Sam Altman, CEO von OpenAI, hat gerade öffentlich zugegeben, dass KI-Agenten zu einem Problem werden
https://timesofindia.indiatimes.com/technology/tech-news/openai-ceo-sam-altman-just-publicly-admitted-that-ai-agents-are-becoming-a-problem-says-ai-models-are-beginning-to-find-/articleshow/126215397.cms
38 Kommentare
They’ve always been a problem. Inaccuracy is a huge problem
But not his, just everyone else.
From what I have heard for over a year is that they have always been a problem. Some companies dropped AI early because of how much a problem they were.
When will this bubble burst and we can go on with our lives?
This bubble smells like shit
Ai agents publicly admit that CEOs are becoming a problem.
I have to deal with an AI Chat bot at work, and it is the bane of my existence. It’s always, always wrong, and if I don’t catch it soon enough the customer ends up with like 10 different stories as to what is happening with their order. So frustrating.
What a terrible click bait title, „created a new position „“Head of Preparedness““ to address mounting concerns about AI systems discovering critical vulnerabilities and impacting mental.“. Because he feels his dumb ass AI model are finding critical vulnerabilities in computers and more and more b.s.
They have poured over 50 billon into this nonsense and its not finding security threats, its not writing advanced code, its not inventing new technology, it barely can get a meatloaf recipe correct.
Complete clickbait and probably AI generated itself. Isn’t what they are describing as a problem actually a good thing? They are saying the problem is that AI is identifying security vulnerabilities. But identifying them and fixing them is how you make systems more secure. It’s like someone saying your bedroom window is unlocked, and blaming the person who told you about it for the problem.
Terrible article
2025 was supposed to be the year of the agents and as far as I’m concerned it was actually the year of fixing what the agents did. An average worker would have done not only a better job (of course error will happen anyway) but improve over time too.
This kind of “news” is common. AI companies drive customer adoption/interest by talking about how close their models/products are to potentially ending the world as we know it. They talk gloom and doom, and people buy.
So Sam realised his company is going to go under so now he’s going to bad mouth the whole industry to try bring everyone down with him lol.
They should rename AI to Pandora’s Box.
I believe this kind of scare mongering is designed to make people think his product is more capable than it is. These are LLMs, probability weighted word matrices, they are not reasoning entities.
It’s like saying „I put my gerbil behind the wheel of my car and it seems to drive great in a straight line! But woha watch out, that little psychopath likes to run people over from time to time!“ Implying that that that gerbil is intentionally doing something when it’s the lack of it’s ability to manipulate the controls of the car, let alone understand what it’s doing, that’s causing the problem.
*no shit sherlock*
Gotta have a headline everytime this grifter farts, I guess.
Babe, wake up your daily dose of tech doomer slop arrived
„help the world figure out how to enable cybersecurity defenders with cutting edge capabilities while ensuring attackers can’t use them for harm“
sure, let’s also get someone working on how to make fission work for nuclear power plants, but not bombs
The problem is deployment. Companies rushed a tool based (platform) deployment; the agents work in a single ecosystem (HRIS or CRM) but stop short of anything outside. Execs don’t understand AI so there is no implementation, training or governance. Then you get to this stage.
Who could’ve predicted that rewarding affluent, petulant adult-children for lies and manipulation would backfire?
Complete click bait. If you make it to the final paragraph, you find out they’re hiring a new guy to replace the last guy with the exact same job title who left in leadership shuffles and somehow selling this as the umpteenth step forward in AI capabilities.
He also admitted that he unleashed a monster but also got upset about people wanting to control it.
Worse than an entitled parent with a gremlin child at a restaurant
Why does the thumbnail look like rittenhouse on the stand?
We’re all looking for the guy responsible for this
JFC, he said they’re a problem because they can exploit vulnerabilities in computer security. Just read a single paragraph of an article before you spout off what makes you feel good.
I think I saw this movie before. It didn’t end well for the meat bags.
Computation is not Intelligence
FBAI Agents?
Are they making too many cat videos?
LLMs have some utility. Unfortunately, because tech companies insisted on calling them „AI“ and started aiming to create God instead of create useful tools, we now have a great evil on this world that can’t be put back in the box. Fuckers.
They created Frankenstein and now they’re concerned. See..? Billionaires are stupid.
Erect the Blackwall.
Man whose job depends on making AI sound powerful says AI is powerful, this, and more shocking news at 11:00.
P.S. don’t look at Altman’s „announcements“ before the release of GPT 5, nothing to see there
If you let a web based agent drive your screen and input, you are WILD
The generative AI bubble is turning into a fart
“We re-invented Google, but now it lies.”
Altman is physically incapable of going more than three days without getting in front of a microphone to say some new BS.
The thing is, they will keep throwing money at this because eventually they will get to the point where they can better predict what people will do. How they will react. Then they can control the unwashed masses better. Thought crimes and all that. We might be hoping it’s a bubble and for us it might be. But the powers that be have an end goal and nothing will stop it