That's amazing! I've got the same combination on my luggage!

Might be a reasonable position except you know this guy is 115% against loosening zoning regulation.

How to get away with murder (obviously this is Florida):

https://www.nytimes.com/2017/10/18/u...onal&smtyp=cur

How is that not manslaughter? I get the lack of intent, but come on.

That AlphaGo Zero breakthrough is pretty amazing, remembering that AlphaGo itself was considered a huge breakthrough. After defeating the original AlphaGo 100-0 after training for 3 days, it went on to train for a further 37 days and is now EXTREMELY strong, like relatively much stronger against top humans than the top chess engines are. The top human player is rated 3667 and AlphaGo Zero's ELO is in excess of 5000 now, that advantage is roughly on the order of Magnus Carlsen's advantage against a bad casual player.

Lee Se-dol thought he failed when taking on AlphaGo last year, but in fact his one win that match will be remembered as the last time a human managed to win a game against the best available computer opponent. In the space of two years we moved from Go being a difficult game for computers to play, to being a game where computers are so strong that humans have no chance to ever win a game.

I'd like to understand what "merging policy sets with value sets" means. Like I can understand that big blue just tries every possibly future move combination during each move (with priority tweaking for strategy I'm assuming).

But obviously if that worked for Go they'd have done it a long time ago. What's different about how AI approaches the problem?

I got pretty into a dumbed-down video game version kind of like Go called Ataxx. It was different than a lot of games in that you had to keep pumping in more quarters for the time you used on your turn only. So you had a huge incentive to act fast. Being a broke college student helped with the motivation. But you also lost your quarter if you lost to the ever toughening bad guys. So I kind of have a vague idea of the strategy concepts. Maybe.

Computers are ruining board games.

Poker is next

Obviously bots are already a huge problem online; the thought of a future when you can't tell which of your live opponents has hidden SuperPokerAI helping them is depressing

You keep the computer in a box, you don't give it a shark's head that shoots laser beams.

Like the supercomputers in the Hitchhiker's Trilogy.

yeah well someday someone is going to give a learning ai a task like "make me rich" then let it loose on the internet and then we're all ****ed.

Nah, the superAIs are gonna kill themselves when they find out how stupid we are.

I would be surprised if people in the finance industry ar not already doing this.

Person: Super-universe-brain-AI-computer, solve world hunger and poverty.

SUBAIC: Ok, how about the people with all the stuff give some of their stuff to the people that don't have any stuff?

Person: Yeah, uh, we're not gonna do that.

SUBAIC: ...

They are, but it's "AI" not generalAI.

SUBAIC: Here, I developed a pill that will cure and prevent any and all disease.

Person: OK, you didn't understand the question. I asked how to fix obamacare so the premiums don't go up.

SUBAIC: ...

Cool!

Go is a game that's more about pattern recognition. The number of possible moves is exponentially greater than chess, so doing a brute force calculation of every move tree isn't feasible. There's big demand for this kind of problem solving for things like facial recognition and getting computers to read handwriting and so on.

Only in the same way that cars ruined foot races. Now there's a lively competition afoot to make the best chess/go AI. And ofc human vs human matches are still alive and well.

There are tool assisted speed runs that complete video games in a matter of seconds, but humans playing through speed runs is still the majority of what people give a crap about. Go figure

No one actually cares about speed running video games.

My incredibly high-level and uninformed understanding is that original AlphaGo basically had two parts. There was a "policy network," which was originally trained on a huge corpus of human games, then refined further with self-play, and whose function was to pick out promising move candidates. The value network was (I think) totally self-trained and was used to decide when the search had arrived at a good outcome and stop looking. So basically, the AI would take current game state, ask the policy network for some likely-looking moves, then see if making any of those moves would put it in a great or terrible position, and, if not, iterate the process again until it runs out of time or finds a winner. I don't really understand what it means either to merge those two together.

Here's a great article about the old, obsolete version of AlphaGo:

https://www.wired.com/2016/05/google-alpha-go-ai/

Keep in mind that thousands, and eventually millions, of humans have been developing Go strategies for thousands of years. That was all a dead end. Asking a human to play go is like asking a horse to be your financial advisor. The only way to get a competent approach is to hand it wholly over to the AI.

It's horrifying that people are allowed to do this **** without any government oversight. This kind of research needs to be locked the **** down and only conducted in high-security government labs.