This is a really great conversation. Kofinas asks "What are people still good at"? Presumably as machines get good at doing some of the things that people do now, the people will move on and do things that require even more intelligence.
With much more separation than you could ever get on roads, with ≥2 qualified professionals in the cockpit who'll lose their jobs if they mess up, and with infrastructure for routing, positioning & landing #weakanalogy
One important case can be made for very fast recursive algorithms re-adapting an evaluation in a situation change. As in a bicycle just probably /possibly swerving as needed, or a premature braking of a driver or after any other human intervention. Here, a human would slower to correct his previous decision, or perhaps a previous reflex may interfere. Intuitively, I would like an A I to know people-behavior and thinking as well as specifically me and my emotions and reflexes before I'd let it brake or swerve, cutting me off from decision making. Fine if you can best me at chess, but know me first, and work with me and that drunken, computerless idiot ahead of me, before you and take joint responsibility on the road. This way, in paired and shared interdependence, the sum total of all together will be maximized. For now, this could be developed only by a central, specialized "brain-compound" in constant communication, rather than a resident 'brain' in each car.
I've heard the story of Robert Moses building low bridges on purpose to keep blacks out of the beaches many times, but it's probably not right: www.citylab.com/transportation/2017/07/how-low-did-he-go/533019/
lets say a group of people surround the car and want to hurt the driver and/or steal the car. Would the algorithm be able to give people with worse qualities (like those trying to steal cars and hurt people) a lower-weighted value in the associated neural nodes?
I spoke too soon. The idea that capitalism is THE driver for research and development is wrong. So much of what we think of as having been developed by capitalists has been state funded by one country or another for a long time. This fact has been raised by many well respected economists and political thinkers. 59:05
Total Harm = Severity X Frequency. 'Solving' the trolley problem marginally reduces severity at best. It may be more effective to reduce harm by focusing on preventing accidents in the first place. This may explain engineers dismissing trolley problem.
The was a story kicking around that Chinese car makers were going to use social-media credit scoring as a factor in deciding who lives and dies in a Trolley Problem situation. Better behave yourselves online folks!
To have the courts to be as fair and as unbiased as they can be, a good start would be to use a blind judge who never gets to see nor speak with the accused. Because there isn't much justice in letting the judge _judge_ a person for her character, how well she can portray herself or how much caucasian/african/asian she is. In that context of a blind or apart judge an expert system could help the judge digest the known hard facts that are really the only thing that matters regarding a crime and help detach the professional from his own emotions regarding the cases. Hunch based judgments could get lost in the process, but maybe in the big scheme of things that'd be a good thing.
If the trolley is approaching, and I am the only person on one track, and there are five people on the other track, I will be concentrating on getting out of the way, not worrying about the ethical design of the control algorithm. I couldn't blame the algorithm for choosing the path of least death and destruction, even if that meant it had to head for me. My job would be to skedaddle. If I can't do that in time, it's my fault, not the algorithm's.
Computers are bean-counters and statisticians, nothing more. It was a great mistake to name them "INTELLIGENCE". Whoever thought it up couldn't have been very intelligent, having no idea of its disastrous consequences. 🥴 (My laptop didn't write this.)
How is it even possible to think of humans (who are imperfect) creating/programming algorithms, to then provide us with information that we are told is the absolute gospel. "Humans are idiots."
PEOPLE: It's not the machine. Its the pogrammers that are coding the algorithms, running the machine. Which in turn, ends up influencing/controlling your decision or life. Put the blame on the programmers. Not the machine, when things start going wrong.....
Mmmmmm....... Hannah Fry ticks all of my boxes....... * Beautiful (looks and personality) * Stunning red hair * Has a voice as smooth and warm as melted chocolate being poured onto ice-cream....... :)
Good conversation with a public intellectual and thought leader
lol lol
Wow! 6 years ago yet so relevant today. Great interview : so glad providence led me to this.
Love to all from England
Such a great interview, its so refreshing to hear interesting conversation where the host has clearly done his research on her work
Great interview, host and guest! Thorough, thoughtful, well done! Thank you
She has such a beautiful voice. A refined British accent adds 20 points to your IQ.
I could listen to Hannah forever, shes great and so smart. 'Magic Numbers' is a great watch for those who have not seen it!
Same, bro.. same. Hannah is just effin awesome..
What an awesome discussion! Thanks Demetri (I am a big fan) and Hannah. Good luck with the book.
This is a really great conversation. Kofinas asks "What are people still good at"? Presumably as machines get good at doing some of the things that people do now, the people will move on and do things that require even more intelligence.
What a woman!... I could listen/watch Hannah for days.
What an interesting lady. Thanks for sharing.
I should listen to this on Spotify
Fantastic host. Keep it up!
The comment section is awesome.
I have just reserved this book at the library. Sorry Hannah, I am bit too skint at the moment to actually buy the book.
Keep it up.
Airplanes have been flying autonomously more or less for around 20 years.
In general this has resulted in less air accidents.
They've been flying autonomously with a pilot and copilot behind the stick. That's obviously not the same thing
With much more separation than you could ever get on roads, with ≥2 qualified professionals in the cockpit who'll lose their jobs if they mess up, and with infrastructure for routing, positioning & landing
#weakanalogy
Μπράβο Μιτσάρα συνέχισε έτσι και με παρόμοια θεματολογία!
One important case can be made for very fast recursive algorithms re-adapting an evaluation in a situation change. As in a bicycle just probably /possibly swerving as needed, or a premature braking of a driver or after any other human intervention. Here, a human would slower to correct his previous decision, or perhaps a previous reflex may interfere. Intuitively, I would like an A I to know people-behavior and thinking as well as specifically me and my emotions and reflexes before I'd let it brake or swerve, cutting me off from decision making. Fine if you can best me at chess, but know me first, and work with me and that drunken, computerless idiot ahead of me, before you and take joint responsibility on the road. This way, in paired and shared interdependence, the sum total of all together will be maximized. For now, this could be developed only by a central, specialized "brain-compound" in constant communication, rather than a resident 'brain' in each car.
I've heard the story of Robert Moses building low bridges on purpose to keep blacks out of the beaches many times, but it's probably not right: www.citylab.com/transportation/2017/07/how-low-did-he-go/533019/
lets say a group of people surround the car and want to hurt the driver and/or steal the car. Would the algorithm be able to give people with worse qualities (like those trying to steal cars and hurt people) a lower-weighted value in the associated neural nodes?
He is so in love with this fascinating, beautiful woman.
Oh, wait, that's me.
That aside, I have to get that book.
I spoke too soon.
The idea that capitalism is THE driver for research and development is wrong.
So much of what we think of as having been developed by capitalists has been state funded by one country or another for a long time.
This fact has been raised by many well respected economists and political thinkers.
59:05
Total Harm = Severity X Frequency. 'Solving' the trolley problem marginally reduces severity at best. It may be more effective to reduce harm by focusing on preventing accidents in the first place. This may explain engineers dismissing trolley problem.
The was a story kicking around that Chinese car makers were going to use social-media credit scoring as a factor in deciding who lives and dies in a Trolley Problem situation. Better behave yourselves online folks!
To have the courts to be as fair and as unbiased as they can be, a good start would be to use a blind judge who never gets to see nor speak with the accused. Because there isn't much justice in letting the judge _judge_ a person for her character, how well she can portray herself or how much caucasian/african/asian she is. In that context of a blind or apart judge an expert system could help the judge digest the known hard facts that are really the only thing that matters regarding a crime and help detach the professional from his own emotions regarding the cases. Hunch based judgments could get lost in the process, but maybe in the big scheme of things that'd be a good thing.
I like my skills. I resist automation x
Is it really a trolley problem if one of the people on the track is you?
If the trolley is approaching, and I am the only person on one track, and there are five people on the other track, I will be concentrating on getting out of the way, not worrying about the ethical design of the control algorithm. I couldn't blame the algorithm for choosing the path of least death and destruction, even if that meant it had to head for me. My job would be to skedaddle. If I can't do that in time, it's my fault, not the algorithm's.
Numberphile woman!!!
Computers in general disenfranchise people, especially older people that have never used a computer before.
Wrong.
Computers are inanimate and have no intention, or motive.
People's unwillingness to learn disenfranchises themselves.
IBM cheated in that match with Kasparov. There was an excellent documentary made about it.
Computers are bean-counters and statisticians, nothing more. It was a great mistake to name them "INTELLIGENCE". Whoever thought it up couldn't have been very intelligent, having no idea of its disastrous consequences. 🥴 (My laptop didn't write this.)
How is it even possible to think of humans (who are imperfect) creating/programming algorithms, to then provide us with information that we are told is the absolute gospel. "Humans are idiots."
PEOPLE: It's not the machine. Its the pogrammers that are coding the algorithms, running the machine.
Which in turn, ends up influencing/controlling your decision or life.
Put the blame on the programmers. Not the machine, when things start going wrong.....
Mmmmmm....... Hannah Fry ticks all of my boxes.......
* Beautiful (looks and personality)
* Stunning red hair
* Has a voice as smooth and warm as melted chocolate being poured onto ice-cream....... :)
Her accent sounds like Norwich(UK), is involved somewhere.
Alan Garland She’s from Essex 🙂
@@GraemeMarkNI
Ah, my guess is not very far amiss then.
100% agree.. I'd marry her on the spot - no questions asked - all terms accepted in advance..
Well I think she's married. At least she has a boyfriend with which she has a kid.
Amazing how being a liar for the system gets you mass exposure
Do you not know how use a compressor? This video is quieter than a library... with no people in it!
I'm not sure what is wrong with your sound.
@@HiddenForces Well, it is true that the sound is very quiet, so, the compressor or just increasing of volume of the sound would be better.
.
Hannah is the only intellectual with charm and beauty matching theirs intellect.. I have a hardcore crush on Hannah..
Did he ask this stooge about Haslemere?