Cool:
(My older son and I have done Folding@Home on our PCs in the past. I wonder if it is still valuable – any feedback from you professionals?)
1 Like
So, now the AI’s are teaching us new science as well as doing better engineering and teaching us how to play our hardest strategy games better than before. The idea that computers are not capable of intelligence is becoming more and more absurd. Sentient? No. They are not alive or conscious. But they are able to learn new things – they are not just calculators anymore.
It was the logical next step for Deep Mind in applying these new AI learning algorithms. Congrats and kudos to Deep Mind! I expect we will be hearing more of such breakthroughs from them.
2 Likes
It is just a powerful scientific tool.
1 Like
Intelligence is powerful tool for a lot of things including science. But it is just a tool.
But making this some kind of measure of either humanity or divinity is wrong.
I wasn’t. ‘Artificial’ is not part of the moniker for nothing, however
Probably not. But perhaps I was for much of my life. My parents made it seems SOOO important all the time.
My first reaction is… so…? And then I wondered… what does that mean to you exactly? Why is the computer’s intelligence artificial and our intelligent is not? explain that one.
This is really cool.
I’m looking forward to the future developments of this.
I always am curious about how AI verses I comes into play. After all if AI can learn new things then perhaps they can learn how to make a choice between two things and if that’s possible maybe they can develop preferences and if that’s possible
they can develop likes and dislikes and if one can then do can others. We don’t really know what AI will look like or be capable of in 1,000 years from now. The most common argument I hear is that AI has set parameters. But don’t we all? Likes and dislikes for us is based off of parameters. None of us likes the smell of rotting decaying roadkill. Why do some of us like horror and metal and some likes comedy and pop. It’s not like we just sat around and decided on this or that.
I feel like we don’t fully understand it in our own species and experiences. It’s still debated on if free will even exists. So for me there are to many unknowns on the future of AI. It’s very exciting though. Skynet! Jk.
The protein calculations in itself is cool. I heard on , Common Descent podcast, a while back that there is also a computer that after being fed a lot of data was able to correctly predict a series of possible mutations , including some that was discovered in fossils after the fact.
Processors is probably a better choice. But do they possess intelligence in the same sense we do, or do they rather reflect the intelligence of their programmers?
No. That at least we can say for sure they do not.
It is more that intelligence is a complicated collection of abilities and some of those the computers simply do not have while others they exceed our own. Certainly when it comes to many kinds of thing which look like intelligence to us, computers do better than we do.
They don’t just calculate better in chess and other strategy games, they actually teach us how to play better – showing us how we were doing it wrong. In chess for example, our professionals chess players focused on the innate power of the pieces they would have left, the computers beat the professionals by ignoring this and focused more on the mobility of the pieces they had left and upon immobilizing the pieces of the other player.
In engineering they find solution which we would never think of.
We look at the results of evolution and the designs look like the product of a superior intelligence. So Dawkins invented the word “designoid” for this. Computers have mastered the same kinds of algorithms to surpass us on many different kinds of challenges, proving that they have the kind of intelligence we think we are seeing in the products of evolution. So in that sense at least, it looks like intelligence superior to our own. The reality is more complicated but I think we are justified in concluding that to a first approximation computers have it.
I think it is a greater distortion of the language to exclude these things from intelligence rather than to narrow intelligence down to what the computers can do so well. And instead to look for our own value in other things which computers don’t do so well, in those aspects which haven’t been the primary focus of what we have called intelligence. We most likely need a bigger vocabulary. But I at least have tentatively chosen to distinguish sentience from intelligence to distinguish what humans can do from what computers can do so well.
So what is it that we can do better which I am identifying with sentience: assigning ourselves new tasks, thinking outside the box, breaking the rules? Computers excel at following rules, and achieving the tasks given them, searching and finding better solutions. It might be tempting to call our version creative intelligence, but that doesn’t quite hit the nail on the head – remember their mastery of the evolutionary algorithms which certainly look very creative to us.
I noticed. Very even handed I think. Heck, English is so flexible, why not characterize what they do as intelligent?
Computers are really good at parallel processing, iterating, and calculating. This makes them really good at finding associations hidden deep in data sets. However, they aren’t that great at tasks most humans find simple, such as depth perception in 2d photographs. Intuition seems to be the one thing computers have a difficult time emulating. At the same time, humans use intuition to fill in gaps which can make us worse at some tasks.
1 Like
GPT-2 is spooky. It’s able to write better than most humans.
These recent advances is AI are fairly new and we don’t know limits of it yet. Some of those things you mention may simply be things we haven’t got to yet and we will soon get computers to master them. I suspect that finding what we can really do better at than computers is going to tell us something about the nature of living things as opposed to machines.
1 Like
Just don’t turn your back on them or you may find yourself locked out of your spaceship being apologized to by a very condescending sounding AI!
Tackling the difference…
Obviously computers are limited by their hardware and programming. We expand those limitations by altering the hardware and programming. This became capable of surprising us when this gave them enough flexibility to go beyond what we could predict. There are many math problems or calculations which we cannot solve without iterative methods using computers. With the learning algorithms this simply went so much farther beyond what we could predict.
With living organisms the ability to go beyond what we can predict is much more basic as part of the nature of life in its simplest and earliest forms. Thus it was able to evolve without engineers working to expands its capabilities. The separation between software and hardware isn’t there as far as the evolutionary algorithm is concerned or to put it another way the evolutionary algorithm is part of the very structure of life and how it operates. We can at least theoretically close the gap to some degree by putting computers in charge of designing the hardware of the next generation of computer – don’t know if that is a wise thing to do.
This topic was automatically closed 6 days after the last reply. New replies are no longer allowed.