I am puzzled by all the hype (including by Google itself) around the Google Go Bot beating the world’s best human Go player


I am puzzled by all the hype (including by Google itself) around the Google Go Bot beating the world’s best human Go player.
Did we care as much when a computer could do mathematical operations faster than a human or a database could hold much more structured information than a human brain?
A much more interesting test would be whether the Google Go Bot could beat a machine-aided human player. That seems like a much more realistic setup in terms of how humans leverage computational capability.
More broadly speaking, this is my big issue with how we have been talking about AI over the last couple of years. It feels like a lot of the positioning is about humans vs. computers as opposed to humans+computers.


Christian Bateman
December 28, 2017

Sounds like a op-ed to me

Joe Ho
December 28, 2017

Agreed. AI is to augment Humans, not replace.

Yishan Wong
December 28, 2017

So stage an exhibition chess match or something.

    Aditya Agarwal
    June 8, 2017

    Sure, I would be happy to take on a machine+human vs. machine at what I do every day (eng management) 🙂

    Yishan Wong
    June 8, 2017

    No, I didn’t mean it like that. I mean just find a human chess master and see if they want to play a match like that that you set up.

    Aditya Agarwal
    June 8, 2017

    Ah ok. But I am not in the business of proving this. I just have the privilege of making observations :). Imagine if you had to prove every hypothesis that you post on FB!

    Daniel Chai
    June 8, 2017

    Kasparov gave a recent TED talk where he says roughly the same thing (the future is AI-assisted humans) and indicates that machine-aided chess competitions are already a thing. https://www.ted.com/…/garry_kasparov_don…/transcript…

    Anand Manikutty
    June 8, 2017

    Agreed.

    What should be added to this analysis: we have already more or less accepted that homo sapiens can no longer compete against A.I. in some realms. We are not even competitive any more. Now, the question is– how competitive are we in other realms?

    It is already the case that A.I. has won the battle in entire industries. What about other industries?

    What about automating the job of the managerial class in Industry X, Y and Z? Does the defeat of the world’s top Go player imply other capabilities of A.I. that should cause concern?

    I will post more on my own Facebook page.

Aditya Agarwal
December 29, 2017

Sure, I would be happy to take on a machine+human vs. machine at what I do every day (eng management) 🙂

Yishan Wong
December 29, 2017

No, I didn’t mean it like that. I mean just find a human chess master and see if they want to play a match like that that you set up.

Aditya Agarwal
December 29, 2017

Ah ok. But I am not in the business of proving this. I just have the privilege of making observations :). Imagine if you had to prove every hypothesis that you post on FB!

Daniel Chai
December 30, 2017

Kasparov gave a recent TED talk where he says roughly the same thing (the future is AI-assisted humans) and indicates that machine-aided chess competitions are already a thing. https://www.ted.com/…/garry_kasparov_don…/transcript…

Anand Manikutty
December 31, 2017

Agreed.

What should be added to this analysis: we have already more or less accepted that homo sapiens can no longer compete against A.I. in some realms. We are not even competitive any more. Now, the question is– how competitive are we in other realms?

It is already the case that A.I. has won the battle in entire industries. What about other industries?

What about automating the job of the managerial class in Industry X, Y and Z? Does the defeat of the world’s top Go player imply other capabilities of A.I. that should cause concern?

I will post more on my own Facebook page.

Molly McKinnon
December 28, 2017

Humans have a hard time getting along with themselves let alone things that can best them.

Joshua Goldbard
December 28, 2017

Humans + computers still beat computers at chess if I’m not mistaken. I believe the same will hold true for go for the foreseeable future.

Erik Klein
December 31, 2017

do you know how the human works with the computer? I wouldn’t imagine it would be funky if it was consensus of the move..

Joshua Goldbard
December 31, 2017

Erik: https://en.wikipedia.org/…/Human%E2%80%93computer_chess…

Hooman Radfar
December 29, 2017

Couldn’t agree more

Ali Partovi
December 29, 2017

Augmented Intelligence > Artificial Intelligence

Chris Varenhorst
December 30, 2017

I’d say this is hyped because of Go’s historical context in AI. Go has been a classic and well-known example of the type of problem we don’t know how to make programs that are better at than people, even though on the surface it sounds like something we could do. It used to always be a really clear example of how the mind computes in a way we don’t know how to duplicate in software.

Though agreed that at this point we shouldn’t be surprised by any further successes. Given how good Alpha Go is at this point, my uneducated guess is that a top Go player with an Alpha Go assistant wouldn’t be any better than pure Alpha Go but it’d certainly be interesting if I’m wrong!

    Aditya Agarwal
    June 8, 2017

    Ah, that’s good context that Go was one of the white whales of AI folks.

Aditya Agarwal
December 30, 2017

Ah, that’s good context that Go was one of the white whales of AI folks.

Kayur Patel
December 30, 2017

I don’t know if human+machine is a *much more* interesting test. Game playing is a problem that people can relate to. And Go is a hard game for computers to play well — a long standing problem. Now it’s a milestone that we passed.

Human+machine is just another milestone for progress.

    Aditya Agarwal
    June 8, 2017

    I get that it’s an interesting milestone to pass but the coverage around it is so disproportional to its significance per my post.

    At the very least human+machine vs machine seems much more realistic about how we use technology.

    Kayur Patel
    June 8, 2017

    Yeah, you don’t need to sell me on human+machine. I spent way too much time thinking about that problem. 🙂

    The ML hype is deafening, I agree. But it’s kinda everywhere. Every keynote seems to mention ML a ton. Go may be too hyped.

    I compare it to the Grand Challenge. Setting a goal focuses work and the process generates technology that folks can use. The PR helps get the public interested.

Aditya Agarwal
December 30, 2017

I get that it’s an interesting milestone to pass but the coverage around it is so disproportional to its significance per my post.

At the very least human+machine vs machine seems much more realistic about how we use technology.

Kayur Patel
December 31, 2017

Yeah, you don’t need to sell me on human+machine. I spent way too much time thinking about that problem. 🙂

The ML hype is deafening, I agree. But it’s kinda everywhere. Every keynote seems to mention ML a ton. Go may be too hyped.

I compare it to the Grand Challenge. Setting a goal focuses work and the process generates technology that folks can use. The PR helps get the public interested.