Agreed
this post was submitted on 10 May 2025
107 points (80.2% liked)
Memes
9934 readers
1040 users here now
Post memes here.
A meme is an idea, behavior, or style that spreads by means of imitation from person to person within a culture and often carries symbolic meaning representing a particular phenomenon or theme.
An Internet meme or meme, is a cultural item that is spread via the Internet, often through social media platforms. The name is by the concept of memes proposed by Richard Dawkins in 1972. Internet memes can take various forms, such as images, videos, GIFs, and various other viral sensations.
- Wait at least 2 months before reposting
- No explicitly political content (about political figures, political events, elections and so on), !politicalmemes@lemmy.ca can be better place for that
- Use NSFW marking accordingly
Laittakaa meemejä tänne.
- Odota ainakin 2 kuukautta ennen meemin postaamista uudelleen
- Ei selkeän poliittista sisältöä (poliitikoista, poliittisista tapahtumista, vaaleista jne) parempi paikka esim. !politicalmemes@lemmy.ca
- Merkitse K18-sisältö tarpeen mukaan
founded 2 years ago
MODERATORS
- You can't have unbiased AI without unbiased training data.
- You can't have unbiased training data without unbiased humans.
- unbiased humans don't exist.
If you want AI agents that benefit humanity, you need biased training data and or a bias inducing training process. E.g. an objective like "Improve humanity in an ethical manner" (don't pin me down on that, just a simple example).
For example, even choosing a real environment over a tailored simulated one is already a bias in training data, even though you want to deploy the AI agent in a real setting. That's what you want. Bias can be beneficial. Also if we think about ethical reasoning. An AI agent won't know what ethics are and which are commonly preferred, if you don't introduce such a bias.
show your work. 1 especially seems suspect. Especially since many AIs are not trained on content like you are imagining, but rather trains itself through experimentation and adversarial networks.
Even how it trains itself can be biased based on what its instructions are.
Yes, and? If you write a bad fitness function, you get an AI that doesn't do what you want. You're just saying, human-written software can have bugs.
You’re just saying, human-written software can have bugs.
That's pretty much exactly the point they're making. Humans create the training data. Humans aren't perfect, and therefore the AI training data cannot be perfect. The AI will always make mistakes and have biases as long as it's being trained on human data.
was the ai trained on reddit commenters? just asking.
It’s weird to hold the belief that AI won’t oppress us while showing it that it’s fine to oppress animals as long as you’re smarter
Every sci fi work : oh no, the technology is bad
Reality : the assholes using the tool are making it do bad things
I do think that the best government would be one run by AI.
I do not think the AIs we currently have could run a government, though.
That's just what they want us to think! /s 😜
Wait a minute... oh no no no no no no, that is what they want ~~to sell us~~ us to think! (as they game the system and control the AI, no /s no cap!)
Literally the plot of every sci-fi show with an "overseer".
https://en.m.wikipedia.org/wiki/The_Evitable_Conflict
If nearly perfect computers controlled government.
Absolutely.
Every time I hear someone question the safety of self-driving cars, I know they've never been to Philadelphia or NJ.
I mean, the US really isn't a good example for road safety. Even Germany got better drivers, and we like to drive 140-200 kmh. It's a matter of good education, standards and regulations (as always).
In the end self-driving public transport is the way the future of mobility should primarily be imho. Self-driving cars… as long as there always is a steering wheel in case of unexpected circumstances or to move around backyards and stuff it'll probably me fine. Just don't throw technical solutions at cultural problems and expect them to be fixed.
I mean, the US really isn't a good example for road safety. Even Germany got better drivers, and we like to drive 140-200 kmh. It's a matter of good education, standards and regulations (as always).
I didn't want to believe it as well, but it seems to be factually correct, as per this wonderful Wikipedia list.
Ah yes the good regulations on german public roads
They're so well regulated that they can safely drive on roads with no speed limit, whereas the US for example has pretty low limits and multiple times the fatal crashes (proportionally to population)
This. Of course it would be even better with limits on the Autobahn, and in fact a majority of people are in favour of such a change (especially if the limit is at 130-140). Our governments are in the pocket of the car industry though, politicians act as if our whole freedom is endangered talking about it (now where do we know that from? 🙃). Things can always be better, but A.I. definitely doesn't improve an absolutely shitty mobility system like the US has (which is basically nothing but cars). If anything it will make shit even more… off the rails. 😏
Yeah but then what's the point of visiting Germany as a tourist slash petrolhead?
Jokes aside, I'm of the opinion that existing freedoms are generally best left alone. Besides, Germany has a lower rate than Estonia and we have much lower speed limits. 120 on newly built separated highways in the summer (actually these might have 120 with good conditions in winter too - they have digital signage), 110 on old separated highways and in October or so, they go and collect all the 110 signs and replace them with 100... And up to 90 everywhere else.
There's a good chance the limitless autobahn is actually part of what makes German numbers so good. It just requires stricter training and policing, stricter TÜV and for people to always check their mirrors before switching lanes. And just good lane discipline in general. You don't get that in a lot of Europe. People switch lanes whenever because they're going 10 over the speed limit and can't possibly imagine someone else is going faster than them, potentially very close behind, in the other lane.
PS: traffic fun fact: Did you know that in Latvia, a two lane undivided highway has up to four active lanes? There's the law abiding citizen lanes (known as shoulders in the west) and the BMW/Audi lanes in the middle, marked by the white lines.
I mean TBF, they don't trust the average person in New Jersey to handle a petrol pump—so much so that it's legally prohibited.
I'm not at all surprised that they shouldn't be trusted with the vehicle itself, given that
AI judges make a lot of sense, that way everyone is treated equally, because eveey judge thinks literally the same way. No corrupt judges, no change in political bias between judges, no more lenient or strict judges that arbitrary decide your fate. How you decide what AI model is your judge is a whole new can of worms, but it definitely has lots of upsides.
Perhaps when we have real AGI, but I wouldn't want an LLM to decide someone's fate.
You have been found guilty of jaywalking. I hearby sentence you to 90 days of community service as unicorn titty sprinkles from Valhalla. May Chester have mercy on your handkerchief.
- JudgeGPT, probably
You could get away with murder if your lawyer talked the charges out of it's context token limit.
if it's designed right, it would be great. otherwise it would suck
And how will this be done? A proper legal system needs impartiality, which an AI still varies as much or more than a human judge. Not to mention, the way it's trained, the training data itself, if there are updates to it or not, how much it thinks, how it orders juries and parties, etc.
If, in theory, we have a perfect AI judge model, how should it be hosted? Self host it? Would be pretty expensive if it needs to be able to keep up. It would have to be re-trained to recognise new legislation or understand removals or amendments of laws. The security of it? If it needs to be swapped out often, it would need internet access to update itself, but that produces risk for cyber attacks, so maybe done through an intranet instead?
This requires a lot of funding, infrastructural changes and tons of maintenance in the best case scenario where the model is perfect and already developed. There would be millions, or ideally, billions in funding to produce anything remotely of quality.
All I see are downsides.