I am from Sydney which is a city in the state of NSW, in the country Australia. Very interesting discussion Scott. You did well to incorporate all the questions into your seminar.
i hate to sound rude, but people were asking a lot of stupid questions. how could you not see the value of Vicarious' program hitting the last pong block whereas Deep Mind could not? "but they could train Deep Mind with a million end-game examples"... yea they could, but the idea is to create something that does not need to be trained at all. as soon as you train for that, you will have cascading trillions of others to train for, too.
The audience made me crazy. It's like Phoenix is slowed to an irritating staccato by these folks who wanted to hear themselves talk rather than listen to the breakout work Vicarious is doing. Y'all hush and learn. Dang it.
The audience sounded like they had a bone to pick with the technology industry. Some of them were obviously "fishing" for points of criticism. However, one legitimate criticism that, for some reason, nobody had pointed out is that at an arbitrary point in the future, AGI's will be able to do every tasks a human can, thereby nullifying new labour niches in the market (since by that time there will be no niche that cannot be filled by an AGI.)
Hey Scott, the thing about having an AI read captcha is posable. There are a limited number of fonts, all with ten numbers and twenty six letters. each has a hight to width ratio. there are only a certain number of combinations that will result in your phrase coming up with the dimensions, making it an image. Have the AI try to reproduce the dimension from the available fonts and make it an image recognition problem instead of a language recognition problem.
It is an image recognition problem. All things in life have a limited amount of possibilities, the AI being able to sort through these possibilities largely depends on the number of possibilities. The ways in which you can write out letters are so grand that you'd never be able to draw all of them if I gave you hundreds of years.
You are correct. If you try to teach it to recognize whole characters. There are simple geometric shapes in every letter that distinguish it from every other letter in the alphabet. An A next to an O is distinguishable very simply. IT wont look for every possibility, only the difference between one character and the next. There are only 26 plus ten digits. Find these and the computer will easily solve the problem.
In order for the immense job loss to actually be apocalyptic, a great many things have to go wrong at the same time, and this is extremely improbable. Therefore a job-apocalypse will probably not happen.
what about the growing monster debt? Does this guy have any idea what will happen when treasury is unable to service the bond debt that it has grown since 1913?
18:15 ahh, so people should not be allowed to think the thoughts they want to think. their wrong think should be corrected by what the overlords who control the ai should agree with thank you for correcting my wrongthink, big brother
I agree with some of the people in the audience. Humans are still very ignorant of the interrelationships and the social interactions that take place within species. What was mentioned about animals working by "instinct"... that is to say moving around like robots is no longer accepted. Animals are more intelligent that we imagine.
Really good presentation. But what's with the audience? They are like stupid "OLD BRAIN" version. xD I feel bad for this guy. I hope some of them will read this.
Well, that got ...political. It seems the presenter oversimplified the state of the art. Possibly taking into account the make-up of his audience, whom heapparently failed to convince.
As someone who has very little experience in this field of knowledge, I really appreciate you guys posting this. I learned a lot. Thank you very much
I totally agree!
I am from Sydney which is a city in the state of NSW, in the country Australia. Very interesting discussion Scott. You did well to incorporate all the questions into your seminar.
Thanks for the clarification :) that turned into a funny argument at the end for sure.
The speaker was very good; I enjoyed the lecture part. I wish the questions had been saved to the end.
Thanks for sharing this with us Scott. Great talk.
i hate to sound rude, but people were asking a lot of stupid questions. how could you not see the value of Vicarious' program hitting the last pong block whereas Deep Mind could not? "but they could train Deep Mind with a million end-game examples"... yea they could, but the idea is to create something that does not need to be trained at all. as soon as you train for that, you will have cascading trillions of others to train for, too.
rockytunes89 because they’re bias nationalists and anti-government
The audience made me crazy. It's like Phoenix is slowed to an irritating staccato by these folks who wanted to hear themselves talk rather than listen to the breakout work Vicarious is doing. Y'all hush and learn. Dang it.
Great presentasi on
Great seminar, can you please provide some web links, where I can read more about casual learning and may be apply it on minute scale
The audience sounded like they had a bone to pick with the technology industry. Some of them were obviously "fishing" for points of criticism. However, one legitimate criticism that, for some reason, nobody had pointed out is that at an arbitrary point in the future, AGI's will be able to do every tasks a human can, thereby nullifying new labour niches in the market (since by that time there will be no niche that cannot be filled by an AGI.)
Hey Scott, the thing about having an AI read captcha is posable. There are a limited number of fonts, all with ten numbers and twenty six letters. each has a hight to width ratio. there are only a certain number of combinations that will result in your phrase coming up with the dimensions, making it an image. Have the AI try to reproduce the dimension from the available fonts and make it an image recognition problem instead of a language recognition problem.
It is an image recognition problem. All things in life have a limited amount of possibilities, the AI being able to sort through these possibilities largely depends on the number of possibilities. The ways in which you can write out letters are so grand that you'd never be able to draw all of them if I gave you hundreds of years.
You are correct. If you try to teach it to recognize whole characters. There are simple geometric shapes in every letter that distinguish it from every other letter in the alphabet. An A next to an O is distinguishable very simply. IT wont look for every possibility, only the difference between one character and the next. There are only 26 plus ten digits. Find these and the computer will easily solve the problem.
In order for the immense job loss to actually be apocalyptic, a great many things have to go wrong at the same time, and this is extremely improbable. Therefore a job-apocalypse will probably not happen.
Those old people asking really annoying questions that go on complete tangent are pretty frustrating....
FR
what about the growing monster debt? Does this guy have any idea what will happen when treasury is unable to service the bond debt that it has grown since 1913?
48:43
If anyone starts to do question? Skip. Thanks me later.
5 comments visible out of 7 comments total? I wonder what those other comments were...
22:00 I'm out. Off topic political debate. No thanks
18:15
ahh, so people should not be allowed to think the thoughts they want to think. their wrong think should be corrected by what the overlords who control the ai should agree with
thank you for correcting my wrongthink, big brother
I agree with some of the people in the audience. Humans are still very ignorant of the interrelationships and the social interactions that take place within species. What was mentioned about animals working by "instinct"... that is to say moving around like robots is no longer accepted. Animals are more intelligent that we imagine.
A⚡I
Pizza?!!!!!!
good point, except the political part. US is not hostile to immigrants. It's the illegal immigration thats the problem. Nothing to do with AI
Oh, and don't legally have a visa to pay for education here and take our AI! ;)
Christ, having this guy in a room and asking him questions about the taxis/car sharing slide? What an outrageous waste of opportunity. Knuckleheads!
Really good presentation.
But what's with the audience? They are like stupid "OLD BRAIN" version. xD
I feel bad for this guy.
I hope some of them will read this.
Focus on actual topic, not on political/economic bullshit!
Did these people walk in with a soap box.....if so surely they should have had it removed in the interest of thinking and learning......
ua-cam.com/video/Og8e1C-uqhk/v-deo.html same lecture without the annoying audience
Well, that got ...political. It seems the presenter oversimplified the state of the art. Possibly taking into account the make-up of his audience, whom heapparently failed to convince.
52:30