This is the massive elephant in the room; AI needs human-made work, published openly to further develop, and simultaneously heavily discourages human-made work to be published openly, either because the people who previously did it now use AI to do it or because the people who still do no longer have any incentive to publish it openly (and lets be honest, the incentives were pretty meagre before).
But craziest of all is that the best plan this trillion-dollar industry can come up with is "we'll figure something out later". Maybe we will, maybe AI will be able to train off synthetic data or we'll achieve the perfect model that never needs improving or we'll simply hit a wall with how much training data we need, but for a problem that an existential threat to LLM's, that's a lot riding off a few maybe's
This is awful. It’s already hard to trust information online and this will make it even more opaque. Imagine you are curious about a war or political movement or event for example and the only way to get informed would be the chat box of a handful of mega corporations.
The real elephant is that we've "funded" "content creation" (barf) for 20 years on the back of advertising. That is the long-term unsustainable problem, not AI.
Having dealt with many family members just like this, I can tell you they certainly can use Google for search, they just don't want to. When forced to they use it fine.
If I were to guess, a lot of people, particularly older people, see computer use outside the web browser as an amateur might a chop-saw; a dangerous tool not to be used by inexperienced hands, so an AI that can also directly do the tasks for them might solve that problem (assuming it came preinstalled or someone else did that for them). Though then comes the question of whether they'd be willing to pay for it
Plus hallucinations. Most "regular people" questions aren't inherently easy to verify. And when regular people do have to verify the answers, the value of "AI" drops like a rock.
> Having dealt with many family members just like this, I can tell you they certainly can use Google for search, they just don't want to. When forced to they use it fine.
Exactly! I noticed this when I moved out of my parents' house, when suddenly my mom became an electronics whiz who can hook up VCRs and TVs just fine. I always thought I was the smart person in the house, but now I realize I was just taken advantage of.
Not to justify those problems, but there are direct comparisons of those problems in the US. More recently with the human rights issues with ICE, though this is certainly not a new issue it has been ramped up over the years, but territorial ambitions, though less direct, have been US policy for over 60 years. Recently you have Iran and Palestine of course, but the US has been overthrowing democratically elected leaders in particularly South American countries for a while now. As I say, less direct than sending troops in and taking over literally, but installing a puppet leader in these countries is certainly grounds for "territorial ambition".
As I say, I don't want to dismiss what China is doing as "well the US are doing it too", it's just worth pointing out if that's a red line for a global power
Parent's comment was "Let's hope China are benevolent world superpower" - I provided grounds why they (China) are not, you've provided grounds for why the US is not. The two are mutually exclusive insofar as the US' state of benevolence has no bearing on China's.
Deliberately or not, your comment is an example of whataboutism [0]
> The two are mutually exclusive insofar as the US' state of benevolence has no bearing on China's.
All due respect, I literally said that, twice. I suppose you could still call it whataboutism by technicality, but it doesn't quite have the same ring to it.
My point was that replacing a benevolent superpower with one that is not is quite a different situation than a "meet the new boss, same as the old boss" one, and I was just pointing out that this is the latter, not the former.
> disclaiming a whataboutism doesn't somehow negate elaborating on it
I think it largely does. Look at the stated purpose of it in the wikipedia article you linked:
> It is an informal fallacy that the accused party uses to avoid accountability—whether attempting to distract by shifting the conversation's focus away from their behaviour or attempting to justify themselves by pointing to the similar behaviour
Pointing out that China is still responsible for its actions and the US doing it too doesn't justify them makes it whataboutism without a purpose. Like I say, whataboutism by technicality, rather than as a counter-argument.
> I didn't get that from your comment; it rather came across as a laundry list for why the US is not benevolent.
I think it can be quite easily inferred. We don't want China as a superpower because of X. The US is a superpower. The US also does X. Therefore...
Again, I said this at the end:
> it's just worth pointing out if that's a red line for a global power
> The training data will degrade in quality if more people rely on LLMs so I'd expect models to eventually start getting worse even if they currently enjoy apparent improvements in output.
Its not a direct analogy, but this seems to me as like saying "I already know maths and english, why do we still hire and pay teachers". Its a massive issue that just seems to be hand-waved away as "we'll figure it out later".
To compound on the problem as well, even those of us who do still write code by hand are actively discouraged from sharing it, knowing no human will ever read it and we'll never be credited for our work, it'll just go in the pockets of a few big AI companies. I suppose there will be some that take some pride in knowing their code is helping AI models advance, but thats a small pool of an already small pool, assuming most people decide to just use the AI to write the code for them. A minority of a minority.
I guess this is the chickens coming home to roost of this whole attitude towards growth now, fallout later
I gave away my base spec (8gb ram, 256gb ssd) M1 air just a few months ago and it's still working fine, I just needed something higher than base spec for work but the new owner of the M1 loves it, for general purpose use it may as well be an M5.
I do wish they had removable storage though, I originally got that Macbook as it was the cheapest way I could try out MacOS and M1, once I decided I liked it, I was stuck with 256gb or spending all that money (and then some) again. This time, I went with 1tb.
From what I can tell, it's the thing they notice more than almost anything else. I was recently shopping for a new phone and my brother, who has no interest in tech, kept telling me I need to get the one with it (I forgot the exact term he used, it wasn't 120hz admittedly).
Off topic, but it's interesting how Recraft might've accidentally produced the closest thing to actual art from an AI model at least that I've ever seen. Is there any meaning to it at all? Is it taking "controversial take" as an unflattening image? Theres actually something to engage with it.
Unfortunately for the pro-ai-art people, even in this case it wouldn't make them the artist, as the "mind" of interest is the AI itself.
You might be interested in the Pebble Index 01, thats pretty much what that is, plus it's a smart ring so it's easy to keep on you, and can do local transcriptions. Downside is that it's disposable, it has roughly a years battery life and cannot be recharged
For some it certainly is, I remember it being very controversial (especially for Pebble) at release, but it's not something I'm particularly concerned about. Considering the current options out there, it's the best for privacy and thats more of a concern for me. If they stop producing them by the time mine dies, I've lived long enough without it, I'll be fine.
Can certainly see why others wouldn't be such a fan though
But craziest of all is that the best plan this trillion-dollar industry can come up with is "we'll figure something out later". Maybe we will, maybe AI will be able to train off synthetic data or we'll achieve the perfect model that never needs improving or we'll simply hit a wall with how much training data we need, but for a problem that an existential threat to LLM's, that's a lot riding off a few maybe's
reply