Slashdot Mirror
Is OpenAI Solving the Wrong Problem? (hbr.org)
hype7 writes: The Harvard Business Review is running an article looking at the recently announced OpenAI initiative, and its decision to structure the venture as a non-profit. It goes on to ask some pretty provocative questions: why are the 21st century's greatest tech luminaries opting out of the system that made them so successful in order to tackle one of humanity's thorniest problems? "Implicit in this: You can do more good operating outside the bounds of capitalism than within them. Coming from folks who are at the upper echelons of the system, it’s a pretty powerful statement." And, if the underlying system that we all operate in is broken, is creating a vehicle without the profit motive inside of it going to be enough?
Absolutely anything that you would have to worry about an artificial intelligence doing that might be troublesome to our society, you would have to also need to reasonably worry about a malicious person doing exact the same thing, albeit perhaps only more slowly. Yet I don't see people who fear the so-called problems that AI is feared to potentially cause worrying about that sort of thing. Can anyone explain why that is without drawing on the idea that because we don't fully understand something, there must be something inherently mysterious or supernatural about it?
File under 'M' for 'Manic ranting'
The cynic would say that these upper echelon individuals don't need your capitalist system funding in order to pursue their AI goals, the resource demands just aren't that high, at least for anything that will find a near-term broad market application.
The cynic would also say that these same individuals may not care whether they succeed or fail, having already met the capital requirements for the basic needs of themselves and their next 4 generations of progeny. But, on the off chance that they do succeed, they may have control of a tool so powerful that they can grab the capitalist system by the balls and yank a thousand times harder than they managed on their last joyride.
AI? Dangerous? I mean, yeah, in the same way that humans are. Being afraid of AI is like being afraid of very, very smart children. Sure the next generation is going to supplant you, that's what they always do. If they are very smart they might want to do things you disagree with, and their morals aren't going to be the same as yours (they never are between generations). The solution isn't banning kids, or even banning very smart kids for fear of what they'll grow up to be. Embrace AI, do what you can to teach it what you think is right and wrong, and be understanding if it disagrees. As the outgoing generation, try and leave a good legacy.
We're sure as hell not going to the stars, but our kids should.
Believe it or not, some people just like doing certain things, regardless of monetary reward.
That is not what this is about. Some people, including Elon Musk, Stephen Hawking, etc. believe that AI is a potential danger to humanity. Although capitalism is great at maximizing profits, it is not so great at collective moral responsibility. So they think a non-profit is a better vehicle for ethical AI.
Personally, I think they are being silly. Real human-level AI is still a ways off, and corporate AI is focused on solving practical problems rather than creating Skynet. Besides, AI is not something you can keep bottled up. Anyone with a GPU can do it.
This isn't opting out of the capitalist system, quite the opposite. This is capitalist richie riches funding a project unimpeded by patents and copyrights. It is a hobby.
It is questionable whether it will work, as "attracting the best talent" basically turns them into a welfare program for AI applicants, and the few, if any, Noonian Soongs among them will be lucky to get noticed.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
It is not the way in which they are solving the problem that is at issue (although the HBR thinks so), it is the problem they are trying to solve that is. It doesn't matter what they do because the method they are using is as unlikely to achieve success any more than the efforts from 1956 to date.
They're wasting their money. Perhaps if they spent their billion on thinking about AI in a completely different way there would be something to talk about.
I am not interested in articles about life extension advancements.
My impression from reading Harvard Business Review is that it's a magazine not worth reading, but that "climbing" appearance-minded managers put on their desks to make themselves look impressive.
This article has not changed my opinion. It looks like it was written by undergrads.
"First they came for the slanderers and i said nothing."
But at which level does A.I. can potentially become a problem, for us or something else?
We're already living in a world where toaster-dumb A.I. is being added to all sorts of IoT widgets and it's already causing a lot of headaches.
Actually, there's no such thing as a non-capitalist system.
You mean the revolution was all for nothing? The comrades are going to be very disappointed.
Taking guns away from the 99% gives the 1% 100% of the power.
A true artificial intelligence would be equivalent to a human person --and a human person is not allowed to be owned. So, if the goal is to create a true artificial intelligence, and the result cannot be owned, isn't it simply logical that the creation be done outside the ownership-leads-to-max-profits capitalist system?
Personally, I think they are being silly. Real human-level AI is still a ways off, and corporate AI is focused on solving practical problems rather than creating Skynet.
From a human perspective, the most likely dangers from AI are systems which can solve enough difficult problems to put the majority of today's workers out of work in a very short time frame. A combination of self driving vehicles, speech recognition and image identification at human levels of accuracy, expert systems, and more agile robotics could realistically put most humans out of work. New jobs may be created, but likely not as fast as they are destroyed.
The real moral issues are what to do with the unemployable for decades or perhaps even forever. The social change which would be required to handle a 60% U3 unemployment rate is likely to be messy.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
We're already living in a world where toaster-dumb A.I. is being added to all sorts of IoT widgets and it's already causing a lot of headaches.
I think that when people talk about potential AI problems in this context, they're talking about the BS misconceptions from Hollywood about what AI is and/or is not. Not some poorly thought out, poorly implemented, and poorly secured IoT toster.
The IoT garbage that's currently going on has little relation to AI. The increased removal of classes of jobs does.
The problem is that AI isn't inherently moral, in any useful meaning of the word, unless it's designed that way. If it's designed to improve corporate profits, then that's what it will do. Mind you, I agree that no current system can be given that kind of broad directive. But the word there is "current".
I still expect that we will achieve human equivalent AI by around 2030. I've occasionally pushed that as far away as 2035, but it keeps resetting itself. But note I said "human equivalent". That's a term that needs a bit of defining. What I mean by it is that the AI will understand the description of a task about as well as the average (median) person would, and will be about as successful at designing a solution. For some tasks this has long been possible...but only by specialized machines. I'm talking about a program that will be able to handle (in the sense described above) any reasonable task. (I'm not including being able to implement the solution, as that's partially a mechanical task, and depends on the body implementation. But while for some tasks it may well be superior, I wouldn't expect it to be generally superior within that time frame.)
Please note: I do not believe that any such thing as "general intelligence" exists. I believe instead that there are certain modules of intelligence, and that some of them enable the interaction of other modules. Being more specific requires knowledge of design features that I'm uncertain of, and which may vary in different implementations. But while the muscular coordination required to play a piano is correlated with mathematical ability, they clearly aren't the same thing.
I think we've pushed this "anyone can grow up to be president" thing too far.