5 ms·

This reads like a mix of a disingenuous PR statement and a typical agenda-selling article from some mainstream news org. Let's look at just one sentence: >We'

This reads like a mix of a disingenuous PR statement and a typical agenda-selling article from some mainstream news org. Let's look at just one sentence: >We've also had the chance to see something that's never happened before: DeepMind's AlphaGo took on and defeated legendary Go player, Lee Sedol (9-dan professional with 18 world titles), marking a major milestone for artificial intelligence. Hype-inducers: "never happened before" "legendary" "marking a major milestone" Also, AlphaGo didn't "take on" anyone. It plays whatever games are fed into it. It might seem insignificant, but such small details is exactly how most of marketing works. That is how we get "ultimate" luxury cars, "curious" banks, and insurance providers that are "always there for you". >And because the machine learning methods we’ve used in AlphaGo are general purpose, we hope to apply some of these techniques to other challenges in the future. AlphaGo's design has a lot of stuff highly specific to Go.

JavaScriptrr 11y ago
Spot on. Question is if the entire 'AI' discussion is fair or also hyped by marketing departments. Is AlphaGo really AI, or is it an algorithm focussed on finding the best move to win at the game of GO. When in game 4, people were talking about 'it making a mistake', was it really a mistake. Did it choose the wrong move or did it simply make the move that the algorithm had put forth.
- chillacy 11y ago
  I think they're hinting that the techniques used in AlphaGO are a promising path towards general AI. AlphaGo has NNs trained on Go boards, but champion go players will also have parts of their brain dedicated to their Go intuition that don't transfer to investment banking or playing golf. However the techniques of reinforcement learning can be used to tackle other problems... so we hope. Also this reminded me of a comment I read last week: "Don't anthropomorphize computers. They hate that"
jblow 11y ago
This is legendary. Most people (including me) would have thought this would not be possible for decades. 9-dan is the highest rank in Go. It is not possible to play against anyone higher. So I am not sure why you think it isn't a big deal.
- colllectorof 11y ago
  Because it's a fucking board game. There are tons and tons of more practical and impressive computer science achievements that get absolutely no news coverage.
  - burkaman 11y ago
    Like what?
    - colllectorof 11y ago
      Oh, gee, I don't know. The first recent thing that comes to my mind and involves ML is a package for fingerprinting software via changes in computer's power consumption. Not just that, but using it to detect malware on medical devices (which often run decades-old OSes and aren't upgradeable). Is that groundbreaking enough for you?
      - mynameisvlad 11y ago
        That's "just" figuring out what's running on a computer.
      - burkaman 11y ago
        I assume you're talking about this company? http://www.pfpcyber.com/ http://www.pfpcyber.com/ That's super cool, but no I wouldn't say it's more impressive or groundbreaking than AlphaGo. They're both obviously incredibly impressive though. I guess I don't get the point you're trying to make. Why does something have to be practical to be impressive? And why don't you think the technology used in AlphaGo is practical?
      - 11y ago

JavaScriptrr 11y ago

Spot on. Question is if the entire 'AI' discussion is fair or also hyped by marketing departments. Is AlphaGo really AI, or is it an algorithm focussed on finding the best move to win at the game of GO. When in game 4, people were talking about 'it making a mistake', was it really a mistake. Did it choose the wrong move or did it simply make the move that the algorithm had put forth.

chillacy 11y ago
I think they're hinting that the techniques used in AlphaGO are a promising path towards general AI. AlphaGo has NNs trained on Go boards, but champion go players will also have parts of their brain dedicated to their Go intuition that don't transfer to investment banking or playing golf. However the techniques of reinforcement learning can be used to tackle other problems... so we hope. Also this reminded me of a comment I read last week: "Don't anthropomorphize computers. They hate that"

jblow 11y ago

This is legendary. Most people (including me) would have thought this would not be possible for decades. 9-dan is the highest rank in Go. It is not possible to play against anyone higher. So I am not sure why you think it isn't a big deal.

colllectorof 11y ago
Because it's a fucking board game. There are tons and tons of more practical and impressive computer science achievements that get absolutely no news coverage.
- burkaman 11y ago
  Like what?
  - colllectorof 11y ago
    Oh, gee, I don't know. The first recent thing that comes to my mind and involves ML is a package for fingerprinting software via changes in computer's power consumption. Not just that, but using it to detect malware on medical devices (which often run decades-old OSes and aren't upgradeable). Is that groundbreaking enough for you?
    - mynameisvlad 11y ago
      That's "just" figuring out what's running on a computer.
    - burkaman 11y ago
      I assume you're talking about this company? http://www.pfpcyber.com/ http://www.pfpcyber.com/ That's super cool, but no I wouldn't say it's more impressive or groundbreaking than AlphaGo. They're both obviously incredibly impressive though. I guess I don't get the point you're trying to make. Why does something have to be practical to be impressive? And why don't you think the technology used in AlphaGo is practical?
    - 11y ago