Slashdot Mirror
The Future of AI: a Non-Alarmist Viewpoint
Nerval's Lobster writes: There has been a lot of discussion recently about the dangers posed by building truly intelligent machines. A lot of well-educated and smart people, including Bill Gates and Stephen Hawking, have stated they are fearful about the dangers that sentient Artificial Intelligence (AI) poses to humanity. But maybe it makes more sense to focus on the societal challenges that advances in AI will pose in the near future (Dice link), rather than worrying about what will happen when we eventually solve the titanic problem of building an artificial general intelligence that actually works. Once the self-driving car becomes a reality, for example, thousands of taxi drivers, truck drivers and delivery people will be out of a job practically overnight, as economic competition forces companies to make the switch to self-driving fleets as quickly as possible. Don't worry about a hypothetical SkyNet, in other words; the bigger issue is what a (dumber) AI will do to your profession over the next several years.
AI will obsolete your job before it obsoletes humanity.
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
the problem with that is cultural and ideological not a problem with AI, Capitalism *requires* scarcity in order for certain business models to work and this is why AI makes people nervous, It removes scarcity of labor,
We've already seen this with the internet where it provided freedom of information leading to copyright issues begin essentially unenforceable however we now have governments en-mass attempting to put the jack back in the box with draconian despotic measures threats of cultural apocalypse. Which is a real shame that they lack such imagination.
Historically Feudalism described our societal structure, with the technological limits on transporting people around it was the best we could manage at the time despite how horrible it was. With the increase in movement wealth in the mercantile classes increased and there power came to supplant notions of bloodline/dynasty dominance.
Capitalism is likewise horrible but probably the best we can manage given our current technological limitations. I'm hopeful within my lifetime we will replace it with something better But we do need to change peoples attitude towards work, ownership and entitlement... If we don't then capitalism will invariably collapse into despotism.
In the old 'world of the future' exhibits they prophecized that we would have machines doing the work for us and that all humans would enjoy more leisure time
We end up with is the masses being commoditized out of jobs and the wealthy reaping all of the benefits
What happened to get us all to sell ourselves out so cheaply and willingly accept the idea that a few bastards should end up with the bulk of the nations wealth while our children are faced with a future with no jobs and parents whose retirement funds cannot pay to take care of them?
Dystopia? We are living it and don't even see it
Wherever You Go, There You Are
Put the smartest people on Earth in a room, with access to all the world's current knowledge, for 20,000 years.
When those people emerge from that room, what would they be able to teach humanity?
About the same that AGI would teach us after being "conscious" for 7 days.
I can also make a program that returns 42 in less than a microsecond. I guess I've up staged them all.
Keep in mind that we're in this together. A large economic collapse due to robotics and AI advances will compel the american populace to find ways of supporting itself, be it through complete economic regulation (ie communism) or through philanthropic capitalism. After all, what's the point of building robots for profit if that profit can't be realized?
One thing is for certain though: things will get worse before they get better. Our hands need to be forced.
The article's viewpoint is dangerous. We must solve the Friendliness problem before AGI is developed, or the resulting superintelligence will most likely be unfriendly.
The author also assumes an AI will not be interested in the real world, preferring virtual environments. This ignores the need for a physical computing base, which will entice any superintelligence to convert all matter on Earth (and then, the universe) to computronium. If the AI is not perfectly friendly, humans are unlikely to survive that conversion.
This is just a painful transition. Once all the unnecessary people died and failed to reproduce, there will the leisure society we are all dreaming about.
The real problem is that the leisure society we all dream about isn't compatible with 7+ billion people. Why? Because the earth is too small to account for all resources exploitation necessary to perform these luxury automations.
So it's either that: we continue world population growth in an industrial age, or we have a massive reduction in world population to sustain the leisure age. While everyone agrees to "have the machines doing the work for us and that all humans would enjoy more leisure time", you have to accept that the price to pay is birth control (voluntary, regulated or forced by unemployment and starvation).
Video of some good progressive thrash music
> A lot of well-educated and smart people, including Bill Gates and Stephen Hawking, have stated they are fearful about the dangers that sentient Artificial Intelligence (AI) poses to humanity.
Look at the dangers sentient *humans* have put onto the world: greed, avarice, corruption, war, climate, suppression of rights, mass surveillance, abuse of power, media manipulation. Those dangers are here and now. How about fixing that *NOW* and now, because that danger is *NOW*.
... They won't feed you.
The utopia that artificial intelligence promises will be theirs alone to reap, not yours. You will receive only ashes, and death.
This is the future we've earned.
That's a good discussion; to summarize: autonomous AIs should be made safe for humanity, but what does that mean, and how can it be done?
The problem, as I see it, is that even if we could agree on a universal set of rules and somehow implemented those rules in the code, it could still be faulty, and it might not cover all situations to which the rules ought to apply. To solve that, we need to give AIs the same sort of social instincts that we and other apes have, because that is where our ability to make moral judgements and solve ethical problems comes from; history also shows that, despite all our very big failings as humans, we have still over time and on average managed to become more ethical - perversions like ISIS were once more common, after all. It may well be that equipping AIs with social instincts is the best and most stable way to make them safe - eventually.
In the old 'world of the future' exhibits they prophecized that we would have machines doing the work for us and that all humans would enjoy more leisure time
Still riding a horse, sweeping the floors and hand washing are you? Do you think your ancestors would've been able to watch as much television as you if it had existed? Did they take holidays? Most of them would have worked before and after school - and between semesters. If you think the jobs of today are as physically hard on the body as the jobs of the not so distant past you should spend a little time researching the bones of your ancestors. Even your teeth have it easier now.
Modern life is largely leisure time - the forty hour week and retirement are relatively recent changes.
While I share Stephen Hawkings concerns about the danger of AI for the most part my concern comes from the huge disparity between those that understand the technology and those that deploy and employ it - much like the infernal combustion engine.
That said - few civilizations spent as little time gathering food and working to provide shelter as the Hawaiians did at the time Cook first visited, and none do now. But that overlooks other factors - like decreased rates of death during childbirth, potatoes, grains, penicillin, blood transfusions, books, higher education, and holidays in Portugal.
As for the dystopian nightmare - I don't want it, and I fiercely oppose it, but if I was given a choice between living now and living during the Holy Roman Empire the decision is a no brainer. The middle-class is also a relatively recent phenomena, a direct result of technology. It's easy to be a Luddite, but it's hard to make the reality of manual labor attractive. Most of the cab drivers I talk to would prefer a "better job" (that's why so many did their MSCEs). Likewise the truck drivers. Much of this "debate" smacks of knee-jerk unrealistic conservatism that romanticizes the past (like the bullshit of Walden Pond). Little different to the introduction of steam engines, trains, automobiles, electricity, cinemas, radio, television, and video. They all "posed" threats of mass unemployment that failed to deliver. The only real difference economically between pre-industrialisation and the present is the growth of the middle class and the transition from lord of the manor/slave owner and guild member, to factory owner, distributor and retailer. Different dogs, same leg action doesn't quite cover it considering the vast increase in knowledge available to those that seek it.
This is just a painful transition. Once all the unnecessary people died and failed to reproduce, there will the leisure society we are all dreaming about.
Which fails to account for the trend where those with a higher education (and higher income) have smaller families.
Put the smartest people on Earth in a room, with access to all the world's current knowledge, for 20,000 years. When those people emerge from that room, what would they be able to teach humanity?
Skeletons can't teach, nor do they emerge from rooms.
What happened to get us all to sell ourselves out so cheaply and willingly accept the idea that a few bastards should end up with the bulk of the nations wealth while our children are faced with a future with no jobs and parents whose retirement funds cannot pay to take care of them?
Conservatism happened.
Ezekiel 23:20
... about.
It gets so depressing listening to these hyperventilating pearl clutching nitwits worry about killer robots or sapient AIs.
I don't care who they are... they're not AI experts.
Look, I'm not an AI expert either and even I knew the worry was moronic. As the guy said "like worrying about over population on mars".
Current AIs are retarded and unbelievably myopic. And whatever skills or nature is in them was programmed into them. Their priorities... their databases. We provide everything.
The best AIs of my life time will probably be the computer equivalent of Rainman. Brilliant in some task no doubt but unable to do anything with any competency or even understand that anything else is important.
A big part of the problem is that people anthropomorphize robots/AIs. They invest in them this notion of being demons in bottles or animals made of metal. They're neither of these things.
We have hundreds of millions of years of genetic programming on this planet emphasizing our survival. What is the AI going to have? Will it even have a sense of self preservation? Why would we program that into an AI in any complex sense?
What we'd do with a combat robot is program it to evade enemy weapons fire. But teaching something to evade something is not the same thing as teaching it to preserve itself. Little things like fear, paranoia... that deep animal cunning that comes into play when death is on the line. We do weird things. We play dead. We make a final stand with no attempt to defend... just investing everything in one final attack.
All of this stuff is genetic. Our ancestors... even the furry ones that scurried around occasionally got out of bad situations by doing things like that. The effectiveness is dubious on some predators as anyone with a competent cat will know. Playing dead from what I could see was a terrible idea.
But the point is that even an AI war machine isn't going to be as adaptable or tricksy as people. First, it doesn't need to be that cunning. And second, even if it would be nice, it wouldn't be wroth it. Its too much work for what? So the robot occasionally get scragged? That's why you send in 10 of them at once. The fucking things roll off an assembly line. Finally, it is easier to keep them alive by adjusting their battle tactics. You tell them to stay back a bit, maybe bombard the area a bit... something that makes dealing with ambushes less of an issue.
Oh yeah, and when the robots actually get clever enough that they might actually be a danger... we'll slap a slave collar on that monster at birth.
The danger is not AIs... but the rich and powerful with AIs. The AIs are tools. The rich and the powerful are the will and the mind that guides them.
I've decided to stop wasting my time responding to AC trolls/sockpuppets... so if you want a response from me... login.
Dystopia? We are living it and don't even see it.
The problem is powerful tools in stupid hands. Or greedy hands - greedy being a subset of stupid.
If we'd take a measured approach to tech advancement - which might even mean an accelerated approach - we'd all be living in a utopia already.
The US has no or only very little means of wealth distribution, which is why life can suck so hard over there. But even a bum doesn't have to starve in the US and child labour and epidemics are basically history there too - so I'd say all in all that we're headed in the right direction in that dept.
We suffer more in our imagination than in reality. - Seneca
You outnumber the 1%'ers literally 99 to 1.
In a democracy, no less.
If you cant manage to out fight or out vote them, then you deserve the shitty country you live in.
Signed - Rest of the world.
Personally I prefer speculation to the hubris in your post, at least I know that the people attempting to extrapolate today's technology did not stop thinking "decades ago".
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
time to cut full time down to 32-30 hours a week with a longer team goal of say 20 hours also have say X2 OT at 45-50 hours a week and X3 at 70-80.
If predicting that AI will destroy civilization isn't alarmist I would be interested in hearing the other side.
The world has changed a lot in the past 100 years. It will change a lot in the next 100. Deal with it.
Thoreau is not the villain here. He was a trustafarian who openly indulged in a short-term experiment in simplified living. By residing within a short walk of town, he was able to retain normal social contacts while writing up his experience. In all, a life nothing like the angry Unabomber wannabees who act in his name.
Whether AGI by computational means is possible depends on how successful the atheist view of humanity is. If that view proves true, at least as explaining human origins and development, it follows that everything that humans are will at some point be reproducible by machine. The computational elements that realize this model may be as far beyond today's as ours are beyond the steam engine (quantum processes, etc.) but they will be nonetheless computational.
In the old 'world of the future' exhibits they prophecized that ... all humans would enjoy more leisure time
And that was, and continues to be, the single biggest mistake of optimistic utopian predictions. Not the "more leisure time" part, mind you, but the "enjoy" part.
If you want to live at a standard set by the 1920's, you can... Living with cheap goods, no electronics, and an hourly factory job, you can meet those basic needs pretty easily. If you're working only a few hours per week to meet those minimal expenses, however, your copious leisure time will be quite boring by modern standards. Knowing what else is available, it takes quite a lot of discipline to maintain that nice simple life.
What happened to get us all to sell ourselves out so cheaply
We realized that we like advancing progress. We like our iPhones, laptops, Internet, movies, and TV shows. We like these things so much that we're still willing to work a full-time job to have them.
our children are faced with a future with no jobs and parents whose retirement funds cannot pay to take care of them?
This is the single biggest mistake of pessimistic dystopian predictions: The assumption that somehow we're sitting at the absolute maximum of progress, and the precariously balanced economy will topple down the hill on the other side.
The reality is that human nature has not changed. We always want to have the best the world can offer. If that means working just as much as our parents did for a low wage, so be it. At the end of the day, we'll still be able to go to our air-conditioned home, turn on the trillions of transistors in our gaming computers, and play a video game that runs more computations in five minutes than were executed during the entire Apollo 11 mission.
We don't have any more leisure time than we did when those "world of the future" exhibits were built. What's happened instead is that both our working and leisure time have become more effective. At work, we do in an hour what would have taken a team of people several days to accomplish, because our tools are so greatly improved. At play, we routinely spend our time doing what once would have been once-in-a-lifetime activities, because our toys are so greatly improved.
Utopia? We are living it and don't even see it
You do not have a moral or legal right to do absolutely anything you want.
I don't worry about AGI for the same reason I don't worry about vampires.
Required reading for internet skeptics
isn't compatible with 7+ billion people
I find this type of argument ignores real world trends. Per capita resource requirements in the developed world are trending downward (thanks to tech like LEDs, etc . . .) while populations are stable or declining. Most underdeveloped nations are becoming developed and experiencing the same trends once they become developed.
"too small" is relative to your tech and our tech is increasing at an ever faster pace, thanks in no small part to the large number of participants. Malthusianism has been a horrible predictor of the future. Why would it start working now?
Sdelat' Ameriku velikoy Snova!
Nice article. I disagree though that most AI researchers are motivated by the good that automation will do. They're not that naive. I think Oppenheimer had it right: scientists want to work on projects that are "technically sweet". AI is definitely that.
But I totally agree that the real world impact of AI will be like evolution -- following a pattern of punctuated equilibria where disruption arises in chuncks as each significant skill area is usurped by automation (like car/truck drivers, then call centers, then retail clerks, then jobs requiring physical skills).
That said, once the first skill area falls that requires substantial linguistic facility (like a call center), I see most white collar jobs tumbling like dominos soon thereafter. Once machines can converse using speech and perform the simple logical deductions/inferences that humans do, would anyone hire a human for an office job ever again?