AI Doesn't Threaten Humanity. Its Owners Do | Hacker News Reader

AI Doesn't Threaten Humanity. Its Owners Do

scientificamerican.com

96 pointslaurex45 comments

The "Owners" will use it to get rich[er], at any cost, and we'll only realize too late the damage it did. Government regulation could stop it, but politicians will only use their position to get rich[er] instead.

The issue with "AI" and tech companies is they have incredible ability to influence, going against these companies might mean losing an election. It's a major conflict of interest when it comes to regulation of these "knowledge companies".

Good point, bad rant.

Right now, the two real AI issues are 1) improved surveillance in a broad sense, and 2) job elimination. Both relate to what the owners of the tech want.

The political problem is that few governments are willing to effectively regulate 1) or 2).

Not only improved surveillance, but vastly increased surveillance, as everyone rushes to capture as much data as they possibly can for training.

Job elimination is not a new thing. People like to bring up textile workers because of how the term Luddite has come to refer to anti-technological sentiment, but the change in jobs has been fairly consistent.

I presume that in a few years electric cars will result in significant job losses amongst those who produce internal combustion engines. Renewable energy will reduce the number of employed coal miners.

I agree that this should be a role for government to assist people in a changing world.

there is a quite clear well and obvious difference between job elimination and job transformation and job change. But the express purpose of creating textile machines was not job elimination, it was output increase, meanwhile AI is not at least in business cases presented as an output increaser.

AI is presented and designed as an employee replacer, plain and simple.

Textile machines were attractive to owners since they could employ less skilled people who could be replaced if they tried to organise and strike. This enabled the factory owners to take more of the profit and treat workers badly. It was the loss of high skilled jobs and the consolidation of wealth that drove the Luddite movement.

Car factories that make ICE vehicles can be switched to produce electric vehicles. The job losses in car plants you hear about are regional and caused by plants been moved to areas with cheaper workers and weaker labour law (Mexico for example).

Also less than 44 thousand people work in the coal mining industry in the US. Renewable didn't take those jobs, resource depletion, heavy machinery and fracking did.

> Car factories that make ICE vehicles can be switched to produce electric vehicles.

Although that doesn't seem to happen much. Most electric cars are produced in new plants. The notable exception is Ford's truck line, where the Ford Transit and the F-150 both come in electric and gas models.

> Job elimination is not a new thing.

No, but this time it's come for the chattering classes from the Ivy League, who are freaking out.

3) Diffusion of responsibility, continuing the trend of the past few decades. For example creating ever more layers of disempowered reps and rule-executing paperwork pushers, while cloistering away anyone who has any decision authority. Next it'll be LLMs running such interference.

This is the same semantic argument of "Guns don't hurt people". Of course, guns hurt people. A physical object can be harmful without a human being to give it intent. It's a straw man argument that delves into the meaning of words than actually addressing if something is actually harmful or not.

I think we can all agree that AI is itself a tool, that can be used for good or bad. However, we can certainly make a judgement call on it, as a whole, by examining the current state of how it's being used, and by looking at what kinds of human behaviors it amplifies.

Once you invent the gun you can't really stop them proliferating though, people with the means and determination will own them. Allowing one group of people to own them and not others is just asking for the group of owners to dominate and rule over the ones who do not own them.

Yeah, in the last few years I realized somewhere along the way that I stopped believing in laws. It's a strange feeling and hard to think about how I used to feel in retrospect; it's like I used to believe there was some force that existed throughout the country that should control our behavior, but at some point I realized that there just isn't anything there. It's a bit like losing religion I suppose.

Spirit or letter? Is it fair to assume you agree killing other human beings is still bad? Is there an alternative system you prefer?

Reading back my comment, I probably should have said I have lost faith in "the law". In this context, I mean the idea of gun control, but it's just generally a feeling that there really is no one in control of (American) society at this point.

Without getting into a whole thing about the context of killing other human beings and how that matters, I'll just say I know what you mean and say yeah, it's bad to kill people. I'm not talking about personal morality; I live by one that many people would probably find overly restrictive (I certainly do).

Great response, and you're definitely not alone there, that's why I ask! I too find that the intent of "the law" is noble, but it seems to collect hard edges in practice.

> I'm not talking about personal morality; I live by one that many people would probably find overly restrictive (I certainly do).

Well now I'm interested!

You might be interested in Bruce Schneier's take on trust and what prevents predatory behaviour. In his book Liars and Outliers he gives a taxonomy of 4 elements: law, reputation, morality, and security.

Reputation and law are similar in that they're imposed by a community, but differ in that law follows a formally codified code, and of course the consequences imposed tend to differ. Morality is the internal motivation not to wrong others. Security describes physical/electronic measures that cause the predatory behaviour not to be practical in the first place, or at least raise the bar.

Schneier's own summary: https://www.schneier.com/blog/archives/2012/03/liars_and_out...

A review, hosted on Schneier's site: https://www.schneier.com/news/archives/2012/12/liars-and-out...

You’re losing touch with reality if you actually don’t think laws modify people’s behaviors, at least in most of the developed world.

Reality and I have had a pretty tough relationship, I'll concede that. I think my problem is that I've often seen the other way around - that peoples' behaviors modify the laws. I'm sure quite a bit could be written on that about Donald Trump who is currently in a trial that everyone knows cannot enforce any consequences to him, but for me I think the real moment it came apart was during the pandemic when I was struggling through some bureaucracy to update my car registration. I had the thought that I couldn't believe that the people who refuse to put on a mask to get groceries or whatever actually go do the nonsense that I was trying to get through right then. In contrast, covering your face with a piece of paper seemed really trivial, and yet to so many it was an affront to their identity, and so the mask mandates were ultimately rendered toothless.

You’re definitely onto something there. I highly recommend the book The Metaphysical Club which follows a specific group of thinkers (the Pragmatists) who formed much of the foundation of American legal theory.

The TLDR: Every legal system is governed primarily by the “moral culture” and constructed to defend/enact/perpetuate that culture. When the moral culture changes enough, the laws will change (or the interpretation or application of the laws for the same de facto effect).

So the question for all of us is how to steer the moral culture in preferred directions, which is why I was calling out your initial comment! It’s not good for people to come to the belief that laws don’t matter — such a belief is self-fulfilling at certain scale.

This is literally empirically false. Guns haven’t proliferated across the world. There are countless examples of locales that decided to and succeeded in not having guns proliferate in their region.

What do you mean? Name 'countless locales' without guns. The only one that comes to mind are things like freight ships preferring water cannons over guns. There's also countless of examples where not having apropriate self defence is a death sentence to a culture, country or person living in a dangerous area

Not OP by there are 18 countries that have mostly unarmed police forces [1]. I live in one of them - Ireland.

[1] https://www.worldatlas.com/articles/countries-where-police-d...

Maybe the patrol cops, but your rulers still keep heavily armed SWAT teams ready to be deployed in case the patrol cops encounter real resistance.

Every country, including democracies, have deployed surveillance against their own people. Giving your rulers, and big companies, exclusive access to this technology is just asking for an authoritarian nightmare.

> Giving your populace easy, uncontrolled access to this technology is actually causing rampant shootings.

Notice how one is “just asking” for a hypothetical risk while the other is an everyday fact of life.

The relationship between availability of guns and the amount of gun violence is extremely well-established. I personally am sympathetic to the view that an armed population is a meaningful deterrent against tyranny, but it’s hard to compare a hypothetical to a growing number of deaths. And the odd fact that firearms’ primary political role in our society is to intimidate people while voting doesn’t help either.

>The relationship between availability of guns and the amount of gun violence is extremely well-established.

Its also hard to prove which causes which. In some capacity, violence itself necessitates guns. Both statements could also be true at the same time. or playing the devils advocate, the increase in violence causes higher gun ownership, but once the ownership is there, reducing violence might not reduce gun violence itself.

I don't think there are many places where there is a complete absence of guns, but there are plenty of countries where gun ownership among civilians is outright banned. Japan comes to mind - and the gun related crime statistics seem to back up the fact that even criminals can't really get their hands on one.

Do you feel the same way about nuclear weapons? How about engineered viruses and pathogens?

This argument will fall apart when school shooters gain access to deadlier weapons.

There are (at least, partial) solutions, we just don't like them and they are really really hard:

1) totalitarianism and top-down technology control (and knowledge proliferation control)

2) scientific (prob through genetic engineering) or cultural attempts to alter human species to be more non-violent / more peaceful / abhor conflict, or resolve social conflict / scarcity / prevent the means by which people _want_ to kill each other.

Engineered viruses and pathogens are very simple to make. Within reach of PhD students in biomedicine fields, probably many physicians. I believe that simpler variants are within reach of skilled hobbyists and particularly able high school students.

Totalitarianism won't protect from these things. They are remarkably straightforward.

What has protected us from these things is the fact that people who like biology and medicine are typically relatively nice.

This isn't some deep secret knowledge, these are things so easy and un-novel that you can't publish a paper on them.

I live in New Zealand. Plenty of people are allowed to have guns (those with gun licenses) and plenty of people are not allowed to have guns (those without gun licenses). However, it does not appear to me that we are ruled by a gun-toting minority.

>This is the same semantic argument of "Guns don't hurt people". Of course, guns hurt people. A physical object can be harmful without a human being to give it intent. It's a straw man argument that delves into the meaning of words than actually addressing if something is actually harmful or not.

Given the proposition that 'objects' harm people without human involvement, that does not in any way improve the null-hypothesis of wanting to prevent harm. If an earthquake or tsunami hits we do not shrug it off as an act of nature or god. We have an intent to prevent the disaster even if no human was involved. The comparison to guns is an ill fitted one. A gun that no man knows about cannot hurt someone. A person still has to wield it, compared to AI, people are definitely the ones wielding the weapon. Weapons, like AI is a tool. whereas guns have a very clear target and modus operandi, AI has no such thing. AI is by definition innocent of every crime in the book since you cannot demonstrate it to have morality in the first place. Which is a problem if someone has used it to commit crimes. Obviously AI doesn't kill people, the people using AI want someone dead.

In the context of the article, it means we have to take responsibility for what the AI creates regardless of its intent, precisely because there might or might not be someone behind with intent.

The article doesn't seem to make any attempt at all to defend the first half of its thesis.

I watch a fair few Youtube Video Essays and one thing I notice is that there is quite a discontent with AI. Unfortunately there is a lot of preaching to the choir and not enough making a case. Nevertheless that discontent itself is a signal, and I think this article touches on why.

People are quite disillusioned. There is growing wealth inequality, political division, Dark patterns (which should simply be called abusive software). We now have a generation that does not expect to ever own a home.

You often hear the term now "Late Stage Capitalism" Which while admittedly was a clever term of speech when it was originated, has become a lazy expression of nihilism. It evokes the idea of capitalism as a cancer at a stage where the symptoms are obvious and debilitating, but also that it is at a point of no return and that it will kill itself and its host and there is nothing that can be done.

I rarely hear a detailed complaint about the problems of AI without the phrase "Late Stage Capitalism" cropping up. The irony is, of course, that the devices, platforms, and information used to express these ideas were in a large part due to innovations that were enabled by a capitalist economy. I think without the anger and frustration of their current situation, many of those railing against capitalism would admit that what they would like is a well regulated socialist/capitalist system. Something that works for society but affords people the means to innovate.

The problem for AI is that people fear that the empowering capabilities will go to the already powerful. As with wealth inequality, the expectation is that the power distribution of AI will also be unequal.

There are numerous other complaints about various aspects of AI, but I feel like these are things that should be argued individually but are often too heavily influenced by the issues above.

This feels only vaguely less affronting than "guns don't kill people, bullets do".

AI is, for the foreseeable, the work of extremely vast troves of resources. Countless of the most expensive chips on the planet, the best interconnects, the highest paid expert teams. Whatever the output is, these inputs are the firing mechanism that creates & keeps the AI firing.

I can't believe OpenAI employees threatened to mutiny so their CEO could exploit them for more personal wealth. Meanwhile Anthropic now has the highest-rated model.

> We shouldn’t fear AI as a technology. We should instead worry about who owns AI and how its owners wield AI to invade privacy and erode democracy.

The entire problem with this article is that it is arguing against technological determinism without even knowing it, and so it has no argument against it. But with the seductive power of technology and the ability for it to provide short-term marginal advantages to people, and combined with the fact that humans are horribly poor at acting to stop long-term detriments and tragedy, it is far more likely that AI and advanced technology will spread even if we can see the death at the end of the tunnel.

The only way we will gain more wisdom against advanced technology is to gain more wisdom without using advanced technology, because the immense power it offers is precisely the thing that makes it even harder to acquire that wisdom, simply because the power is too seductive.

Sorry to say, but we cannot handle AI and that is abundantly clear: the short-term advantages are too great, even though it promises a horrendous future. Sort of like burning fossil fuels.

It's the old Marxist means of production concept.

"Instead of enabling a world where we work less and live more, billionaires have designed a system to reward the few at the expense of the many. "

They say this as if it is a novel outcome.

To me, this idea that an AI can specially manipulate humanity into it's own destruction, only comes from the most cynical of people who are already in some form of business that relies on manipulation towards self destructive habits in order to generate profits.

I think it's an incredibly myopic view that willfully ignores everyone outside of the small bubble of capital clutching psychopaths that they mistake for a representative sample of "humanity."

I genuinely fail to appreciate this idea that AI, particularly as embodied through very expensive and poorly trained LLMs, is somehow "dangerous."

some of the many links in that article

https://apnews.com/article/google-chrome-privacy-lawsuit-set... https://apnews.com/article/private-incognito-browsing-explai... https://carnegieendowment.org/2023/09/07/facebook-telegram-a... https://epic.org/issues/consumer-privacy/data-brokers/ https://lithub.com/surveillance-capitalism-and-the-internet-... https://mitsloan.mit.edu/ideas-made-to-matter/machine-learni... https://news.harvard.edu/gazette/story/2019/03/harvard-profe... https://proton.me/blog/privacy-washing-2023 https://theconversation.com/explainer-what-is-surveillance-c...

https://www.abc.net.au/news/2018-06-25/healthengine-sharing-... https://www.apa.org/monitor/2024/04/addressing-equity-ethics... https://www.axios.com/2017/12/15/sean-parker-unloads-on-face... https://www.brennancenter.org/our-work/research-reports/clos... https://www.buzzfeednews.com/article/meghara/period-tracker-... https://www.commonsense.org/education/articles/a-majority-of... https://www.computerweekly.com/blog/Data-Matters/Data-is-not... https://www.courtlistener.com/docket/17216783/1096/brown-v-g... https://www.ftc.gov/news-events/news/press-releases/2018/01/... https://www.ftc.gov/news-events/news/press-releases/2023/05/... https://www.nbcphiladelphia.com/news/national-international/...

https://www.propublica.org/article/you-snooze-you-lose-insur... https://www.reuters.com/technology/biden-crack-down-us-data-.... https://www.scientificamerican.com/article/how-to-teach-comp... https://www.scientificamerican.com/article/information-overl... https://www.scientificamerican.com/article/robot-be-good/ https://www.scientificamerican.com/video/what-is-machine-lea... https://www.stitchdata.com/resources/predictive-analytics/ https://www.technologyreview.com/2024/03/05/1089449/nobody-k... https://www.theatlantic.com/ideas/archive/2021/10/facebook-p... https://www.theatlantic.com/ideas/archive/2021/11/facebooks-... https://www.theguardian.com/books/2019/oct/04/shoshana-zubof... https://www.vox.com/2018/2/27/17053758/phone-addictive-desig...

In other news, guns don't kill people. People do. etc etc.

I dislike guns or arguments that because there are guns we need more guns. I expect we’re ideologically aligned on guns.

Except you’re supporting the point.

AI is more like the state with its army and high tech rifles than a predator with a pistol.

AI will deeply centralize power if computing resources and data availability dominate.

> AI will deeply centralize power if computing resources and data availability dominate.

I'm sure the compute resources will remain centralized, but I'm growing increasingly doubtful of the current data landscape's ability to keep the data "clean" and not prone to Hapsburg AI scenarios or threat actors working to corrupt this data.

>> guns don't kill people. People do.

> I dislike guns or arguments that because there are guns we need more guns.

If people are ones doing the killing with guns, not guns in itself, it means that owners of guns are the ones responsible for killing. So it sounds like argument for regulating gun ownership. Like with driving license.

Guns make violence worse, so regulate them.

The prime actor in violence is a person, and that person still has the capacity for violence without a gun.

But let’s still prefer they don’t have one.

Reply on news.ycombinator.com