Should we / I worry about AI "taking over the world "?
-
Or it emphasizes his point -- there is zero "intelligence" in AI, it's just a monkey learning colors and passing matching blocks. When I execute a search, I must review the answers, as some are obviously wrong (for my need), so I have to exert myself to locate the most correct answer(s) I can. The biggest danger of so-called AI is that people will begin to trust it, when it's just a monkey doing color matching. "Training" reduces the number of obviously wrong answers, and possibly increases the number of "may be correct" answers. Intelligence and experience is required to determine the validity, and AI does not have that.
There is a tremendous amount of intelligence in a monkey learning colors and passing matching blocks. It's not enough intelligence to write an epic poem or wonder what's beyond the horizon perhaps, but might be enough to take over the world if left unsupervised around just the right kind of blocks. An AGI that gets the answer wrong sometimes could still be as successful as, say, a politician or CEO. I think there is far more to be concerned about than you may realize. Of course, YMMV.
-
Says a guy on the world wide web (the same web that 5 billion other people will use this year) that barely existed 30 years ago. The same guy that probably has a smart phone that has 5000x more computing power than the fastest super computers of the 1980's in 1/5000th the space. Most 12 year old kids have the same smart phone... Never mind other AMAZING things our parents never dreamed of as kids: flash drives, SSDs, fiber optics, the human genome project, graphene, WiFi & Bluetooth, Large Hadron Collider, AbioCor artificial hearts, artificial joints, stem cells, gene editing (CRISPR), laser/robotic surgery, GPS, MRIs, facial recognition, cheap drones, 3D printing, etc...
fgs1963 wrote:
Says a guy on the world wide web
One success does not mean that every other prediction is also a success. Cars did not exist two hundred years ago. But flying cars (the ones and usage actually predicted) do not and never will.
fgs1963 wrote:
Never mind other AMAZING things our parents never dreamed of as kids
I gave you a long list of other predictions that do not, and are unlikely to, ever exist. Betting on one stock which makes one a millionaire does not mean that betting on all will make everyone a millionaire. That is why people do those seminars to teach others how to invest. Because getting paid for those seminars does make one wealthy.
-
jschell wrote:
Future being what exactly?
Sometime after now. :rolleyes:
jschell wrote:
Development of "AI" started in the 1950s.
If you haven't noticed, human technological advancement a) sometimes happens in leaps and b) is escalating at a geometric rate not linear. In the 1950's there was severely limited hardware* and only a handful of developers working on AI. Now we have exaflop super computers and thousands of very well financed developers working on AI. If you don't see AI becoming radically better over the short term (ie. the next few decades) then you might want to open your eyes. *The 1960 Cray CDC1604 was the fastest super computer ever made at the time. It was 48bit, 192kB of memory and operated at 0.1 MIPS.
fgs1963 wrote:
sometimes happens in leaps and
No I haven't noticed that. Actually, since I am more aware of it now, I am come to be aware that it never happens that way. It just seems like that since people are not looking at the full process. Technology is not built on the shoulders of giants. It is built on the shoulders of very normal size people who made incrementally improved perhaps just one thing. Moreover technology is never what drives that. Rather businesses do. Ford (the person) did not invent anything. He put an bunch of bits and pieces (some at least decades old) and then publicized it for very specific business reasons. Same for Edison. Cell phones have a very specific progression from the utility of satellite phones. Computers have a very smooth transition from larger computers and the very real need and demand by businesses as the saw the benifits. Applying technology to agriculture has been something that has been going on since agriculture existed. The Wright brothers were absolutely not even close to being the first ones to 'fly'. They were not even the first ones to put a human on a motor propelled flying machine (that happened in France.) For close to one hundred years that was how it was depicted though. Look even at the following which specifically mentions that they used a wind tunnel but in fact that was invented by someone else. 1903 Wright Flyer | National Air and Space Museum[^] Cell phones have had so much impact not because of the technology but rather because it is just so cheap to put up the towers. Businesses world wide could see the need and desire to communicate and providing that was so cheap and profitable that they did it. The technology allows it but it does not do it. If it was just a matter of technology then the first Mars-Earth war would have already happened.
fgs1963 wrote:
In the 1950's there was severely limited hardware*
And yet there was significant investments in that hardware made just so AI research could go on. It wasn't faltering due a lack of hardware.
-
Says a guy on the world wide web (the same web that 5 billion other people will use this year) that barely existed 30 years ago. The same guy that probably has a smart phone that has 5000x more computing power than the fastest super computers of the 1980's in 1/5000th the space. Most 12 year old kids have the same smart phone... Never mind other AMAZING things our parents never dreamed of as kids: flash drives, SSDs, fiber optics, the human genome project, graphene, WiFi & Bluetooth, Large Hadron Collider, AbioCor artificial hearts, artificial joints, stem cells, gene editing (CRISPR), laser/robotic surgery, GPS, MRIs, facial recognition, cheap drones, 3D printing, etc...
fgs1963 wrote:
Says a guy on the world wide we
Another way to say what I said... Hindsight does not make one a genius at predicting the future.
fgs1963 wrote:
Never mind other AMAZING things our parents never dreamed of as kids
Naming the past does not have anything do with assuring that a future prediction will come to pass. And it definitely does not mean all future predictions will come true.
-
fgs1963 wrote:
Says a guy on the world wide we
Another way to say what I said... Hindsight does not make one a genius at predicting the future.
fgs1963 wrote:
Never mind other AMAZING things our parents never dreamed of as kids
Naming the past does not have anything do with assuring that a future prediction will come to pass. And it definitely does not mean all future predictions will come true.
Nor do any of your examples assure that AI will fail. My point is that technology is advancing at an alarming rate and AI is likely to be part of that advancement. If you disagree... that's fine. I won't be alive in the time frames I've mentioned (~50 years) so unless it happens a lot sooner I'll never know if I'm right or wrong.
-
Self-driving cars will last exactly as long as it takes the Great Unwashed to realize that they're buying machines which can choose to kill them based on an algorithm!
-
As opposed to driving on roads where other humans can kill them on purpose or by accident. :doh:
Yes, but you're not choosing to pay the other humans to do it.
-
Yes, but you're not choosing to pay the other humans to do it.
True, but people already die in cars they bought. Most cars already have things like throttle by wire, cruise control, power steering, power brakes, anti-lock brakes, etc... High end cars are adding automatic collision avoidance, lane centering and automatic parking to manual drive cars. High end pickups will automatically center your tow ball to your trailer hitch. Full self drive is the ultimate goal. It will get here eventually - its just harder than most people think.
-
Kate-X257 wrote:
we need to prepare and act pro-actively to detect and actively counter active threats. If one group creates a malicious AI, you basically need a more advanced and specialized AI system to actively counter that threat.
So we should focus on that one AI versus the potential millions of people around the world that continuously seek active harm through technology?
To give an example: Phishing and social engineering are examples of threats. Due to AI being employed, these threats will scale up in scope and effectiveness. I propose we use AI to counter these threats more effectively. Nowhere do I claim we should focus on AI as a threat. I explicitly started with: it's not a threat.
-
True, but people already die in cars they bought. Most cars already have things like throttle by wire, cruise control, power steering, power brakes, anti-lock brakes, etc... High end cars are adding automatic collision avoidance, lane centering and automatic parking to manual drive cars. High end pickups will automatically center your tow ball to your trailer hitch. Full self drive is the ultimate goal. It will get here eventually - its just harder than most people think.
The difference between those things and self-driving cars is this: Imagine your self-driving car is driving down a narrow road lined with cars, and a couple of small children jump out right in front of it. The only way it can stop itself from killing the kids is to crash into a parked car, potentially killing the driver (or passenger, or whatever you choose to call it). Would you buy something which might choose to do that? Could you honestly say you'd trust it not to do that for the sake of a large dog, instead? It's not the same as lane centering.
-
The difference between those things and self-driving cars is this: Imagine your self-driving car is driving down a narrow road lined with cars, and a couple of small children jump out right in front of it. The only way it can stop itself from killing the kids is to crash into a parked car, potentially killing the driver (or passenger, or whatever you choose to call it). Would you buy something which might choose to do that? Could you honestly say you'd trust it not to do that for the sake of a large dog, instead? It's not the same as lane centering.
...and again - this happens already with human drivers. People have killed themselves and others in car accidents caused by an effort to avoid a squirrel. It happens. A self driving car might actually perform better than the average driver in a panic. On the other side of the coin, you should consider human faults that routinely cause accidents today that would be avoided with a self driving car. Drunk drivers. Driving while using a cell phone. Medical events (stroke, heart attach, seizure, etc...)
-
...and again - this happens already with human drivers. People have killed themselves and others in car accidents caused by an effort to avoid a squirrel. It happens. A self driving car might actually perform better than the average driver in a panic. On the other side of the coin, you should consider human faults that routinely cause accidents today that would be avoided with a self driving car. Drunk drivers. Driving while using a cell phone. Medical events (stroke, heart attach, seizure, etc...)
You're missing the point. Yes: a human driver can do all these things. But you don't go to the human driver store and deliberately PAY MONEY to get one of them to do it!
-
You're missing the point. Yes: a human driver can do all these things. But you don't go to the human driver store and deliberately PAY MONEY to get one of them to do it!
-
Oh right. Yes: I'd forgotten that those cost $100,000 or whatever a Tesla costs these days.
-
Oh right. Yes: I'd forgotten that those cost $100,000 or whatever a Tesla costs these days.