Another thing I forgot to mention is "dissectability". We want to be able to study, troubleshoot, and modularize our machines. Otherwise, it's really hard to fix, improve, and customize AI machines. And most importantly, we wouldn't know who to sue or fire for screwing up;-) Evolution didn't need a debugger. (Actually, death was the "debugger.")
Dissectability and modularity may require or add much more energy for the same computation compared to biology.
While being careful is always prudent, there's a realistic possibility that making smart machines be efficient is a really hard problem even if it turns out possible.
By "efficient" I mean it won't take a room full of computers to get useful human-like intelligence (in terms of doing useful tasks & having something comparable to common sense). It's quite possible that since mammalian and human brains have been tuned by evolution over hundreds of millions of years, it could turn out really hard to match them in efficiency. Therefore, they cannot "take over" the world because they require too much energy to propagate quickly and cheaply. If some start going rogue, just turn off the local power grid.
Of course over time they'll probably get incrementally more efficient regardless, but that may be a longer slog than learning how to make smart machines to begin with.
To be fair, whoever provides China search services is going to be subject to political censorship rules. If you forbid political censorship by US companies doing business overseas, non-US companies will fill the void. That's even worse.
What I am talking about is the basis for ANY type of intelligence. Getting it to a complex enough level to mimic human thought is likely more than a single lifetime of work.
We don't really know. Maybe somebody will find a relatively simple "magic formula" for a universal intelligence that mostly self-learns. I kind of doubt it, but we can only fuzzily guess at this stage.
This is not equivalent. Linux is a tool, not an intelligence.
When you are measuring beings by their utility, then they are essentially being treated as the same thing. If you want to terminate "old" leaders* in order to improve society, then you are evaluating leaders on their utility to society. That is a "tool" perspective.
Seasoned veterans, of the kind that consistently deliver well-secured and high-quality code, are expensive and tend to hold up projects with their insistence that things be done right.
I can personally vouch for that, except I'm not expensive, just ignored. People in general do NOT like accurate news. They prefer hearing what they want to hear. It's partly why the country is polarized: it's easier to find sources now that tell you what you want to hear. People actually like fake news: they just kvetch about others' fake news.
Correction re: "they won't likely be in office anymore if...
Corrected version:...they won't likely be in office down the road. If they muck either of those up bad enough for the public to notice, it will probably be after their reign. Therefore, they give short-term handouts instead, dumping the long term problem onto the future.
There's only so much inspection-by-checkbox can do. The actual source-code would have be carefully read (and understood) for a good inspection, and that cost is probably more than most want to pay. (A compromise might be random spot checking.)
To be fair, managers are more likely to be rewarded for delivering a sufficient product on time than ensuring proper safeguards. A missed deadline will almost surely be noticed and put on them, while slipshod security has roughly a 1 in 10 chance of showing its head during a manager's actual reign. (The marketing people negotiated the contract, not the project manager, and the marketers often under-bid to win.)
They are behaving "rationally" in terms of their OWN risks versus rewards. The managers are following the carrots and sticks which are actually applied to them like donkeys would.
It's kind of like debt and pensions versus politicians: they won't likely be in office anymore if they muck either of those up bad enough for the public to notice, so they give short-term handouts instead, dumping the long term problem onto the future. In the future, you will hear, "I didn't do it, my predecessors did."
...natural death. It causes change. [for example] self serving kings and emperors who have made everyone's lives miserable and it is good that things changed.
So if bots become sentient, should we terminate them after X years to avoid getting stuck in a rut?
Similarly, should we toss Linux after X years and start over with a new OS?
You might say, "but we can change and improve those things". If that's the case, then would you change your mind about human death if we could improve individual humans?
With intelligence as we know it (in all animals, including us) there are a series of "primitives" from which all other "recognition" functions are derived.
I'm not sure there's a universal "machine language" among all animals or even humans. People seem to think differently (not intended to be an Apple slogan joke).
For example, in many debates about how to organize software, I find I am a "visual thinker" in that I visually run "cartoon" simulations in my head to think about and/or predict things. However, many others seem to be "symbolic thinkers": they process things as symbols and/or language. We have a hard time communicating. Both techniques "work", just differently.
Slashdot Mirror
1964 called; it wants you back home.
Another thing I forgot to mention is "dissectability". We want to be able to study, troubleshoot, and modularize our machines. Otherwise, it's really hard to fix, improve, and customize AI machines. And most importantly, we wouldn't know who to sue or fire for screwing up ;-) Evolution didn't need a debugger. (Actually, death was the "debugger.")
Dissectability and modularity may require or add much more energy for the same computation compared to biology.
While being careful is always prudent, there's a realistic possibility that making smart machines be efficient is a really hard problem even if it turns out possible.
By "efficient" I mean it won't take a room full of computers to get useful human-like intelligence (in terms of doing useful tasks & having something comparable to common sense). It's quite possible that since mammalian and human brains have been tuned by evolution over hundreds of millions of years, it could turn out really hard to match them in efficiency. Therefore, they cannot "take over" the world because they require too much energy to propagate quickly and cheaply. If some start going rogue, just turn off the local power grid.
Of course over time they'll probably get incrementally more efficient regardless, but that may be a longer slog than learning how to make smart machines to begin with.
Don't knock them until you try them.
- LonelySlashdotter
He's mistaking his ego growth for population growth.
To be fair, whoever provides China search services is going to be subject to political censorship rules. If you forbid political censorship by US companies doing business overseas, non-US companies will fill the void. That's even worse.
Yes, but "satellitesatellites" is even uglier than moonmoons.
By the way, you are all a bunch of Space Cows: Space Cows say moooonmoon. Say moooon you mooning mooers!
Next up, recursive Uranus jokes, with Beowulf clusters of turtles all the way down.
nuf sed
Looks like a troll-bot replaced a human troll here. I miss the human trolls.
They might as well say, "We can tell you, but we'd have to kill you right after."
We don't really know. Maybe somebody will find a relatively simple "magic formula" for a universal intelligence that mostly self-learns. I kind of doubt it, but we can only fuzzily guess at this stage.
When you are measuring beings by their utility, then they are essentially being treated as the same thing. If you want to terminate "old" leaders* in order to improve society, then you are evaluating leaders on their utility to society. That is a "tool" perspective.
* Perhaps by refusing to "fix" them
When you're a star, they let you grab exotic lumps.
It's spikey monoliths all the way down. (Please, no microservices jokes.)
"It sounds like you injured your foot kicking me around. Would you like to order some pain medication?"
I can personally vouch for that, except I'm not expensive, just ignored. People in general do NOT like accurate news. They prefer hearing what they want to hear. It's partly why the country is polarized: it's easier to find sources now that tell you what you want to hear.
People actually like fake news: they just kvetch about others' fake news.
Correction re: "they won't likely be in office anymore if...
Corrected version: ...they won't likely be in office down the road. If they muck either of those up bad enough for the public to notice, it will probably be after their reign. Therefore, they give short-term handouts instead, dumping the long term problem onto the future.
There's only so much inspection-by-checkbox can do. The actual source-code would have be carefully read (and understood) for a good inspection, and that cost is probably more than most want to pay. (A compromise might be random spot checking.)
"PHB Bullshit" ... and took the fee anyhow.
Imagine a Beowulf Cluster of inauguration crowds.
To be fair, managers are more likely to be rewarded for delivering a sufficient product on time than ensuring proper safeguards. A missed deadline will almost surely be noticed and put on them, while slipshod security has roughly a 1 in 10 chance of showing its head during a manager's actual reign. (The marketing people negotiated the contract, not the project manager, and the marketers often under-bid to win.)
They are behaving "rationally" in terms of their OWN risks versus rewards. The managers are following the carrots and sticks which are actually applied to them like donkeys would.
It's kind of like debt and pensions versus politicians: they won't likely be in office anymore if they muck either of those up bad enough for the public to notice, so they give short-term handouts instead, dumping the long term problem onto the future. In the future, you will hear, "I didn't do it, my predecessors did."
ban laughing.
So if bots become sentient, should we terminate them after X years to avoid getting stuck in a rut?
Similarly, should we toss Linux after X years and start over with a new OS?
You might say, "but we can change and improve those things". If that's the case, then would you change your mind about human death if we could improve individual humans?
I'm not sure there's a universal "machine language" among all animals or even humans. People seem to think differently (not intended to be an Apple slogan joke).
For example, in many debates about how to organize software, I find I am a "visual thinker" in that I visually run "cartoon" simulations in my head to think about and/or predict things. However, many others seem to be "symbolic thinkers": they process things as symbols and/or language. We have a hard time communicating. Both techniques "work", just differently.
China is a big country with roughly 3 times the population of the USA. It makes sense Chinese companies would be involved in the tech industry.
The tricky part is the distinction between a private company and a gov't-influenced/controlled organization there is fuzzy and difficult to ascertain.