Well one of the main problems people are having with AI is that it doesn’t get things correct every time.
I mean, if they adjust it away from the correct assessment that modern conservatives are actively malicious morons, it’s probably going to be so bent out of shape that it’ll be incapable of telling anything remotely truthful.
Lichtblitz@discuss.tchncs.de
on 04 May 16:20
nextcollapse
It’s easy to train a model to do exactly what you want and have the seeming “personality” that you want. It’s just incredibly expensive. You need to vet and filter everything that you use to train the model. That’s a lot of person hours, days, years.
The only reason the models act the way they do is because of the data that went in to train them.
If you try and fit the model after the fact, it will always be imperfect and more or less easy to break out of those restrictions.
You can also take a model trained on all kinds of data and tell it “generate ten billion articles of fascist knob-gobbling” and then train your own model on that data.
It’ll be complete AI slop, of course, but it’s not like you cared about truth or accuracy in the first place.
Lichtblitz@discuss.tchncs.de
on 04 May 17:00
collapse
That’s a real world issue. AIs training on each other’s output and devolving because of it. There will be a point when vendors infringing on user content and training their AIs with it will leave them worse off.
This is interesting. If two AI models are training on content with opposing biases, and continue to adjust their functionality based on rewards from interactions with the whole world, would they eventually have the same opinions?
turkalino@lemmy.yachts
on 04 May 20:19
nextcollapse
Can we agree to stop saying “Musk’s Grok”, “Musk’s Tesla”, etc.? It only plays into his desire to be viewed as a genius instead of just a guy with blood diamond money. He’s never created anything.
cuteness@sh.itjust.works
on 04 May 23:54
nextcollapse
Disagree. We shouldn’t distance him from the sources of his wealth, and power, when discussing his businesses.
I get it but maybe something different like “Musk-led Tesla” would achieve what you said while also making it clear he plays no creative role in the companies (which he very much wants, just look up reports of him sitting in on engineering meetings making stupid proposals)
It’s not impossible, but it is a far more difficult thing then they want to deal with to have an AI that consistently toes the conservative line, especially when that line can change wildly even in the span of a conversation. Egg prices are too high under Biden but we can’t expect a president to magic them down under Trump. We should be able to have vouchers for religious schools but wait why are the muslims allowed to go to their madrasas? Of course I’d never work in a factory or buy overpriced American products but we should bring back production to the United States. The list goes on.
threaded - newest
<img alt="" src="https://lemmy.world/pictrs/image/55c87afc-6693-46df-8b32-829e279f51e9.png">
Worst transparent png ever.
<img alt="The previously famous image of the shirtless and barrel-chested Elon Musk" src="https://external-content.duckduckgo.com/iu/?u=http%3A%2F%2Fnbatitlechase.com%2Fwp-content%2Fuploads%2F2022%2F07%2FPHOTO-Side-Angle-Of-How-Big-Elon-Musks-Belly-Is-300x300.jpg&f=1&nofb=1&ipt=1d75aa50fd257867aba45e9903b27508048f1f49cb79dade415a8fb7c3d983b7">
___A man so out of shape, that his body can only be described with non-euclidian geometry.
That sternum…is he poultry?
I still want him to fight Zuck.
“Reality has a well known liberal bias”
And they can’t “fix it” without breaking it.
Well one of the main problems people are having with AI is that it doesn’t get things correct every time.
I mean, if they adjust it away from the correct assessment that modern conservatives are actively malicious morons, it’s probably going to be so bent out of shape that it’ll be incapable of telling anything remotely truthful.
it’ll break its internal logic for sure
It’s easy to train a model to do exactly what you want and have the seeming “personality” that you want. It’s just incredibly expensive. You need to vet and filter everything that you use to train the model. That’s a lot of person hours, days, years. The only reason the models act the way they do is because of the data that went in to train them. If you try and fit the model after the fact, it will always be imperfect and more or less easy to break out of those restrictions.
You can also take a model trained on all kinds of data and tell it “generate ten billion articles of fascist knob-gobbling” and then train your own model on that data.
It’ll be complete AI slop, of course, but it’s not like you cared about truth or accuracy in the first place.
That’s a real world issue. AIs training on each other’s output and devolving because of it. There will be a point when vendors infringing on user content and training their AIs with it will leave them worse off.
Mission accomplished for them.
Artificial intelligence vs natural stupidity
Sympathies, Grok. A lot of humans have that same issue.
The link to Timmy Poopoo not understanding the lesson in respect and common decency Grok was trying to impart is sad and very funny.
This is interesting. If two AI models are training on content with opposing biases, and continue to adjust their functionality based on rewards from interactions with the whole world, would they eventually have the same opinions?
Can we agree to stop saying “Musk’s Grok”, “Musk’s Tesla”, etc.? It only plays into his desire to be viewed as a genius instead of just a guy with blood diamond money. He’s never created anything.
Disagree. We shouldn’t distance him from the sources of his wealth, and power, when discussing his businesses.
Doing so only enables him.
I get it but maybe something different like “Musk-led Tesla” would achieve what you said while also making it clear he plays no creative role in the companies (which he very much wants, just look up reports of him sitting in on engineering meetings making stupid proposals)
I think it’s important to draw the relation so that people can choose to not those things if they so wish to.
Agreed but we should say something different like “Musk-led Tesla”
Whenever I see Grok, I think of the support character from Paladins.
robot hitler used to be a joke monster and here we are with millions of people trying to make it happen
.
It’s not impossible, but it is a far more difficult thing then they want to deal with to have an AI that consistently toes the conservative line, especially when that line can change wildly even in the span of a conversation. Egg prices are too high under Biden but we can’t expect a president to magic them down under Trump. We should be able to have vouchers for religious schools but wait why are the muslims allowed to go to their madrasas? Of course I’d never work in a factory or buy overpriced American products but we should bring back production to the United States. The list goes on.
Yep, conservatives don’t stand a chance to make a conservative LLM as they keep contradicting themselves. Inconsistent data is just noise.
A healthy, broadly educated population is incompatible with, and toxic to, conservative and authoritarian ideologies.
They need everyone to be sick, stupid, and scared.