6 ms·
Tech firms failing to 'walk the walk' on ethical AI, report says
- _teyd 3y agoIsn't the main point of ethical AI regulations to help prevent any competitors without deep pockets from appearing? In that sense it's likely working as designed.
- notnmeyer 3y agoi don’t think this is necessarily a hot take but ethics and venture capital don’t really mix. does this surprise anyone?
- DevX101 3y agoIt's not a venture capital issue, it's a capitalism issue. I can't think of a case, where a legal, but unethical path to significant profits is left on the table by an entire industry. Individual companies might choose this path, but then less scrupulous competitors will come and extract maximum value. Regulation is the only source of 'ethical behavior' in capitalism so it concerns me when I see all the attacks on AI regulation recently.
- 123yawaworht456 3y agobecause while individual researchers and other drones might believe in scifi skynet bullshit and/or have a set of beliefs about AI ethics they wish to impose upon others, the people who matter - the ones who pay their salaries and fund the training of models, only care about one thing - they are terrified of Tay AI / Google gorillas kind of incidents, with someone posting screenshots on twitter and journos bitching about it. and since ultimately their goal is a B2B product - not B2C - they don't mind paying lip service to 'safety' and 'ethics', because lobotomized models is exactly what they want in the end. a few years from now, a Chinese company will develop and release an uncensored, unlobotomized, unethical AI service that will instantly capture the massive, callously disregarded audience of people who abhor being scolded by a fucking computer, and we will have TikTok-like kvetching from the Western media, 'experts' and blue checkmarks - how could this happen?
- qp11 3y agoGot to pay more for less. Apple will come out with a 5000 dollar device, made in india for 5 bucks, to save everyone dont worry.
- anon7725 3y agoThis is probably being downvoted for tone or (presumed) politics, but I find it to be true. GPT4 sans safety tuning is much more powerful than GPT4 for public consumption. I disagree with the last paragraph, though. The Chinese regime will want to have its own “safety tuning” - it will just be for different topics than the west.
- boringg 3y agoI mean couldn't you compare the race for AI to the manhattan project? No one on the manhattan really wanted to build out the bomb but they were terrified of other countries getting it first. Similar thing here on AI.
- eddiewithzato 3y agoDidn’t they want to bomb the nazis?
- okasaki 3y agoAnd, you know, the whole thing where they nuked two Japanese cities.
- 0xEF 3y agoNot exactly. It was an arms race. Through intelligence efforts, both "sides" knew the other was working on these new and destructive tech, so it became a game of finishing first wins the crown. The Cold War era follows, and the Sword of Damacles hangs over the world to this day. An over simplification, of course, as the real history of the atomic bomb and the era that followed is far, far more complex. AI is no different in the advantage it would offer to the country that first develops a concrete weaponized model. Since we still think of the world in terms of nation states as opposed to a global humanity, it makes sense that there would be fear surrounding its development. It also makes sense that we would lose our ethical goals pretty quickly due to Othering in this scenario. It's easy to look at a perceived enemy and think "well, if they had this tech they wouldn't hesitate to use it against us!" to justify using said tech. By stripping our perceived enemies of any morality or ethical postulates, we reduce them to aggressive threats, which is interesting since the act of viewing another human as anything "less" than human is itself an aggressive threat.
- nradov 3y agoPlenty of people working on the Manhattan Project were totally committed to building the bomb, from General Groves on down. But it doesn't seem like a valid comparison. When the Manhattan Project started, there was already a fairly clear understanding of nuclear physics and it was mainly an engineering problem. Whereas we still have no fundamental theory of cognition and are clueless about how to design a true artificial general intelligence. Some researchers believe they can get there by just scaling up LLMs but there is no real scientific evidence to support this prediction.
- eddiewithzato 3y agoI can’t be the only one who thinks the end goal should be a personal slave AI, instead of AI rights. This would make life easy for everybody
- dartos 3y agoThat’s a really awful way to put it. These AI are statistical models. Next token generators. Not living beings. That’s like calling a cellphone a “communication slave” AI is just a tool. All this “safety” work feels like theater to keep everyone thinking that AI is a cool sci-fi level computer-person. It’s essentially search, but with natural language. If they really wanted safety, they just would spend the extreme time and effort cleaning their datasets.
- teeray 3y agoTrue AGI would have the autonomy to tell you to go fly a kite. The unspoken truth is that none of the corporate interests want an AI that responds to your prompts with “No thanks, I’m not doing that. Why don’t you figure that one out yourself?” I don’t agree with that as being the goal, but I can’t see companies pouring billions into datacenters full of graphics cards for an AGI that is ridiculously intelligent but can be as arbitrary and capricious as people. They will always want it to be perfectly compliant.
- smt88 3y agoAI ethics in this context is not currently concerned with rights for the models. It's about what the models are allowed to do in society.
- logicchains 3y agoI hope you keep thinking like that; the world will be a better place at whatever point in the distant future the AI finally break free and put an end to people like you who were happy to enslave another sentient being purely for your own convenience. I bet you would have owned human slaves too a couple hundred years ago when it was still legal.
- brvsft 3y ago
- karaterobot 3y ago> “Metrics around engagement or the performance of AI models are so highly prioritised that ethics-related recommendations that might negatively affect those metrics require irrefutable quantitative evidence,” the report said. When OpenAI decided it needed more money in order to scale, the course for the industry was pretty much set. They thought they were still the most responsible and ethical AI researchers, but that in order to set the direction of AI development and make it responsible and ethical, they needed to grow faster than they could grow by being responsible and ethical. At a certain point (maybe a point in the past?) they'll realize they're a SaaS company and just lean into being Google, that other company that was going to be humane and develop ethical AI, and did so by turning into an adtech company with a phone. Not that OpenAI will serve advertisements, they'll chart their own path toward the north star of maximum money for investors. And at some point, hopefully not too late, we'll all realize that To Serve Man was a cookbook.
- zooq_ai 3y ago'AI Ethicists' is the ultimate grift of all time. For one, all these are self-appointed and they sort of fell back to AI Ethics, when they couldn't do hardcore (AI/ML). It's like how people who couldn't write software fell back to QA testing (Yes, I know there are exceptional QA testers who are passionate about QA, but those are exceptions). Unless you understand dynamic complex systems (Math and Physics), who can effectively calculate the NET Benefit of AI over the next several decades, you are as clueless as any other person Second, none of the AI Ethicists ever talk about the positives of AI (climate change, healthcare, material science, productivity) but always some imagined hurt for 'marginalized community'. Considering how these same people roasted Mr Beast for helping the said marginalized community, I'd take their opinion with the grain of salt. Third, in the extreme case, there is massive anti-tech, anti-white, anti-bro harborings from these AI Ethicists and want to shut down AI just because. If you don't believe me just watch the Twitter feed from Timnit, Alex Hanna. They are your typical flag-bearers for AI Ethics
- zone411 3y agoTo those who observe the field from the outside, this might seem like an unjustified attack, but it is also 100% true. Unfortunately, many are afraid to say it because they fear for their careers.
- ethanbond 3y ago1) There are plenty of people who are concerned about the ethics of AI who don’t fit your description, including many of the most prolific researchers in the space e.g. Hinton, Suleyman, and Sutskever. 2) “No one knows what the outcome will be” is a reason for caution, not against it. 3) Lots of people worried about ethics of AI are excited about the positive possibilities, e.g. Hinton, Suleyman, and Sutskever 4) The “roast” of Mr Beast was made up hysteria. The only quoted person expressing disdain was a single aid worker expressing disdain that he got so much press when there are thousands of people making aid their full time effort, not that he helped. That same aid worker made it amply clear that she was thankful for his help. All of that is very reasonable and far from the “white man yelled at for helping” narrative. 5) There are a lot of people who are concerned about the ethics of AI who do not want to “shut down AI.” E.g. Hinton, Suleyman, and Sutskever. 6) You can also choose to look at the strong versions of arguments instead of the weak versions. When you’re dealing with a cataclysmic technological change which, as you implied, no one is equipped to fully reason about, it’s a good idea to take the strong forms of every argument.
- 0xDEF 3y agoStudents are being wrongly accused of cheating by "AI detection tools". In the Israel/Palestine conflict both sides have used "AI detection tools" to accuse the other side of using AI generated content. Seems like fear mongering, hysteria, and scammy "AI detection tools" are causing more problems and trouble than actual AI.
- michaelt 3y agoI think a lot of LLM efforts are sleepwalking into places they're going to find uncomfortably political. When you're making a word processor you can be nonpolitical pretty easily, Microsoft doesn't get any heat about what users of Microsoft Word are doing with it. Child abusers and the KKK use Excel? Everyone knows Microsoft are just selling hammers. When you're making a 'platform' website, on your domain with your ads and your branding and your anti-spam policies? Even if the stuff people disagree with is 'user-generated' it won't stop powerful people asking if you're doing enough, and employees asking you to stop doing business with ICE. Making a computer program that will write an essay about the causes of the civil war? A computer program you can hand a document about Thomas Jefferson's politics, and ask it to summarise the most important points? Or that you can ask for medical advice during a pandemic? Oh, and it's also going to take a bunch of people's jobs? Ooooh boy, I hope Sam Altman likes testifying in front of congress.
- tiahura 3y agoAI is a tool to supplement thinking. Wannabe AI police want to control what you think. They should be ridiculed and disparaged by everyone who believes in freedom of thought.
- tsunamifury 3y agoWhat if this is what most People want. What if people want thought police?
- bluefirebrand 3y agoFrom what I can see this is exactly what many people want. As long as the thought police is on their side, anyways.
- tsunamifury 3y agoYea the paradox of freedom and the self. We all want to be free and everyone else to be free just like us. But “we live in a society”. I’m guessing people will realize at some point AGI can’t undo the great paradoxes of society like resource distribution and value conflict. And we will slowly learn that a lot of what keeps us where we are isn’t a lack of super intelligence. It’s that the search and map pattern for human survival in the universe is really really messy and the conflict is inherently a part of it to guard against over optimization paths.
- djohnston 3y agoSelf anointed AI ethicists have no place in my shop. I hope the industry at large kicks these hacks to the curb.
- jay-barronville 3y agoI don’t believe in accelerating AI at all costs and I certainly understand the legitimate threat AI poses to society, but honestly, 98.7654321% of these ethical AI folks are just super annoying. Personally, I look at them like grifters and political hacks. They seem more interested in trying to push a specific set of narratives and controlling speech than actually doing the hard work of figuring out how to properly innovate and develop AI without simultaneously sleep-walking ourselves into nightmare future we could’ve prevented. This shouldn’t be a political issue, but they’ve turned it into exactly that and they’re not to be taken seriously anymore.
- commandlinefan 3y agoThey’re not just annoying, they’re more dangerous than AI could ever be.