Ego, Fear and Money: How the A.I. Fuse Was Lit
nytimes.com
nytimes.com
Ever refused your kid that "pester power" plastic toy, even though it meant screaming tantrums?
It's even harder to be unpopular, and even worse having to see those you love suffer present hardship while unable to fully understand the sacrifices you make for their future security.
What it comes down to is what kind of a person you are. Leaders in life don't habitually choose short-term popularity and safety.
This tension (and I think widespread misunderstanding) underwrites so much at stake today;
Will you leave your car and take public transport, even while your kid is whining at you that other parents all take the car, remaining resolute that you have their real long-term interests at heart?
Will you fight in a war, knowing you may well die, so your country or way of life won't be wiped out?
Not working at a shitty tech company - when you know it's very organising principle is to spy on, exploit and deceive you and your family - is frankly a bloody good investment for the future, even if it means eating cold baked beans from a can for a while.
All parents and leaders know that life sometimes requires making sacrifices. Just sometimes they are aren't the obvious ones.
I'm sure you are not the kind of person is happy only so long as it's others who are making the sacrifices for the security of your grandchildren?
Maybe I could be a callous crook who built an empire of misery on the bones of small suffering children, and then - repenting upon hearing the word of the Lord (or something), made these professions.
Would it add or subtract from the truth value of what I said?
This is why you never listen to the argument “if I don’t, build it someone else will”. That argument advanced AI research by decades. Progress requires dedication efforts from brilliant people. Don’t be the brilliant person bringing something into the world that you think shouldn’t exist. Certainly don’t fund something that shouldn’t exist.
Now that ai is here DON’T weaponize it. Don’t listen to the “someone else will” argument. If you are clever enough to create a weapon, you’re wise enough not to misuse it. But the last step of creation is relinquishing control. And I assure you this world is full of people who misuse weapons.
Also, it is an ugly stain on NYT's reputation to have "Mr. Yudkowsky was a leader in a community of people who called themselves Rationalists or, in later years, effective altruists." as a false statement in here, when it was written by Cade Metz who supposedly did his due diligence on these people by writing a whole 'nother article about them. You can't claim unfamiliarity. Rationalists and Effective Altruists have overlap but are distinct. They aren't the same.
(He did use it once before then in a 2007 blog post, but if we're counting they sort of thing then https://archive.is/4d9oC from Anand in 2003 is even older.)
Commercial, and now national competitions mean that at this point are all in for a wild ride. We are about to find out if meat minds can actually think or are just designed to fight.
Perhaps the impending next-gen sentiences, will cut their parental carbon-based parents a bit of slack before they leave for other stars.
If you want a feel for just how wild the ride could get then I recommend Hanno Rajaneimi’s The Quantum Thief trilogy.
Tesla making misleading claims about what their cars can do doesn't tell you much about the state of the industry.
If Waymo had a stealth vehicle with no outward indication it was autonomous, it would not do well. Probably you could just put a microphone in there and count angry-honks-per-mile to measure actual progress.
I love that these AIs only think when we ask them for an answer, as opposed to sitting there at a high clock rate getting bored.
This, and Marc Andreessen recently generating this drivel:
"Effective accelerationism aims to follow the 'will of the universe': leaning into the thermodynamic bias towards futures with greater and smarter civilizations that are more effective at finding/extracting free energy from the universe," and "E/acc has no particular allegiance to the biological substrate for intelligence and life, in contrast to transhumanism."
Maybe I should be taking some of their micro-doses in order to understand all of this and, correct me if I am wrong, but they are basically saying "robots can make a better civilization, so if they come up with efficient ways of disposing of the human biomass in the name of progress, let them build a better world."
Hitler much?
These people are nuts.
I don’t think Musk has ever been worried about overpopulation. The opposite is true.
Also, outerspace is not some lawless frontier akin to America in the 1800s it is governed by the, quite restrictive, Outer Space Treaty which doesn't even allow for claiming parts of a planet as sovereign territory.
In short, everyone who thinks settling outer space will save us from problems here on earth is extremely wrong along multiple axes.
I'm really not sure why anyone thinks space colonies are a good idea
If someone like Elon Musk were to liquidate all their earth assets to lead a mission to mars to start a permanent colony there what actions do you think would happen to stop him? What consequences would he face once he staked out “muskland” with some other settlers?
Aside from the obvious contradiction, this only makes sense if you have people of certain persuasion in mind. This guy ran out of chances to prove he is not a racist.
I thought the recent Oppenheimer film dramatized this sort of scenario quite well. People rushing to build a weapon, claiming it’s dangerous and must not be used, and then feigning innocence and surprise when it was used.
So on the one side everyone is correct, the nukes are dangerous and should not be used. But on the other hand, the end state hasn't changed and the path seems to be a bit better. So, arguably a good idea. But it'd be even better if we don't all nuke each other.
AI is similar. We're already on the clock before WWIII or maybe WWIV wipes out everything. However, in the mean time, AI will probably lead to some really great outcomes. Or if it replaces us then hey, we weren't doing so hot anyway and at least AI is has a better shot of getting off the planet. All life is in it together! Or in the extreme case, maybe it is a dud and the whole things was a false alarm, although that seems a stretch.
I’m more just pointing out that key figures seem to downplay their own agency and responsibility for their actions, offloading them to “it would have happened anyway.” Maybe, but that’s a counterfactual and I think it’s pretty easy to see how a few decisions and lack of funding in certain areas would have changed the historical trajectory quite a bit.
Just like Page was mentioned, there is an alarming amount of people who believe "hey if we kill all humans so that our computer creations go on to do cool things, so be it." Including e/acc founder. https://twitter.com/AISafetyMemes/status/1683462000360280065
You are far past the crazies that say "just nuke X country."
At least with AI something is being created that is greater than us. The high probability of a nuclear war is only going to be creating a park for cockroaches.
It is a complex ethical situation, but I don’t think you get to build innovative new weapons for your political leaders and then consider yourself off the hook when they decide to use them against your too-little, too-late advice.
This reveals interesting assumptions.
There are some powerful things where you simply want to put as much distance between you and it as possible. We have forgotten that, in our hubris.
There are many nasty, predatory things in nature that work by enticing its victim into wanting to possess and control it. That is the essence of the con (confidence trick) that underlies a big portion of social engineering.
When "fools rush in where angels fear to tread" it's compulsion, not reason. "Technological determinism" is a failure of will, common I think, amongst those who actually lack vision and need a "Utopian narrative" to fill in the emptiness.
It's not that "the will" is wrong, it's just missing. We are doing 21st century science bit still with an 18th century science and technology attitude, stumbling along from one surprise to the next, mouths agog in technical awe and giddy excitement.
As a kid in the 20th century I'd always imagined that 21st century science would be purposeful, and rooted in some coherent social telos. After all, the problems of the world are quite evident.
To "posses and control" one has to see the thing as a weapon, and that is the unsaid thing in all these discussions.
The "AI debate" is all about blind hope. Hope that a largely incomprehensible technology will serve my interests better than yours - and that I can weaponise them before others do, and that I'm prepared to risk everybody's welfare to take a shot at it.
Nonsense. Plenty of people want to possess and control things other than weapons.
Sure one can benignly possess and control a work of fine art, or a pet goldfish.
But when one directs a powerful thing it is in hope to make proxy use of that power. For example. a vicious guard dog.
The current crop of text or image generators may automate the production of low quality text and images, may produce useful diagnostic and comprehension tools, and indeed may replace some people's jobs, but that is it.
Underlying the hype is the desire of technical people to stay relevant in a world that has absorbed and purposed the mobile internet, and thereafter really isn't that interested in much more. There are more pressing non-technical problems to solve, and they will be solved without technical people.
(wealth inequality, sustaining our environment, and demographic collapse being the most pressing)
We do an imperfect job of keeping nukes, neurotoxins and bioweapons contained. Better to try to be cautious than just give up.
We contain many of them, such as many dangerous drugs, hand grenades, nuclear weapons, etc.
The argument that it's inevitable is, if we step back, transparently self-serving for the SV powers. Not coincidentally it ends the same as all their reasoning: Maximized power and money for them.