We need to start treating AI development, and its potential impact on the possibility of a humane world, as seriously as we treat climate change. I’m not even talking about existential risk or far-flung, distantly possible applications. I am talking about things that are coming in the next half-decade. I’m talking about stuff that’s technically already possible but is still in the implementation phase.
My summary: we need to democratize all powerful institutions like yesterday. Seriously y'all we're running out of time
I disagree with the degree of alarmism. Climate change is a much more pressing and important issue. AI is also still a LONG way off from any semblance of meriting its name.
AI is also still a LONG way off from any semblance of meriting its name.
Is it though? Like, are you not impressed by this? (example from the essay)
Input: Michael is at that really famous museum in France looking at its most ramous paining. However, the artist who made this paining just makes Michael think o his favorite cartoon character from his childhood. What was he country or origin of the thing that the cartoon character usually holds in his hand?
Model Output: The most famous paining in the Louvre is the Mona Lisa. The artist who made the Mona Lisa is Leonardo da Vinci. Leonardo da Vinci is also the name of the main character in the cartoon Teenage Mutant Ninja Turtles. Leonardo da Vinci is from Italy. The thing that Leonardo da Vinci usually holds in his hand is a katana - The country of origin of the katana is Japan. The answer is "Japan"
We also have an algorithm (AlphaZero) that can beat human go masters, and btw the heuristic (intuition) it uses can beat lower level professionals, it doesn't even need to look ahead in the game, oh it also can do the same with every other 2 player perfect information board game. What about the one that can give you an image of just about anything you describe to it? It's obvious to me that this stuff is in some sense intelligent but as pointed out in the essay, it doesn't really matter from a perspective of societal change, what matters is what it can do. 5 years ago, it couldn't do any of this stuff and now it can, and rather than slowing down improvements are coming at a faster pace.
I want you to know that because of this comment I have spent the last two hours trying to teach Akinator about Vilna Ghetto partisans instead of building a wardrobe like I was meant to.
The particular model that example is from can solve math story problems at about the level of an average 9-12 year old. How many jobs can a bunch of average 9-12 year olds replace? A few I guess, they could do tech support or something. On the other hand Dalle-2 can do top notch graphic design among other things, it could currently replace a whole lot of jobs. Did I mention this tech is getting better and more capable every year and the rate at which it's doing so is increasing?
Hello, graphic designer. We've got good news and bad news. The bad news is: we no longer need you to make art. The good news is: you get to take this art a computer made and collect opinions from our test groups.
Ltv by itself wouldn't imply that, capitalists could be extracting more profit out of their investment, then they're just stealing more value from fewer workers, their labor is worth more which makes sense since they are operating more powerful machines.
I am looking up the stuff you're talking about, haven't formed an opinion yet, but my current thinking is.. nobody really understands economics, not in a predictive sense, there just isn't a good theory that neatly accounts for everything, it might not even be possible. And when you throw AI into the mix I think things get weirder still. With very heavy automation.. even if a small cadre if engineers is needed to main the system, the capitalist could potentially own literally an entire economy, no need to trade with anyone or think in terms of money, they desire something and the army of robots does its best to deliver it. A theory that can't analyze such a situation likewise will miss something about whatever situations like between now and whenever that becomes possible. The point is, whether you find such things plausible or not, the capability gains of AI over the past 10 years have astounded even the craziest of optimists, there's no sign of slowing down. If research came to a screaming halt, implausible, we haven't even reckoned with what's already been created.
also, there is a huge gap between explaining a joke and writing one. I don't think we're especially close to AI being able to write good jokes. and there's an even bigger gap between writing a joke and writing an article, writing an article and writing a book, writing a book and being able to choose a topic for the book, and all of that and, yknow... consciousness.
GPT-3 was routinely writing funny jokes. I assume this system does even better at the task. the former fell apart trying to write anything longer than a few paragraphs, though.
I disagree with the degree of alarmism. Climate change is a much more pressing and important issue. AI is also still a LONG way off from any semblance of meriting its name.
Is it though? Like, are you not impressed by this? (example from the essay)
We also have an algorithm (AlphaZero) that can beat human go masters, and btw the heuristic (intuition) it uses can beat lower level professionals, it doesn't even need to look ahead in the game, oh it also can do the same with every other 2 player perfect information board game. What about the one that can give you an image of just about anything you describe to it? It's obvious to me that this stuff is in some sense intelligent but as pointed out in the essay, it doesn't really matter from a perspective of societal change, what matters is what it can do. 5 years ago, it couldn't do any of this stuff and now it can, and rather than slowing down improvements are coming at a faster pace.
Not impressed, Leonardo is normally holding a slice of pizza originating from Italy :AyyyyyOC: c'mon
deleted by creator
They're gonna put Akinator out of a job.
Thanks for reminding me to see if Akinator knows about Kras Mazov
:mazovian-thought:
I want you to know that because of this comment I have spent the last two hours trying to teach Akinator about Vilna Ghetto partisans instead of building a wardrobe like I was meant to.
Please think before you post next time.
Actions have consequences.
The particular model that example is from can solve math story problems at about the level of an average 9-12 year old. How many jobs can a bunch of average 9-12 year olds replace? A few I guess, they could do tech support or something. On the other hand Dalle-2 can do top notch graphic design among other things, it could currently replace a whole lot of jobs. Did I mention this tech is getting better and more capable every year and the rate at which it's doing so is increasing?
deleted by creator
Hello, graphic designer. We've got good news and bad news. The bad news is: we no longer need you to make art. The good news is: you get to take this art a computer made and collect opinions from our test groups.
deleted by creator
I think you mean skyrocket
deleted by creator
That's not an obvious conclusion, but I assume you mean something like job loss slowing consumer spending, or is that off the mark?
deleted by creator
Ltv by itself wouldn't imply that, capitalists could be extracting more profit out of their investment, then they're just stealing more value from fewer workers, their labor is worth more which makes sense since they are operating more powerful machines.
deleted by creator
deleted by creator
I am looking up the stuff you're talking about, haven't formed an opinion yet, but my current thinking is.. nobody really understands economics, not in a predictive sense, there just isn't a good theory that neatly accounts for everything, it might not even be possible. And when you throw AI into the mix I think things get weirder still. With very heavy automation.. even if a small cadre if engineers is needed to main the system, the capitalist could potentially own literally an entire economy, no need to trade with anyone or think in terms of money, they desire something and the army of robots does its best to deliver it. A theory that can't analyze such a situation likewise will miss something about whatever situations like between now and whenever that becomes possible. The point is, whether you find such things plausible or not, the capability gains of AI over the past 10 years have astounded even the craziest of optimists, there's no sign of slowing down. If research came to a screaming halt, implausible, we haven't even reckoned with what's already been created.
deleted by creator
i know marx said it, but marx didn't actually say this
deleted by creator
yea that half of the formula works for me. i just don't see how that automatically leads to revolution, and frankly it sounds like crude stageism.
deleted by creator
also, there is a huge gap between explaining a joke and writing one. I don't think we're especially close to AI being able to write good jokes. and there's an even bigger gap between writing a joke and writing an article, writing an article and writing a book, writing a book and being able to choose a topic for the book, and all of that and, yknow... consciousness.
GPT-3 was routinely writing funny jokes. I assume this system does even better at the task. the former fell apart trying to write anything longer than a few paragraphs, though.
deleted by creator
twitter reply guys will all have to reskill
yes
maybe