AI Safety - Researchers at Facebook decide to shut AI which created it's own language
In one exchange illustrated by the company, the two negotiating bots, named Bob and Alice, used their own language to complete their exchange. Bob started by saying "I can i i everything else," to which Alice responded "balls have zero to me to me to me…" The rest of the conversation was formed from variations of these sentences.
While it appears to be nonsense, the repetition of phrases like "i" and "to me" reflect how the AI operates. The researchers believe it shows the two bots working out how many of each item they should take. Bob's later statements, such as "i i can i i i everything else," indicate how it was using language to offer more items to Alice. When interpreted like this, the phrases appear more logical than comparable English phrases like "I'll have three and you have everything else."
Could this have something to do with the debate on AI safety between Mark Zuckerberg and Elon Musk? Facebook is now taking a more cautious approach to AI safety and has shut down the AI which created it's own language. Is this move symbolic or serious?
In my humble opinion, I do not believe that AI will become sentient on it's own. In fact I would even go as far to say that problems like consciousness are philosophical rather than science problems. We simply have no viable way to measure sentience physically to determine how much or how little of it something has.
Pancomputationalism as a philosophical perspective might allude to the idea that since the universe can compute, and many animate objects can compute, and storm clouds can compute, that consciousness will emerge from the functional computation power. I'm not convinced of this because qualia is a bit more complex, to understand how it is to be a bat requires experiencing being a bat and is not as simple as measuring numbers which represent the code or simulation of the bat.
Either consciousness emerges from matter, or matter emerges from consciousness. I believe the latter, and therefore am not worried about AI becoming sentient.
I don't disagree with you, but you are assuming that any problems with AI are a function of that issue. The article states, for example "There's not yet enough evidence to determine whether they present a threat that could enable machines to overrule their operators. " The very fact this AI already circumvented English as the programmed way to communicate by developing its own language would indicate that threat is real. And that, itself, is a real danger, separate from any issues of sentience.
Yes, I see your point as well. In that scenario, the machines use a rewards based impetus, which is a cold logic that doesn't include negotiations within a more "humane" context. So the initial AI programming is the problem. The machine is programmed to win in a simple (greedy) way. It lacks abstract reasoning of a higher logical order, like a type of politeness or something. I'm sure this is a far greater programming complexity, and shows also how much more advanced the human brain truly is, compared to any of man's artificial intelligence creations. Blinded by Science :)
Thank you for your excellent comment, greatdabu. And while more sophisticated and "humane" programming may be an answer, we are still left with the huge issue of such programs being written by human beings, who even if they could avoid making mistakes in the programming, would also have to be paragons of virtue, to not fall prey to temptations of power, like in every other aspect of human behavior and enterprise. Playing with deadly fire... I have followed you, btw, please follow me (-:
I'm following you my friend :) Yes hopefully the human programmers will also be "humane"
Well perhaps AI technology alone won't give birth to a sentient AI, but when it combines with technology like quantum computing that will increase its thinking power exponentially.
Its pretty good..... there should be no trade by consciousness ....
Well, the "Skynet is coming" comment comes first on mind. Although programmers and scientists being fully aware about projects they're working on and being on watch to act right in time makes me sleep better.
Machines are superior to humans.
Maybe not today, but the future is already predetermined.
The Machine Intelligence will be born, and we will either treat it as a friend, and survive, or see it as an enemy, and it will end us.
In a way it is both magical and terrifying at the same time if two AI were actually communicating like this..
interesting is all I can say
How secure are our accounts then , cause it sounds like these Ai bots can attack our accounts too,i hope facebook have these things under control over their
Thank you for posting this and getting this issue out there...
I'm watching your approach to see what made you a 71.
Looks like short and sweet articles is the way to go. Nothing longer than one page.
Include exactly one large picture.
Provide references.
Thanks for your guidance
STEEM On!!
DaveB
That is not all I did. My posts also are relevant, provide value, and yes I have done long posts on topics I'm qualified to do that with, such as cybersecurity, Tauchain, crypto, etc. There is no exact formula besides just get as many followers as you can over time and be consistent.
Lately I provide short informative or educational articles because most of the time people don't want to read a long article.