Hacker News Reader

Comment by visarga | Hacker News Reader

Shifting goalposts. And humans cost much more energy than the watts contained in their food. Humans take 20 years to "train" and need lots of water, electricity, food, clothes, transportation and teaching time.

If they were to use only the policy net part of AlphaGo they would need a few milliseconds per move and only one computer to run it, achieving the level of 1p. Also, if DeepMind used specialized Go hardware maybe they could reduce the power usage a lot, but they didn't optimize for that.

For example, it would be interesting to calculate how much power is used in image recognition, man vs machine. The brain uses 20-40W and a cell phone only 5W, and we know cell phones can do image recognition, so, probably the computer uses less than the human given that human reaction time is 200ms and the computer could classify much more than 5 images per second, so it uses less time and consequently less energy per image.

View on HN

>The brain uses 20-40W and a cell phone only 5W, and we know cell phones can do image recognition, so, probably the computer uses less than the human given that human reaction time is 200ms and the computer could classify much more than 5 images per second, so it uses less time and consequently less energy per image.

I'm all for praising the enormous developments being made in modern computing, but this is absolutely needless exaggeration. Computers are far, far, far weaker at image recognition than even an 8 year old child. The amount of things humans can continually process, using a fraction of the brain (the brain is constantly doing visual processing without shutting the other centers down), whereas a computer can barely do simple tasks and takes longer to do that than "200ms", as you claim.

The brain is a mind boggling machine. Our computers are getting pretty kickass, but there is still a long long way to go.

I mean, you could keep counting all the energy used in the causal chain as far back as you want. After all, AlphaGo was created by humans whose brains used a heck of a lot of energy.

I still think that looking at marginal energy cost while playing the game is an interesting approach.

AlphaGo is only two years old and is one of a handful of computer AI programs, but humans have had millions of years of evolution. So it's not fair to compare. Computers have evolved too, maybe even more dramatically than biological life, to lower and lower levels of power usage.

In Chess no top player can beat his/her cell phone. This milestone will be eventually achieved in Go too.

> Humans take 20 years to "train"

Just because we're doing things that way doesn't mean there's a requirement to do so.

I read that when the British government wanted people to translate Japanese for them in WWII, they went to their universities and heard, from the relevant departments, that becoming sufficiently conversant in Japanese was a painstaking, multi-year process. Someone else heard about this, thought "pffft, I could train people to be conversant in Japanese in six months", and set up a school. He was right, the university departments were wrong.

You see so many (justified!) complaints that modern schooling is composed mostly of wasted time. Humans don't take 20 years to train.

Train for a specific task? No. But train to be able to acquire new skills, yes. For example, I have explained the basics of Go to my 5yr-old, but he will need years to be able to develop sufficient abstract thinking, patience, etc. to actually play.

Schooling is not primarily intended for academic training, at least until post-secondary levels are reached. Instead, it's primarily about socializing, instilling rule-following, reinforcing social roles, and so on.

> Shifting goalposts.

No, new challenge.

I agree it would be a good challenge, but I disagree that it would be "fairer" and "much more important" than AlphaGo itself.

The fact that DeepMind accomplished this feat, period, is what's most important. The fact that it runs on Google Cloud Platform, not custom-built hardware (à la Deep Blue), is almost as important. When it inevitably shrinks down, that would be icing on the cake.

I mean, I think he's right that it'd be a great achievement to be able to do this with standard hardware and not a purpose-built supercomputer.

IBM's Deep Blue used custom hardware. DeepMind's AlphaGo used Google Cloud Platform: http://googleresearch.blogspot.com/2016/01/alphago-mastering...

It still uses a lot of hardware, but not hardware built for the sole purpose of playing Go.

How much energy do you think went into "training" alphago? Not just electricity and server time to train it's neural networks, but the development team too.

Reply on news.ycombinator.com