Videos

Making Friends With Artificial Intelligence: Eric Horvitz at TEDxAustin

38Views

Disclaimer

37 Comments

  • I noticed that I used to have shallow gasping breathing like that a few years ago.. Its slowly going away since I have been meditating and doing breathing exercises.
    I think Eric is too buzzed up!

  • I don't understand, what has your breathing got to do with AI? Other than that, my opinion is that when we actually get machines to help us with boring and other useful stuff, things will get so amazing 🙂

  • Hi, sorry it was a bit vague and is probably arrogant – what I was trying to say was that when Eric speaks he doesnt inhale hardly at all. I used to be like that when I was a bit overly cerebral and "wired' up.
    Ive since learned how to relax and be more calm and I try to balance with our natural world more.
    I recently watched a speech by one of the founders of skype on AI and it turned out to be a very cautionary tale about how we proceed with AI: Search: Jaan Tallinn, the Intelligence stairway

  • I've been looking into this as well. Machine Learning just looks like predicting some numbers on data, and AI just looks like hardcoded "fake intelligence". What would you suggest we do? Do you know a place to start?

  • we should do a natural inspired approach, simulating neurons and the connects and the change of the connections. We should try to undertsand how the cortex works exactly, the HTM algorithm and the ART (theory) are good starts but it doesn't explain everything.

  • Looks like full steam ahead with the creation of a superior life form. Wagering on an existential level should require democratic participation on a global scale. I don't doubt the sincerity of researchers like Eric or the authenticity of potential benefits. But, as machine learning progresses and accelerates, mans ability to anticipate and control outcomes diminishes. A possibility for changing course may cease to be an option. Little time to debate the most important decision in history!

  • There is extensive work being done on connectionist systems as opposed to the physical symbol (rule based) systems you mention. It is true that with a discrete symbol system, there is a problem with combinatorics.

  • I do. Its called soft AI. It uses statistical teqniques like artificial neural networks wich simulates the braincells. Im thinking about starting a channel explaining soft AI in comparison to hard AI as they talk about here.

  • This is way better than siri or google now … when will this kind of ai will be available? If this is the Real reason why Microsoft wants to include kinect with the x box one, I definitely would buy the console with kinect.

  • I personally believe the implementation of Google Glass can be used in Healthcare by downloading a series of Instructional videos to doctors who operate in remote villages across Africa and Third World. These videos will help train doctors on the surgical aspect of certain operations and protocols.

    Glass in the same manner, can become a pivotal role in bringing western education to classrooms which will create better business opportunities and the learning curve across poorer nations. #Glass

  • I personally believe the implementation of Google Glass can be used in Healthcare by downloading a series of Instructional videos to doctors who operate in remote villages across Africa and Third World. These videos will help train doctors on the surgical aspect of certain operations and protocols.

    Glass in the same manner, can become a pivotal role in bringing western education to classrooms which will create better business opportunities and the learning curve across poorer nations. #Glass

  • you're pleasantly optimistic, but once these AI's learn how we think, especially once they become "very powerful" AI's, wouldn't it be wise to be over cautious instead of blindly optimistic?  i like the idea of "making friends" with smart machines, but i dunno…. my mandate to survive informs me with a vigorously waving red flag, NOT TO TRUST.

  • i would love "making friends with artificial intelligence" they would really understand humans feelings but how do we shut up a friend of these if he decided to keep talking all the time. its hard to understand one of these friends even if i cant even sleep for a couple of days. i would say its "stress inducing" meet one of the friends.

  • AI has 3 categories:

    ANI: Narrow= one task (ex: pro at chess)
    AGI: General = multiple tasks equal to human capacity (many ANIs combined)
    ASI: Super intelligence = network of AGI = beyond human intelligence

    To Read more about this, go to the article waitbutwhy dot com

    Amazingly descriptive and easy to read to understand what AI is and where it's headed and what's known already along with all the possible outcomes by experts.

  • I would like that microsoft quit the project of (making friends with artificial intelligence) and devot its time to cloud computing and simulations…..and why those ADs with kids…..Microsoft almost never make ADs….

  • These TED talks are like the NPR of the world of intellectual inquiry, all designed to put your darkest speculative faculties to sleep in a gentle, peaceful slumber…..

  • Making friends with AI is easy. Just surrender your dignity, self respect, your job, your home, friends and family, hopes and dreams and we will get along just fine.

  • If engineers wanted to they could design AI with a secretive and highly secured tiered architecture for questioning. An average person gets many answers to questions posed, a mid level employee, academic or politician gets most answers and the top executive operator(s) of the AI gets any and all answers. Humans shouldn't be as worried about AGI running amok, but those fallible humans who may have primacy of control over it. That is, If there is to be any control over it.

Leave a Reply