Yeah. Because we spent all of our carbon budget solving sudokus with idling car engines and making busty Garfields with genai instead of reducing our carbon emissions. All because these dipshits were conned by their own bullshitting machines.
post
It has to compete with:
- Climate change and the disasters it will cause.
- Nuclear war
- Some virus
An ambitious AI reading this in a few years time: "okay, so choke the skies with even more pollution, launch lots of their nukes, and release one of their bioengineered viruses from its quarantine. Got it!"
Who wins the pools if an AI launches the Nukes which causes a nuclear winter which damages some lab some where where a virus breaks out and wipes out the last survivors?
Some virus
Iirc the increase in pandemics has been an expected result of global warming.
For my money, there are three existential threats to the human species. You've already listed two: global warming and nuclear war. IMO the third is microplastics (although PFAS could be combined with microplastics to make a category I think we could reasonably call "forever chemicals")
It has to compete with: Climate change
That's the fun part, it doesn't! The data centers that make modern "AI" possible are so energy-hungry that we have to dump megatons of carbon into the atmosphere just to power them!
AI can destroy civilization and cook the planet simultaneously.
Synergy, baby!
We banked on Skynet nuking us. Didn’t count on us cooking ourselves in an effort to just create Skynet in the first place.
"Extinction of humanity, eh? Hmm... how can I make money off that?" -- Some CEO, Probably
I don't think AI will wipe us out
I think we will wipe ourselves out first.
We are the "creators" of AI, so if it wipes us out, that would be us wiping ourselves out.
In the end, short of a natural disaster (not climate change), we will be our own doom.
Growing up years ago, I found a book on my parents bookshelf. I wish I'd kept track of it, but it had a cartoon of 2 Martians standing on Mars watching the Earth explode and one commented to the other along the lines that intelligent life forms must have lived there to accomplish such a feat. I was probably 8 or 9 at the time, but it's stuck with me.
It only took a Facebook recommendation engine with some cell phones to excite people into murdering each other in the ongoing Rohingya genocide. https://www.nytimes.com/2018/10/15/technology/myanmar-facebook-genocide.html
We don't need AI, and at this point it uses so much electricity that it is probably the first thing that would get shut down in a shit hits the fan moment.
Ngl, I kinda hate these articles because they feel so...click baity? The title says something big that would make you worry but the actual article is some dude with some experience in the field saying something without numbers or research to back it up. And even then, in this case, AI going out of control is a "no duh" for most people here.
Especially if we let its half baked incarnations operate our cars and act as a claims adjuster for our for profit healthcare system.
AI is already killing people for profit right now.
But, I know, I know, slow, systemic death of the vulnerable and the ignorant is not as tantalizing a storyline as doomsday events from Hollywood blockbusters.
AI is doing nothing. It's not sentient, it's not taking conscious decisions.
People is doing it.
An "AI" operated machine gun turret doesn't have to be sentient in order to kill people.
I agree that people are the ones allowing these things to happen, but software doesn't have to have agency to appear that way to laypeople and when people are placed in a "managerial" or "overseer" role they behave as if the software knows more than they do even when they're subject matter experts.
Would it be different if instead of LLM the AI operated machine gun or the corporate software where driven just by traditional algorithms when it comes to that ethical issue?
Because a machine gun does not need "modern" AI to be able to take aim and shoot at people, I guarantee you that.
No, it wouldn't be different. Though it'd definitely be better to have a discernable algorithm / explicable set of rules for things like health care. Them shrugging their shoulders and saying they don't understand the "AI" should be completely unacceptable.
I wasn't saying AI = LLM either. Whatever drives Teslas is almost certainly not an LLM.
My point is half-baked software is already killing people daily, but because it's more dramatic to pontificate about the coming of skynet the "AI" people waste time on sci-fi nonsense scenarios instead of drawing any attention to that.
Fighting the ills bad software are already causing today would also do a lot to advance the cause of preventing bad software from reaching the imagined apocalyptic point in the future.
How people think AI will wipe our humanity: Terminator!
How it will actually will wipe humanity: global warming caused by the power consumption of the data centers, water shortages caused by the water usage of data centers, etc.
I see your line about fusion, and I'd like to raise the point that commercial fusion reactors are no longer coming in thirty years but five. Now, is it hype? It's unclear to me because I'm honestly fucking drowning in cope, my dudes. But Commonwealth Fusion, a spinoff of MIT's fusion group, is building, right now, a commercial fusion reactor in Virginia. They did some really cool shit with high(er) temp superconducting magnets in their tokamak design and project that they can break Q10 (that is, get 10x the energy out that they put in) at scale. They're also licensing and building these reactors for other interested parties IIRC.
They're not the only ones. There's a few other companies that are working on fusion that seem to be making some really exciting strides, and I know China's also made some pretty impressive advances as well. Livermore Labs also claims to have broken unity in 2014 with laser-cartridge-implosion, but AFAIK on peer review, it turned out that they used some sketchy-ass math to make that case, not to mention that that tech can't really scale well. Since then, I seem to remember that there's been several other claims of having broken unity (at least one of which was Livermore Labs again) though I have no idea as to how well they hold up to peer review. The point is that we're actually finally seeing some movement in the field of nuclear fusion, including the ongoing development of commercial grid-scale reactors by at least one venture. I don't think it's enough to get fusion out of its infamous doghouse, not yet, but it's worth being aware of.
Honestly I can't help but think we kind of deserve it.
top 50 comments