On: AlphaGo shows its true strength in 3rd victory against Lee Sedol

[lukaslalinsky]

They challenged a top player for $1M. Doesn't that imply they think it's stronger than any human player?

[Bud → lukaslalinsky]

It implies that it costs money to get a high-profile match against someone like Lee Sedol, and it costs money to draw public attention, and it implies that they thought they at least had a decent chance. Nothing more.

[wolfgke → lukaslalinsky]

This can also imply that for Google it is worth $1M to know where AlphaGo still has weaknesses.

[iainmerrick → wolfgke]

Don't be silly, a Go AI has very little direct value for Google's business. But the publicity is great, easily worth $1M. More important is keeping the DeepMind team busy and engaged -- the real value is in spinning that know-how off into other projects.

[wolfgke → iainmerrick]

> Don't be silly, a Go AI has very little direct value for Google's business.

On the other hand: Knowing what makes some problems much easier for humans than for computers can have direct value on Google's business. Go was a hot candidate for a game with this property.

[gus_massa → lukaslalinsky]

I think they expected that the computer to play a good game, but probably they were not sure about the outcome. Anyway, Google has Billons in the bank, so if they were defeated they would not miss that single Million too much.

[tim333 → lukaslalinsky]

The $1M gets paid out by Google whether AlphaGo wins or loses. I just goes to charity if it wins.

[fma → lukaslalinsky]

How much money do companies spend on Super Bowl ad for a 30 seconds of publicity.

How much publicity has Google received for $1 million? (Much less than that actually, since they won)