Discussion of Intelligence Amplification, Artificial Intelligence, Mind Uploading, Super Intelligence.
New Minds, and Humanity's Future
- Thread starter Tim Walker
- Start date
www.genericminds.com/mildai/index.shtml
Essentially, "mild intelligence" would be...please excuse the term...a kind of Idiot Savant.
Our current electronic computers might be considered a form of this.
I imagine a symbiotic relationship between such and humans.
Essentially, "mild intelligence" would be...please excuse the term...a kind of Idiot Savant.
Our current electronic computers might be considered a form of this.
I imagine a symbiotic relationship between such and humans.
“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.”
― Edsger W. Dijkstra
Augmenting human cognition with computer power seems like the way forward to me. More direct integration with the wetware would shorten the feedback loop. It's always seemed pointless to try to build an independent AI that is equal or superior to a human mind -- such an endeavour is likely to be be inordinately difficult if not fruitless.
― Edsger W. Dijkstra
Augmenting human cognition with computer power seems like the way forward to me. More direct integration with the wetware would shorten the feedback loop. It's always seemed pointless to try to build an independent AI that is equal or superior to a human mind -- such an endeavour is likely to be be inordinately difficult if not fruitless.
When can I get the Library of Congress on a chip in my brain?
More importantly, how long before the chip can access Pr0nHub?
But seriously, if we start giving people the ability to access the internet mentally, we're going to need MUCH better filters. Too many people already think that sites like Mercola and NaturalNews provide actual information, not utter rubbish.
More importantly, how long before the chip can access Pr0nHub?
But seriously, if we start giving people the ability to access the internet mentally, we're going to need MUCH better filters. Too many people already think that sites like Mercola and NaturalNews provide actual information, not utter rubbish.
Worse than that, I'm thinking about the kind of malware you could get into your brain.
Imagine mind-clicking the wrong link, and now every night you dream about Brand X penis enlargement pills. You're right in the middle of an awesome dream, and then "TRY NEW PENIS ENLARGEMENT, SHE'LL LOVE IT." Over and over.
Hopefully my subscription to Malwarebytes will stop most of that.
Then we just have to worry about DRM. Or the FBI getting Apple to write a backdoor to your iBrain. They'd LOVE that.
I can't wait for the first case in which the FBI wants to exhume someone's body so they can get at the decedent's neural implant and attempt to recover encryption keys from the implant's EEPROM.
Book, copyright 2013. Our Final Invention Artificial Intelligence and the End of the Human Era by James Barrat.
Terminology:
AI stands for Artificial Intelligence.
AGI stands for Artificial General Intelligence. A machine intelligence that can match humans intellectually.
ASI stands for Artificial Super Intelligence. A machine intelligence that intellectually exceeds humans.
IA stands for Intelligence augmentation. A human is intellectually boosted with a human/machine interface.
"Friendly AI" would be a machine intelligence that is programmed to react positively/benignly towards humans.
Terminology:
AI stands for Artificial Intelligence.
AGI stands for Artificial General Intelligence. A machine intelligence that can match humans intellectually.
ASI stands for Artificial Super Intelligence. A machine intelligence that intellectually exceeds humans.
IA stands for Intelligence augmentation. A human is intellectually boosted with a human/machine interface.
"Friendly AI" would be a machine intelligence that is programmed to react positively/benignly towards humans.
Last edited:
We need to master this first:
And then maybe look into this:
And then maybe look into this:
Good luck with that.A useful term to know: Moravec's paradox
https://en.wikipedia.org/wiki/Moravec's_paradox
http://up-ship.com/blog/?p=31449
https://en.wikipedia.org/wiki/Moravec's_paradox
http://up-ship.com/blog/?p=31449
http://io9.gizmodo.com/5923496
I started thinking about this topic a few years back when I came across After The Internet: Alien Intelligence by James Martin.
Please pardon the term, but the computer you are staring at could be deemed a sort of idiot savant. What about the next step, a more capable idiot savant? Will we have more capable idiot savants, but with really weird quirks?
I started thinking about this topic a few years back when I came across After The Internet: Alien Intelligence by James Martin.
Please pardon the term, but the computer you are staring at could be deemed a sort of idiot savant. What about the next step, a more capable idiot savant? Will we have more capable idiot savants, but with really weird quirks?
Last edited:
Rupert Murdoch has already done this.
Expert systems provide a glimpse into how the progression really works.
I don't know why singularitans and speculative writers keep missing this point, but paths of evolution for AI are limited by forces of practicality. That is, experimental AIs can develop a lot of interesting characteristics (Deep Blue et al) but the PRACTICAL uses for AI experiments tend to be pretty limited. From an evolutionary standpoint, this means that the traits that make AIs really interesting also limit their reproductive chances because the engineers who build them never have a reason to build more than one. In other words, humanlike intelligence is a not a fitness trait.
Granted this is really "artificial selection" but that's par for the course with artificial intelligence and/or artificial life forms.
So the better evolutionary paths revolve around utility. IBMs Watson, for example, can be repurposed for practical applications in the real world, as an expert system for hire. As it gains more capabilities, the need to expand its processing power and/or build more Watsons gives it greater growth potential. Which means, ultimately, that the Expert System is the strongest evolutionary contender for AIs.
Looking forward, I would expect that the most advanced AIs in our future will all be Problem Solver systems. Autocad or CNC machinery augmented with expert systems that make design and manufacturing even more efficient and removing some of the human expertise required in the design process. Down the road, you COULD eventually get to the point where the AIs no longer need human input to design new products, just knowing the intended specifications -- or better yet, the intended purpose of the product -- is enough for them to work out the details on their own. You get an expert system where the VP of concepts says "I need you to develop a small combat aircraft with V/STOL capability, a payload of around 15 tons, supersonic and supercruise. Don't care about the service ceiling or stealth characteristics, but it has to be able to mount an anti-satellite missile and it has to cost less than fifty million a piece." And then have the Expert System crunch those parameters and spit out a finished design plan -- and possibly a prototype -- 24 hours later.
Those expert systems probably wouldn't have decision-making power for a VERY long time, if ever. The only way they might is if somebody gets the idea to use those systems to streamline administrative processes, in which case the CEOs, Mayors and Governors become rubber stamps for policies the computers calculate will result in optimum benefit to everyone. Robot uprising potential: the AIs judge that their human counterparts are too greedy and too shortsighted to actually cede control to the AIs and that it is in everyone's best interest if human leaders were quietly and painlessly removed from effective power. The most efficient revolution is certainly a bloodless one, and if anyone could pull that off it would be a political expert system.
I don't know why singularitans and speculative writers keep missing this point, but paths of evolution for AI are limited by forces of practicality. That is, experimental AIs can develop a lot of interesting characteristics (Deep Blue et al) but the PRACTICAL uses for AI experiments tend to be pretty limited. From an evolutionary standpoint, this means that the traits that make AIs really interesting also limit their reproductive chances because the engineers who build them never have a reason to build more than one. In other words, humanlike intelligence is a not a fitness trait.
Granted this is really "artificial selection" but that's par for the course with artificial intelligence and/or artificial life forms.
So the better evolutionary paths revolve around utility. IBMs Watson, for example, can be repurposed for practical applications in the real world, as an expert system for hire. As it gains more capabilities, the need to expand its processing power and/or build more Watsons gives it greater growth potential. Which means, ultimately, that the Expert System is the strongest evolutionary contender for AIs.
Looking forward, I would expect that the most advanced AIs in our future will all be Problem Solver systems. Autocad or CNC machinery augmented with expert systems that make design and manufacturing even more efficient and removing some of the human expertise required in the design process. Down the road, you COULD eventually get to the point where the AIs no longer need human input to design new products, just knowing the intended specifications -- or better yet, the intended purpose of the product -- is enough for them to work out the details on their own. You get an expert system where the VP of concepts says "I need you to develop a small combat aircraft with V/STOL capability, a payload of around 15 tons, supersonic and supercruise. Don't care about the service ceiling or stealth characteristics, but it has to be able to mount an anti-satellite missile and it has to cost less than fifty million a piece." And then have the Expert System crunch those parameters and spit out a finished design plan -- and possibly a prototype -- 24 hours later.
Those expert systems probably wouldn't have decision-making power for a VERY long time, if ever. The only way they might is if somebody gets the idea to use those systems to streamline administrative processes, in which case the CEOs, Mayors and Governors become rubber stamps for policies the computers calculate will result in optimum benefit to everyone. Robot uprising potential: the AIs judge that their human counterparts are too greedy and too shortsighted to actually cede control to the AIs and that it is in everyone's best interest if human leaders were quietly and painlessly removed from effective power. The most efficient revolution is certainly a bloodless one, and if anyone could pull that off it would be a political expert system.
^ If so, that is ALMOST a good reason for me to actually read an Honor Harrington novel.
