Hacker News Reader

Comment by arcanus | Hacker News Reader

Missing the biggest triumph of 2016 by far: the victory by DeepMind's machine learning algorith at Go over the greatest living human player.

At this time last year it was highly questionable if humans would ever lose to algorithms at games as complex as Go.

Now we have reason to be optimistic that an entire class of problems are amenable to solution through computation, and more evidence super-human synthetic intelligence is indeed possible.

View on HN

...and hundreds of millions worldwide who have been saved from malaria or lifted out of extreme poverty would likely disagree with you: what good is super-human synthetic intelligence several decades from now (assuming that's even a tractable generalization of largely task-specific ML models) if you're not sure where your next meal comes from?

Even in the best of circumstances, with unlimited food for all, those people will be dust in a century. Even incremental improvements in technology will help hundreds of generations of humans going forward.

Not all uses of technology ore a positive influence. HFT, NSA's surveillance are two that come to mind straight away.

I don't want this to come across as if I support these particular programs, but I would imagine some tech comes out through veterans and takes on consumer applications.

Does the NSA file patents? If not, usage of tech is probably enforced by the honor code

"Does the NSA file patents?"

They do actually. They also get defense contractors like SAIC's holding company to do it for them then sue groups doing things like encrypted messaging. They can also slap secrecy orders on patents & block those from coming to market with that sort of thing going up since Bush-Cheney.

"usage of tech is probably enforced by the honor code"

See Snowden leaks. Defense contracting debacles. PROMIS scandal. So on and so forth.

> if you're not sure where your next meal comes from?

This is certainly work in progress, but:

"31. In 1990, more than 60% of people in East Asia lived in extreme poverty. As of 2016, that proportion has dropped to 3.5%."

I'm calling BS. East Asia basically just means China, because it dwarfs other populations. IMHO >60% of Chinese were certainly not living in extreme poverty in 1990, unless economically defined via free market rhetoric and bad statistics. The reality is that people used to grow more of their own food and buy less, so the need for free-market capital was limited. Perhaps this is the source of the mistake.

There was and still is far more extreme poverty in South Asia.

This actually sounds interesting as a fact but I would like to see some evidence/links.

Just look for photos of any place in China in the 1990s, or ask someone who lived there.

Pretty much exactly this. Global reductions in poverty equate to reductions in $2/day income threshold in two countries: China and India.

http://www.business-standard.com/article/pti-stories/china-i...

Accomplished at the cost of tremendously increasing resource consumption, particularly energy:

China: https://m.imgur.com/Vj5VwS9?r

India: https://m.imgur.com/elFJjGP?r

So in your opinion what percentage of Chinese were living in extreme poverty in 1990?

Well, I founded a startup that's using ML to help turn satellite data into credit risk profiles for the world's poorest farmers so we can sell them fertilizer and seed.

So... I think tensor flow is going to be pretty great for the 1K farmers we're working with this year and ~50k we're going to try and reach next year.

This is one of those ideas where I think...how on earth did you come up with the idea of building such a company? Sidenote: sounds absolutely amazing! Congrats!

> help turn satellite data into credit risk profiles for the world's poorest farmers so we can sell them fertilizer and seed

Not sure if satire or serious :].

Completely serious, we're alumni of The Climate Corporation.

Satellite data is cheap or in many cases free, and the cost of sending a trained loan officer out to the field is far too high to justify.

Shoot me an email at earl@apolloagriculture.com and I'd be happy to chat more.

Is it? I mean, a lot of hard work was put into it. And, it's a step that needs to be taken. And people can work on what they want to work on. But how can a machine beating the best human at a game really be the biggest triumph of 2016?

Whenever I think about this stuff, I have a similar nagging question. E.g. if there are entire classes of problems that we now "know" are solvable, and have an actual benefit to mankind, why didn't they spend time on that rather than a game?

This is like asking, "why didn't they just build a new dam instead of spending so much time drawing up blueprints?" Games can be useful abstractions and guides that advance our understanding of the same decision making processes that allow people to create real, effective changes in the world. The balance is shifting to the point in the near future where most of the best solutions to real world problems won't come directly from unaugmented human minds.

I can see games being an impetus, but are they really the blueprint, as opposed to just another application, but one that the public at large can appreciate?

The game itself isn't the blueprint. The source code of the application used to play the game is a blueprint for learning and pattern recognition that can be applied to other things. For example, the DeepMind AI used to play Go has been adapted to diagnose kidney injury by reusing much of the same code which was developed for the game.

https://www.engadget.com/2016/11/22/nhs-deepmind-ai-app/

To me, the biggest events of the year are probably Trump, Brexit, and AlphaGo.

Author seemed to be going to great lengths to get things that were very positive, or at least very difficult to argue as a negative.

Depending on one's political leanings, Trump, Brexit and AlphaGo are all positive things.

There's a difference between those, and the eradication of diseases/increased nature preservation/reduced world hunger/increased literacy.

Basically the post is going way out of its way to show a selection of the good stuff (based on what the author usually covers) that is difficult for the Media to get lots of angry shares based on a polarizing title.

There is no reason to think this is a positive development for humanity. Could very likely be a step on the road to our extinction.

Reply on news.ycombinator.com