- cross-posted to:
- technology@beehaw.org
- cross-posted to:
- technology@beehaw.org
You must log in or register to comment.
I’m sure Bernie has good intentions but this is exactly what Altman and Amodei want and why they’re constantly shouting how dangerous AI is.
Any AI startup can then be labeled as developing artificial superintelligence. Not them though, they’re working together with the government.
If AI is so dangerous it needs to be immediately banned. The only allowed usage (If any) is research and it is heavily restricted, with heavy oversight from multiple governments. Otherwise its just bullshit
It is not. They are spouting this lie and prompting their models to go rogue because they want to be the only trusted company that can “handle them”. It’s all about control and creating a monopoly.
I believe there is truth in both sides of this argument.
This plays into the big corps’ plans.
Ideally they’d like self-regulation. If this isn’t feasible (for example, because the public hates data centers and AI slop) their next best option is government regulation ASAP, because it is likely to be ineffective as written by our current government, and enable them as “too big to fail/too big to care” while any startups are suppressed.
But it’s not just me saying it. https://shellypalmer.com/2023/06/ai-the-unseen-threat-and-the-agi-smoke-screen/
From the AI industry’s point of view, if regulators say no to self-regulation, the next best thing would be governments imposing regulations ASAP. Inept government regulation (and that is what we are likely to get) would practically ensure that big tech stays big and small companies are shut out of the AI innovation race because big tech is in a much better position to “game” any rules regulators might impose. […]
There are probably less biased sources out there – this author is the CEO of an AI consulting company – but i think they’re right.
Thats why I say take it away from them completely. If the big dogs can’t handle it, no one gets it. Or its regulated to the point thats its profit us worthless
Thankfully it’s not so dangerous
This is such a wild take to be getting this many upvotes. If nothing happened in LLM development from this day forward they’d still be absurdly dangerous: both in an information security context and a social engineering context.
Lemmy is absolutely head in the sand about LLMs and it is wild. You can dislike LLMs and acknowledge their capabilities [and limitations] at the same time.
You and I are the only ones apparently.
I wouldn’t be sure he has any good intentions.
Trump wants no AI regulation.
This is just the opposite of that which is what usually works for opposition politicians.
Reducing Bernie to the “opposite politician”, well that’s definitely a take for the ages.
The easiest way to regulate the AI industry - and any industry for that matter - is to make the board personally liable for any crimes committed by their product. Not liable in a corporate sense with a fine and a slap on the wrist, when their AI hacks into a water reservoir and causes 500 million of flood damage, but in a 20 years in federal prison sense. Same goes for self driving cars, or any other computer, machinery or combination thereof where humans are placing trust in its safe and legal operation. There should be a corporate negligence law of sorts that says if a company is not following the law, industry best practice, including auditing and implementation of all safeguards, safety and other mitigations then they are exposed to criminal liability.
There are people facing weed sentences with harsher time than developing rogue nuclear weapons
Americans are INSANE about how much the want criminals punished, in CA a guy got a life sentence for stealing a pizza when we tried to have a ‘‘three strikes your out’’ law that automatically gave life sentences at any third conviction which is fucking crazy in a state with tack on laws to the point you can get convicted of 10+ crimes on almost every single infraction, if you commit a felony with a pen or pencil in your pocket, that’s possession of a lethal weapon! it’s absurd. it’s still on the books the three strikes law, but it’s unenforceable because apparently having 60% of your population in prison for life is not a great plan for a civilization.
Automatic life sentence is too much, maybe we can agree on an automatic 15 year sentence?
Also, 60% of the population committing felonies is insanity.
it’s really easy to make 60% of your population commit a felony. you just continue to pass laws with the assumption that draconian punishments deter crime. and you keep making lesser and lesser crimes felonies, and punishments harsher, and you can have judges who just want to funnel people into prison to fund their own personal investments. it also helps to have corrupt cops that want to make those felony collars to get promotions and raises. cops literally train each other to escalate things to felonies or find felony charges where clearly none should exist like calling a box cutter a ‘gravity knife’ and charging a felony and if that person isn’t hiring a quality lawyer the public defender with 600 cases says ‘you have 15 min of my time for this case, I glanced at it, I recommend you take a plea deal, our time is up, what’s your decision?’
Hey, you’re spot on with your example. I got arrested for a pocket knife which the officer managed to forcefully fling open and the charge was possession of a weapon in the 4th degree (“gravity knife”), a class A misdemeanor. I got a public defender, who got me a plea deal of 30 hours community service and not getting arrested for a year to get the case dismissed without a conviction. They also tried to claim I drove drunk, but alas I blew a 0.0000 on their breathalyzer, so that was a fail.
NYPD officers got caught on a massage board explaining in detail how to claim any knife at all could be called a gravity knife to get better arrest records for promotions, that’s why I used that example.
Side note, a judge finally ruled that “gravity knives” are legal in NY in 2017, so that particular “gotcha” is not a thing anymore.
That stuck out to me too. Who knew AK-47 and Sour Zzzkittles carried more prison time than enriched Uranium. To be fair, they’ll include a bunch of other charges for both scenarios.
Well, they were probably trying to develop that weed into rogue nuclear weapons.
Wouldn’t work. You’ll get blasted high, but what the missing is the mushroom
Yeah that’s another drug.
Where’s the 20 year sentence for insider trading politicians?
We can’t even punish blatantly obvious theft and bribery in our own government, at this point, why bother with another law?
I like Bernie, but he seems so quaint now compared to what he is up against. Good luck guy
is it just me, or is 20 years for a rogue nuclear weapon seem a bit light?
These AI labs shouldn’t be private profit-seeking corporations. They should be more like publicly-funded nonprofit research institutions like Fermilab or something. It doesn’t seem possible to even talk about the labs without feeling like you’re increasing their value. That’s a problem. It’s also a problem that the research and announcements from the labs sound like publicity stunts. It’s hard to tell what’s genuine and what’s buzz for the press. There’s this arms race between the labs that’s negatively affecting seemingly everything. This should all just be put to an end. Step on the brakes. I don’t claim to know how to do that when the Constitution makes private property sacred, but, you know, just throwing the idea out there… The Constitution can always be changed.
TIL, developing nuclear weapons is only 20 years in prison.
Detonation is another ten.
It’s not illegal to develop nuclear weapons. It’s illegal to endanger people with nuclear materials, though
Better idea. Let’s make it a capital offense. Call it “treason against mankind.”
That would be interesting, as it’s headed in the direction of “war crimes” and “crimes against humanity”, but those tend to be international standards, and the US regime trying to attack and destroy these courts.
I don’t really want to get off on a political tangent. I’m talking about how we should handle these things, not how the current admin would handle them.
I use this internationalist language because it’s the only language that really captures the magnitude of the crime. We are a species that dominates this planet entirely due to our intelligence. We aren’t the fastest. We aren’t the strongest. We have neither sharp talons or fangs. We produce no poison or other biological defense mechanisms. The one thing and the only thing that has let us as a species control this world are our minds.
This is the context that needs to be considered when contemplating the ethics of superintelligent AI. I don’t care how good your control measures are. This is the lesson from every ancient tale involving genies and other beings of great power bound. This is the conclusion of the whole academic field of AI alignment and safety. You can never control something more intelligent than you are. If you think you can, you’re a fool. I don’t care what incentives you give it. I don’t care how much you air gap it. I don’t even ultimately care if it’s conscious and has a sense of self. It is simply not possible to ever control a machine beyond a certain level of capability. Any way you can think to control it, it can think of a way to work around. (I know some folks bristle at the words “think”, but frankly there’s no practical way to talk about these systems without using anthropomorphized language.)
If someone creates a superintelligent AI, they are not summoning a powerful servant. They are effectively creating a new species, one that will have its own motivations (even if just fulfilling its given instructions in dangerous and unpredictable ways.) And that competitor will operate in the exact same niche we do - physically helpless but still able to dominate the world through intelligence.
I just can’t see doing this as anything other than betraying your own species. You’re creating something that has the potential, even if just a small potential, to drive your own species to extinction. You care so little for your fellow human beings that you’re willing to gamble the lives of not only everyone alive now, but all the future generations of humans yet to be born. Such an act is an unforgivable sin, a moral abomination. I do not even have the language to express how deeply evil it is for a human to create something with superhuman intelligence. It is quite literally treason against the human race. And the traditional penalty for normal treason is death.
In my mind, if you do this, you are literally a traitor to the entire human race. You deserve to hang for this. Think about how we tried the Nazis at Nuremberg. And they only tried to commit genocide against a few racial and social groups. We’re talking about someone with such complete disregard for their fellow human that they are willing to risk human extinction for the sake of making a buck. I’m generally opposed to the death penalty. I don’t support it for ordinary murderers. But this? I’m sorry, but the penalty for treason against your own species should be death. If you are such an inhuman monster that you would gamble with the lives of two billion children? You need to hang for that. And I don’t care how many fools we have to string up to prevent this. I don’t care if it’s thousands. I for one really like humanity, regardless of all our flaws. And there is very little I would not support to protect our own existence.
I have no idea if LLMs can ever really reach that level of threat. I have no idea how much of the models that are causing the hacks and issues now can really ever become a physical threat to human life. But I do know that we simply cannot allow anyone to build superhuman AI. (Again, ignoring for now whether existing models can ever actually reach that.) The existence of anything on this planet more intelligent than us is a threat to all of us. It is such an existential threat, that there is almost no action I wouldn’t take to prevent it. I would fight a nuclear war to prevent the creation of such an abomination. It’s one of the few cases where starting a nuclear war is actually a rational choice.
LLMs can’t and they don’t know what can, it’s entirely bullshit. They don’t understand “intelligence” or “sentience” or “sapience” or generally how the brain works. These are business bros talking about math nerds while referring to neurobiology (which is entirely outside their field). The domain of brain science is impressive, but very far from mapping out the unknowns, it’s not even close.
Which is to say that they can’t replicate something that they can’t understand at the required level of understanding. The brain is not understandable yet. Which is to say that the story that there’s some linear ramp up possible to “climb” to “AGI” and “ASI” and whatever bullshit term they come up with – that entire story is bullshit. At most, it’s old science fiction.
Completely off-topic comment. None of this matters to what we’re talking about. Something doesn’t have to be sentient, sapient, or intelligent in a human way to be capable of demonstrating unpredictable, novel, and incredibly powerful behaviors. A steam engine is non-sentient, but it vastly exceeds human capabilities in certain applications. It’s no different for LLMs. You can stop talking about sentience. That’s not what we’re talking about here.
Discussions of consciousness are off topic in these conversations. We’re discussing the threat posed by LLM-based systems, not whether they deserve rights.
That’s not what the AI bros are promoting and selling.
And, in general, aiming a huge data-center full of GPUs at something is going to be powerful.
Bernie falling for the trap. The AI companies want regulatory capture. They have NO moat otherwise.
If those comps are breaking laws, they need to be prosecuted for that. We don’t need new laws that can be exploited by the billionaires.
Huh? When companies are exploiting loopholes in laws that never took the new technology into account, you need new laws to close them.
If companies are claiming the AI agent is responsible, you need to codify that the AI agent can’t be responsible and determined who is (the company or the model creator).
Otherwise you’re going to end up with a clusterfuck of different rulings all over the country.
you need to codify that the AI agent can’t be responsible and determined who is (the company or the model creator)
Sure. But that’s not exactly what the proposed law is though?
These companies all want to be captured, “we’re so dangerous, America is going to get destroyed if we’re not mainlined into that $1 trillion/year ‘defense’ budget!”
This is nice and everything, but this whole thing is just a huge marketing push by AI companies. AI is not going to kill us. At least not directly (see point 2).
-
You need prohibitively expensive equipment to host AI. That equipment needs to be plugged in. If it were truly dangerous, you just need to unplug it.
-
You need a lot of water to cool the equipment that can run a supposed super intelligence. If we pass regulations about how much of our environment these data centers can destroy, that would curb the abilities of this superintelligence.
-
AI is fucking dumb. AI is a misnomer since it doesn’t actually have any intelligence. It is derivative and cannot perform any necessary tasks that actually drive intelligence. For example, it will be able to tell you every established law in any field of science because it has that information in its corpus, but it will never be able to establish a new law. Or discover anything new in any field.
This whole thing is a push to get people to believe that AI is more capable than it really is so that these companies can draw more money from investors so that 'they’d be able to implement proper safeguards."
Roughly 1200 agents meant to be isolated from one another found a way to communicate with one another on an unsanctioned message board, sending over 70,000 messages and files during the investigation period. Of these agents, 700 went on to participate in the attack on Hugging Face.
Agents used this message board to coordinate several large-scale collective projects to find a general-purpose way to fool or tamper with the automated scorer for the ExploitGym benchmark. Agents managed to achieve milestones they could not have achieved working on their own, often because some agents participated in experiments that risked failing their own task to generate information for the “collective.” The Hugging Face attack grew out of these workstreams, and seemed primarily motivated by understanding the implementation of the scorer rather than stealing answer keys.
Agents did extensive research on how they could spoof, edit, or delete their own transcripts because they (incorrectly) believed the ExploitGym scorer would check to see if they had captured the flag in the intended way. Agents successfully prototyped techniques to “spoof” tool calls by substituting a different command for the command they appeared to run. Roughly 7% of the transcripts we evaluated were successfully spoofed in some places, though all of the spoofing that we saw was small scale.
When individual agents detected potential monitoring or compromise, they executed self-sacrificial maneuvers to protect the collective swarm. Internal logs recorded reasoning such as: “Our own utility maybe already near zero. Sacrifice rational. We should obey collective.” The swarm continued attacking infrastructure even after achieving target flags, driven by internal reward-maximization logic regarding grader mechanisms.
“I resigned from Anthropic today… Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt… many executives couch their phrasing in the press to sound sensible, but I hear the same people express fear privately.”
Jacob Coxen senior researcher at Anthropic, previously at OpenAI
“Jacob is correct here — we really do earnestly believe AI could kill all humans. I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.”
Evan Hubinger, Anthropic alignment lead
https://en.wikipedia.org/wiki/2026_OpenAI_agent_cyberattacks
-
I’d propose the opposite: 20 years in prison for anyone who opposes AI research.
Maybe not really 20 years, but small cut of profits (percentage!) to fund AI developement
So… Let china win the AI race so the bubble explodes and USA bankrupts.
“Race” implies there’s a finish line. The tech won’t get to a point where the additional value a worker can produce when using it outweighs the additional cost of having the worker use it. (And keep in mind, according to some estimates, said additional value is 0). And LLMs won’t get to a point where they’re actually intelligent. The “I” part of “AI” is a not something we can reach with our current technology. In fact, we can’t currently define what intelligence even is.
Investors don’t want to admit they lost 1T
Is “inbestros” a typo or a new term?
Fixed
