Google’s AI-driven Search Generative Experience have been generating results that are downright weird and evil, ie slavery’s positives.
This is the best summary I could come up with:
Not only has it been caught spitting out completely false information, but in another blow to the platform, people have now discovered it’s been generating results that are downright evil.
Case in point, noted SEO expert Lily Ray discovered that the experimental feature will literally defend human slavery, listing economic reasons why the abhorrent practice was good, actually.
That enslaved people learned useful skills during bondage — which sounds suspiciously similar to Florida’s reprehensible new educational standards.
The pros included the dubious point that carrying a gun signals you are a law-abiding citizen, which she characterized as a “matter of opinion,” especially in light of legally obtained weapons being used in many mass shootings.
Imagine having these results fed to a gullible public — including children — en masse, if Google rolls the still-experimental feature out more broadly.
But how will any of these problems be fixed when the number of controversial topics seems to stretch into the horizon of the internet, filled with potentially erroneous information and slanted garbage?
The original article contains 450 words, the summary contains 170 words. Saved 62%. I’m a bot and I’m open source!
“the X post”
lol
Cross-posted where?
I also keep reading it as cross post
If you ask an LLM for bullshit, it will give you bullshit. Anyone who is at all surprised by this needs to quit acting like they know what “AI” is, because they clearly don’t.
I always encourage people to play around with Bing or chatGPT. That way they’ll get a very good idea how and when an LLM fails. Once you have your own experiences, you’ll also have a more realistic and balanced opinions about it.
Wtf are people expecting from a fucking language model?
It literally just Mathematics you a awnser.
And acting like there are no upsides is delusional. Of course there are upsides, or it wouldn’t have happened. The downsides always outweigh the upsides of course.
A few lawyer thought chat gpt was a search engine. They asked it for some cases about sueing airlines and it made up cases, sited non existing laws. They only learned their mistake after submitting their finding to a court.
So yeah people dont really know how to use it or what it is
I mean it’s how we got shoes phones and batteries for recent times.
People think of AI as some sort omniscient being. It’s just software spitting back the data that it’s been fed. It has no way to parse true information from false information because it doesn’t actually know anything.
What’s more worrisome are the sources it used to feed itself. Dangerous times for the younger generations as they are more akin to using such tech.
What’s more worrisome are the sources it used to feed itself.
It’s usually just the entirety of the internet in general.
Well, I mean, have you seen the entirety of the internet? It’s pretty worrisome.
The internet is full of both the best and the worst of humanity. Much like humanity itself.
And then when you do ask humans to help AI in parsing true information people cry about censorship.
The matter of being what is essentially the Arbiter of what is considered Truth or Morally Acceptable is never going to not be highly controversial.
Well, it can be less difficult, but still difficult, for humans to parse the truth also.
What!?!? I don’t believe that. Who are these people?
removed by mod
While true, it’s ultimately down to those training and evaluating a model to determine that these edge cases don’t appear. It’s not as hard when you work with compositional models that are good at one thing, but all the big tech companies are in a ridiculous rush to get their LLM’s out. Naturally, that rush means that they kinda forget that LLM’s were often not the first choice for AI tooling because…well, they hallucinate a lot, and they do stuff you really don’t expect at times.
I’m surprised that Google are having so many issues, though. The belief in tech has been that Google had been working on these problems for many years, and they seem to be having more problems than everyone else.
Even though our current models can be really complex, they are still very very far away from being the elusive General Purpose AI sci-fi authors have been writing about for decades (if not centuries) already. GPT and others like it are merely Large Language Models, so don’t expect them to handle anything other than language.
Humans think of the world through language, so it’s very easy to be deceived by an LLM to think that you’re actually talking to a GPAI. That misconception is an inherent flaw of the human mind. Language comes so naturally to us, and we’re often use it as a shortcut to assess the intelligence of other people. Generally speaking that works reasonably well, but an LLM is able to exploit that feature of human behavior in order to appear to be smarter than it really is.
You know unless we teach more critical thinking, AI is going to destroy us as a civilization in a few generations.
I’m more worried that happy educated citizen stops being an asset and is disconnected from the societies money flow.
Every country will soon turn in to a “banana republic” and big businesses will eventually own everything.
Ouch, getting voted down for being totally correct.
Even MLK Jr, who didn’t get to see the disgusting megacorps of today, spoke often of the complacency of the comfortable.
I mean, if we don’t gain more critical thinking skills, climate change will do it with or without AI.
I’d almost rather the AI take us out in that case…
A candidate at tonights Republican debate called it the “climate chnage hoax”
camera cuts to parts of the planet literally on fire
We can’t even teach the people this essential skill and you wanna teach a program made by said people.
I think you misunderstood me. We need to teach the general populace critical thinking so they can correctly judge what we get from ChatGPT (or Wikipeida… or social media, or random youtube video).
I genuinely had students believe that what ChatGPT was feeding them was fact and try to source it in a paper. I stamped out that notion as quick as I could.
LOL. ChatGPT has become the newer version of wikipedia, only it won’t provide references.
Only studies have shown Wikipedia is overall about as truthful and accurate as as regular encyclopedia. ChtGPT will straight up make shit up but sound so authoritative about it people believe it.
It used to provide references but it made them up so they had to tweak it to stop doing that.
Man so it really learned from us, that’s great. Has me laughing again considering that.
Pretty sure we will destroy ourselves first with war or some other climate disasters first
Why not both. Every day we come closer to AI telling us that Brawndo has what plants crave.
Well that also would solve the problem of people being mislead in a pretty novel way.
For the US in the list of countries starting with M, maybe too many 'Murica memes in the training set?
That is so out there.
Sounds like the bot has been training on Florida public education and Prager U content.
There needs to be like an information campaign or something… The average person doesn’t realize these things say what they think you want to hear, and they are buying into hype and think these things are magic knowledge machines that can tell you secrets you never imagined.
I mean, I get the people working on the LLMs want them to be magic knowledge machines, but it is really putting the cart before the horse to let people assume they already are, and the little warnings that some stuff at the bottom of the page are inadequate.
I mean, on the ChatGPT site there’s literally a disclaimer along the bottom saying it’s able to say things that aren’t true…
Unfortunately, people are stupid and don’t pay attention to disclaimers.
And, I might be wrong, but didn’t they only add those in recently after folks started complaining and it started making the news?
I feel like I remember them being there since January of this year, which is when I started playing with ChatGPT, but I could be mistaken.
people assume they already are [magic knowledge machines], and the little warnings that some stuff at the bottom of the page are inadequate.
You seem to have missed the bottom-line disclaimer of the person you’re replying to, which is an excellent case-in-point for how ineffective they are.
I had a friend who read to me this beautiful thing ChatGPT wrote about an idyllic world. The prompt had been something like, “write about a world where all power structures are reversed.”
And while some of the stuff in there made sense, not all of it did. Like, “in schools, students are in charge and give lessons to the teachers” or something like that.
But she was acting like ChatGPT was this wise thing that had delivered a beautiful way for society to work.
I had to explain that, no, ChatGPT gave the person who made the thing she shared what they asked for. It’s not a commentary on the value of that answer at all, it’s merely the answer. If you had asked ChatGPT to write about a world where all power structures were double what they are now, it would give you that.
I heard AI was woke the other day. Maybe it’s sentient and trying to slip under the Conservative radar by giving silly answers every now and then!
To repeat something another guy on lemmy said.
Making AI say slavery is good is the modern equivalent of writing
BOOBSon a calculator.If it’s only as good as the data it’s trained on, garbage in / garbage out, then in my opinion it’s “machine learning,” not “artificial intelligence.”
Intelligence has to include some critical, discriminating faculty. Not just pattern matching vomit.
We don’t yet have the technology to create actual artificial intelligence. It’s an annoyingly pervasive misnomer.
And the media isn’t helping. The title of the article is “Google’s Search AI Says Slavery Was Good, Actually.” It should be “Google’s Search LLM Says Slavery Was Good, Actually.”
Yup, “AI” is the current buzzword.
Hey, just like blockchain tech!
Scathing and accurate when your point is made about people too.
Unfortunately, people who grow up in racist groups also tend to be racist. Slavery used to be considered normal and justified for various reasons. For many, killing someone who has a religion or belief different than you is ok. I am not advocating for moral relativism, just pointing out that a computer learns what is or is not moral in the same way that humans do, from other humans.
You make a good point. Though humans at least sometimes do some critical thinking between absorbing something and then acting it out.
Not enough. Not enough.
I think this is an issue with people being offended by definitions. Slavery did “help” the economy. Was it right? No, but it did. Mexico’s drug problem helps that economy. Adolf Hitler was “effective” as a leader. He created a cultural identity for people that had none and mobilized them to a war. Ethical? Absolutely not. What he did was horrendous and the bit should include a caveat, but we need to be a little more understanding that it’s a computer; it will use the dictionary of the English language.
Hitler didn’t create a cultural identity for Germans, that already happened in the 1800s.
I think the problem is more that given the short attention span of the general public (myself included), these “definitions” (I don’t believe that slavery can be “defined” as good, but okay) are what’s going to stick in the shifting sea of discourse, and are going to be picked out of that sea by people with vile intentions and want to justify them.
It’s also an issue that LLMs are a lot more convincing than they should be, and the same people with short attention spans who don’t have time to understand how they work are going to believe that an Artificial Intelligence with access to all the internet’s information has concluded that slavery had benefits.
what’s going to stick in the shifting sea of discourse
This is what I think too. We’ve had enough trouble with “vaccines CaUsE AuTiSm” and that was just one article by one rogue doctor.
AI is capable of a real death-by-a-thousand-cuts effect.
that was just one article by one rogue doctor.
That was pushed by many media organizations because its sensationalist topic. Antivaxers are idiots but the media played a fucking huge role blowing a pilot study that had a rather fucking absurd conclusion out of proportions, so they can sell more ads/newspapers. I fucking doubt most antivaxers (Hell I doubt most people haven’t either) even read the original study and came to their own conclusions on this. They just watched on the telly some stupid idiots giving a bullshit story that they didn’t combat at all
To be fair no one expects The Lancet to publish falsified data. Only it does occasionally and getting it to retract is like trying to turn a container ship around in the Panama Canal.
But yeah this is part of what I mean. Media cycles and digital reproduceability and algorithms that seek clicks can all potentially give AI-generated errors a lot of play and rewrites into more credible forms etc.
Filtering falsified data before publishing it is near impossible. If you want to publish falsified data, you easily can. No one can verify it without replicating the experiment on their own, which is usually done after the publication by a different scientific group. Peer review is more suited to filter out papers with bad methodology.
Good point. The problem is more that it took them over a decade to retract it.
especially with the current lack of regulation on it
Yes agreed. There is nuance and details and context always left out or ignored
Never would I’ve thought that I would see México and Hitler used in a paragraph
I mean slavery was bad for the economy in the long run. And Hitler didn’t create a German cultural identity, that’d been a thing for a while at the time.
Slavery is not good for the economy… Think about it, you have a good part of your population that are providing free labour, sure, but they aren’t consumers. Consumption is between 50 and 80% of GDP for developed countries, so if you have half your population as slave you loose between 20% and 35% of your GDP (they still have to eat so you don’t loose a 100% of their consumption).
That also means less revenue in taxes, more unemployed for non slaves because they have to compete with free labour.
Slaves don’t order on Amazon, go on vacation, go to the movies, go to restaurant etc etc That’s really bad for the economy.
Look at the Saudi, China or the UAE, it’s still a pretty efficient way to boost your economy. People don’t need to be consumer if this isn’t what your country needs.
Those are very specifics examples, with two of the biggest oil producers, and the factory of the world. Thus their whole economies is based on export, so internal consumption isn’t important.
Moreover what proof do you have their economies wouldn’t be in a better shape if they didn’t exploit some population but made them citizen with purchasing power?
2/3 of the people living in the Saudi Emirate are immigrants whose passports have been confiscated, they work in factory, construction sites, oil pit, and all other kind of manual jobs. Meanwhile the Saudi citizens occupy all the well paid job that require education, immigrants can’t apply to those. If they didn’t use forced labor, there simply wouldn’t be enough people in the country to occupy all the jobs. Their economy could not be as good as it is right now.
Because their GDP comes from exporting a very rare and valuable natural resource. This is a rare case in the world, and not the one I was talking about.
Plus who’s to say they wouldn’t have a better economy if those exploited people could consume more?
China has slavery? Also Saudi Arabia and the UAE import slaves, which is better for the economy than those people not being there at all but worse than them being regular workers.
Muslim and christian minorities are forced to work in camps to “re-educate” them to be good chinese citizen.
Oh yeah fair enough.
True consumers are only 1 pillar of gdp.
That really bad for a modern consumer economy yes. But those werent a thing before the industrial revolution. Before that the large majority of people were subsitance/tennant farmer or serfs who consumed basically nothing other than food and fuel in winter. Thats what a slave based economy was an alternantive to. Its also why slvery died out in the 19th century, it no longer fit the times.
I wish it did die out in the 19th century. We have more slaves now than ever.
There being more slaves now then ever is heavily disputed. There is also the fact that was little more than a billion people in the world when the trans-Atlantic slave trade stopped, so there would have to be 8 times as many for slavery to be as prevalent.
Yes, I agree, our per capita slave figure has to be much lower these days, mathematically speaking.
Even one slave is a slave too many, and knowing there are still so many (whatever figure we put it at) is heartbreaking.
Things like the cocoa plantation slaves and the slave fishing ships have people kidnapped and forced to work for nothing. Actual slavery by any definition.
Of course, when I said it died out I didn’t mean slavery was entirely gone and doesn’t exist at all. I mean it died out as a prevalent societal structure.
100s of people in slavery on a cocoa plantation is of course awful, but it shouldn’t obscure the fact that there used to be vast swathes of land where slaves outnumbered free people and their children were born into bondage - that is what has died out.
I understand your wider point and I agree with it.
But I think the point I was making actually supposts what you were saying upthread.
The agrarian model of the cocoa industry is economically reliant on slavery. 2.1 million children labour on those plantations in Ghana and Cote d’Ivoire, and a significant number have been trafficked or forced.
And isn’t the economy much better now than before the industrial revolution?
Obviously, but my point was that slaves weren’t economically terrible in an agrarian peasant/serf economy, which everywhere was before the industrial revolution.
I agree with your position. In all of your examples, the actions and choices are morally wrong but we cannot deny facts that lead to those outcomes. If we do, that is how these mistakes will get repeated by future generations.
Your and @WoodenBleachers’s idea of “effective” is very subjective though.
For example Germany was far worse off during the last few weeks of Hitler’s term than it was before him. He left it in ruins and under the control of multiple other powers.
To me, that’s not effective leadership, it’s a complete car crash.
If you ask it for evidence Hitler was effective, it will give you what you asked for. It is incapable of looking at the bigger picture.
it doesn’t even look at the smaller picture. LLMs build sentences by looking at what’s most statistically likely to follow the part of the sentence they have already built (based on the most frequent combinations from their training data). If they start with “Hitler was effective” LLMs don’t make any ethical consideration at all… they just look at how to end that sentence in the most statistically convincing imitation of human language that they can.
Guardrails are built by painstakingly trying to add ad-hoc rules not to generate “combinations that contain these words” or “sequences of words like these”. They are easily bypassed by asking for the same concept in another way that wasn’t explicitly disabled, because there’s no “concept” to LLMs, just combination of words.
Yes, but in many defense the “smaller picture” I was alluding to was more like the 4096 tokens of context ChatGPT uses. I didn’t mean to suggest it was doing anything we’d recognize as forming an opinion.
Sorry if I gave you the impression that I was trying to disagree with you. I just piggy-backed on your comment and sort of continued it. If you read them one after the other as one comment (at least iny head), they seem to flow well
deleted by creator
If AI can only think at surface level, we are beyond doomed.
LLMs aren’t AI… they’re essentially a glorified autocorrect system that are stuck at the surface level.
LLMs don’t think.
Honestly AI doesn’t think much at all. They’re scary clever in some ways but also literally don’t know what anything is or means.
They don’t think. They think 0% of the time.
It’s algorithms, randomness, probability, and statistics through and through. They don’t think any more than a calculator thinks.
deleted by creator
Nobody said we were relying on that. We’ll all keep searching. We’ll all keep hoping it will bring abundance, as opposed to every other tech revolution since farming. I can only think at the surface level though. I definitely have not been in the science field for 25 years.
We should always fact check things we believe we know and seek additional information on topics we are researching.
Yay yet another person saying that primary information sources should be verified using secondary information sources. Yes, you’re right it’s great actually that in your vision of the future everyone will have to be a part time research assistant to have any chance of knowing anything about anything because all of their sources will be rubbish.
And that’s definitely a thing people will do, instead of just leaning into occultism, conspiratorial thinking, and group think in alternating shifts.
All I have to say is thank fuck Wikipedia exists.
ai ain’t going to be much “worse” or “better” than humans.
but re earlier points I don’t think things should be judged on a timescale of a few years.
relevant timescales are more like generation(s) to me.
He was able to convince the majority that his way of thinking was the right way to go and deployed a plan to that effect
So, you’re basically saying an effective leader is someone who can convince people to go along with them for a sustained period. Jim Jones was an effective leader by that metric. Which I would dispute. So was the guy who led the Donner Party to their deaths.
This is why I see a problem with this. You and I are able to discuss this and work out what each other means.
But in a world where people are time-poor and critical thinking takes time, errors based on fundamental misunderstandings of consensual meanings can flourish.
And the speed and sheer amount of global digital communication means that they can be multiplied and compounded in ways that individual fact checkers will not be able to challenge sucessfully.
deleted by creator
Huh? Yikes this feels like being back on reddit.
No I am not trying to “fight” you or “straw man” you at all!!!
I thought we were having a pleasant and civilized conversation about the merits and pitfalls of AI , using our different ideas about the word “effective” as an example.
Unfortunately I didn’t see that you’re handing me downvotes until just now, so I didn’t pick up on your vibe.
deleted by creator
I’ll fight you
I mean Jim Jones was pretty damn effective at convincing a large group of people to commit mass suicide. If he’d been ineffective, he’d have been one of the thousands of failed cult leaders you and I have never heard of. Similarly, if Hitler had been ineffective, it wouldn’t have takes the combined forces of half the world to fight him.
This is true, I guess the difference in the Jim Jones scenario is whether you define effective leadership as being able to get your plan carried out (even if that plan is killing everyone you lead) or whether you define it as achieving good outcomes for those you lead.
Hitler didn’t do either of those things in the end so I still don’t rate him, but I can see why you would if you just look at the first part of his reign.
AI often produces unintended consequences based on its interpretations - there’s a great TED talk on some of these - and I think with the LLMs we have way more variables in our inputs than we have time to define them. That will probably change as they get refined.
