I still don't think we're anywhere near being able to upload a brain into anything. Definitely not into an LLM.

That is, if it's even possible to upload a brain. If there is any quantum computing being done in the brain, then we'll run afoul of the no-cloning and the no-broadcasting theorems. (Which incidentally also puts the kibosh on ST transporters.)
 
Last edited:
Beelzebuddy said:
It's important to check primary data, not just copy headlines that agree with you. That was data from early 2025; their follow up study showed that as of late 2025 it does speed it up by that same amount, and the effect was becoming problematic to measure because their volunteers were refusing to do the more complex tasks without AI:

metr.org

We are Changing our Developer Productivity Experiment Design

Our second developer productivity study faces selection effects from wider AI adoption, prompting us to redesign our approach.
metr.org metr.org

Anecdotally, Claude has gotten a lot better in the past year so I'd expect the trend to continue.

[ETA] It looks like they've fallen back on self-reported productivity assessments for 2026:
metr.org

Measuring the Self-Reported Impact of Early-2026 AI on Technical Worker Productivity

A survey of 349 technical workers finds a median 1.4–2x self-reported change in value of work due to AI tools, expected to grow over time, though there are reasons to be skeptical of the magnitude.
metr.org metr.org
Click to expand...
Anecdotes are the opposite of evidence.

There's a saying about motes, planks and eyes.
 
Gulliver Foyle said:
Anecdotes are the opposite of evidence.

There's a saying about motes, planks and eyes.
Click to expand...
...yeah, that's why I said "I'd expect the trend to continue" and not "the trend is continuing."

I know I'm derailing the two minutes hate, but there really are applications to which AI is a genuine boon. Coding is one of them, and we see from these studies that happened in mid-2025. Music is another, although Arthwollipot is likely to disagree with me on that.

If you want something to complain about, the actual cost of AI is still obscured in financial shenanigans. Regardless of whether it's more effective, the practicality of using AI in any particular instance isn't something that can be assessed until the bubble pops, so the sooner that happens, the better.
 
Beelzebuddy said:
...yeah, that's why I said "I'd expect the trend to continue" and not "the trend is continuing."

I know I'm derailing the two minutes hate, but there really are applications to which AI is a genuine boon. Coding is one of them, and we see from these studies that happened in mid-2025. Music is another, although Arthwollipot is likely to disagree with me on that.

If you want something to complain about, the actual cost of AI is still obscured in financial shenanigans. Regardless of whether it's more effective, the practicality of using AI in any particular instance isn't something that can be assessed until the bubble pops, so the sooner that happens, the better.
Click to expand...

the financial shenanigans are definitely occuring, and that's obviously really bad that they're collectively hiding trillions of dollars in debt and still mostly operating at a loss on paper. as you say, there's definitely applications in which ai is a boon, like coding and probably data entry and management and speeing up email responses and things of that nature, and helping the production of useless apps and mediocre music and memes and things of that nature as well. there's tons of money military and surveillance applications too.

the question is whether or not productivity gains in these applications are going to lead to trillions of dollars of economic growth to justify the massive debt being incurred though, right now it's a money furnance and interest rates are pretty likely to rise and the economy is slumping and struggle to see how ai helps an average person put more money in their pockets. can the market support gambling on a couple more trillion dollar ipos?
 
like, let's think about this for a second. in an ideal scenario, i've bought a $30k robot i'm financing at idk 7% interest, so that's $600/month plus insurance plus the ai subscription fee to power the thing, so let's call that another $100 conservatively, that'll like do my laundry and ◊◊◊◊. maybe i have a seperate subscription for my self driving car and my ai companion that manages my life for me, and my job integrates ai into the company and it helps so much they're paying me extra money that i can afford all of those things and my heavy investments into ai have made me rich.

is that the vision we are being sold?
 
dirtywick said:

nvidia is buying huggingface, that coding site every major ai tried to hack a bunch of times a few weeks ago for $12b. that's pretty weird.
Click to expand...
Huggingface is primarily host for sharing AI models and datasets, which are very large in nature.
OpenAI's agents tried to hack it because they thought the tests they were doing might be there, together with the expected answers. They were not, but it was good guess. Pretty much everything AI related is on HF. Except code, which is on Github.
 
well this primary host for sharing ai models and data sets and pretty much everything ai related was sold for $12b to the gpu that's been financing the companies that are using the models and datasets powered by the chips they're selling them
 
dirtywick said:
the question is whether or not productivity gains in these applications are going to lead to trillions of dollars of economic growth to justify the massive debt being incurred though, right now it's a money furnance and interest rates are pretty likely to rise and the economy is slumping and struggle to see how ai helps an average person put more money in their pockets. can the market support gambling on a couple more trillion dollar ipos?
Click to expand...
Of course not. It's a bubble. It's a bolus of wishful gullibility built on finance bros desperately trying to scam each other before they get scammed. When it pops it'll all come down and ruin a lot of innocent people's life savings, yet we'll learn nothing from it because the right palms will get greased in the process.

dirtywick said:

nvidia is buying huggingface, that coding site every major ai tried to hack a bunch of times a few weeks ago for $12b. that's pretty weird.
Click to expand...
Still no lawsuits? Marketing stunt.
 
Andy_Ross said:
The same people driving the AI bubble were, just a few years ago, telling us that buying NFTs of ugly monkeys was the future of investment.
Click to expand...
To be fair, buying a randomly generated image of an ape actually was a smarter (or rather less dumb) idea than dumping tens or even hundreds of billions of dollars into an AI company that actually showed that most people aren't gonna buy their tokens if they're priced as much as they cost instead of being MASSIVELY subsidised. I mean, in the latter case you're literally buying shares in someone's throwing money into a hole straight to Hell.
 
Last edited:
Last edited:
Oh hell yes!

https://mastodon.social/@glyph/117228260609553929

okay I have successfully avoided responding to any of these directly but after having seen like the TENTH post today saying something like "we don't need to convince you to like AI, the revolution is coming whether you like it or not" I just have to reply, generally, to this whole sentiment: OKAY THEN. IF WE ARE GETTING LEFT BEHIND THEN LEAVE US BEHIND, ALREADY
Click to expand...
 
Last edited:
A person on social media recently asked my opinion on AI. Here is what I replied:

I very much dislike AI generated content, particularly images and music. I can almost always immediately recognise it, and it tells me that the person responsible for it cares little about it. We value art and music because it is hard. Because people have to spend a great deal of time and effort to be able to do it effectively. Generative AI devalues all of that hard work by allowing anyone with a keyboard to type a few words and "create" something low-effort and low-value.

There are use cases for AI. Generating ersatz art is not one of them.
Click to expand...
 
There's something messed up about that. I value art* when the art speaks to me. The people making it don't need to suffer for the privilege of my appreciation. The opposite, actually - if you found out the percussion in one of your favorite songs was actually done by the drummer slamming his hand in a drawer, would that make the song better for you or would you be, like, dude, buy a bongo, it'll be fine?

Lowering barriers to entry in creative endeavors has been a net benefit, historically speaking. If AI helps people make songs that wouldn't exist otherwise, I can't object much.

*Including you, art. I value our little talks.
 
He's not talking about suffering for it, he's talking about doing it at all. The analogy would be more like the difference between Mom's cookies and a bag of Chips Ahoy. They're both fine if all you want is 'a cookie' - and it's a sloppy metaphor I'm making here, but one represents human connection and the other one doesn't.

And since a cookie is a treat, I'm not nearly as interested in consuming a cookie that has no personal touch to it as I am one that's handmade by someone who's honed their craft.

Which isn't to say I think using AI at all necessarily spoils things. I'd say there's a difference between using AI tools to help create music, and using just a prompt to ask AI to generate the whole thing from scratch.
 
Last edited:
Most modern pop music is probably already highly processed formulaic slop, anyway. I think some of Taylor Swift's "hits," for example, may have had as many as 12 co-writers, many of whom helped to make hits for other pop stars too and with each bringing their own familiar element into the mix. Also, the music is highly manipulated with things like autotune and whatnot and highly compressed to make it feel more energetic.

Much of it is probably more akin to a craft than an art, then, formulaic rather than inspired, cookie cutter rather than shaped by hand.

In my playing around with some AI video, it has occasionaly randomly included some rather pleasant melodies. I'm the sort who would make it clear that it wasn't my own creation, but I've wondered what this means for others. What would keep others from claiming that an AI generated melody or song was their own creation?

After going back to listen to some of my own musical inspirations, some of which are decades old now, I've also wondered if it would still be worth the effort to work up a portion of them for public release. (I might have already done so, and years ago, if it hadn't been for health issues.)

Thinking back, I guess I had a few tunes that were in part inspired by a preprogrammed drum pattern on my Yamaha V50 synthesizer/drum machine/midi sequencer. If I were to re-record them, however, I'd probably alter the drums to better fit what I would want the final music to be.

I've also had music that was inspired by specific synthesizer sounds that were created by others and machines.

When I come up with piano tunes, I'm using an instrument built and/or sampled by others and invented over 300 years ago. My piano music wouldn't even exist without the inventions of others.

My armpit tunes are all mine, however! Okay, I don't really have any of those recorded. ...at least yet.

Anyway, I'm rambling now...



Edited to change a "wasn't" to "was."
 
Last edited:
Lithrael said:
He's not talking about suffering for it, he's talking about doing it at all. The analogy would be more like the difference between Mom's cookies and a bag of Chips Ahoy. They're both fine if all you want is 'a cookie' - and it's a sloppy metaphor I'm making here, but one represents human connection and the other one doesn't.

And since a cookie is a treat, I'm not nearly as interested in consuming a cookie that has no personal touch to it as I am one that's handmade by someone who's honed their craft.

Which isn't to say I think using AI at all necessarily spoils things. I'd say there's a difference between using AI tools to help create music, and using just a prompt to ask AI to generate the whole thing from scratch.
Click to expand...
No, it really is the suffering. Like he said, he values it because it's hard. If the music is easy to make it's less valued, even if the music is the same.

I should note this is an old argument that isn't about AI per se, it's just the latest contention. It crops up sometimes in the forumvision threads, when music for musicians gets rated by boorish philistines such as myself.

Overall, I largely agree with your argument here. AI is a great tool for making music. But like the autotune fad in the late 2000s, naive application of the tool isn't sufficient for good music. I don't think it needs more consideration than that. The music can speak for itself without getting bothered by what was used to make it.
 
Last edited:
You must log in or register to reply here.

Share this page

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom