It looks like you're using an Ad Blocker.
Please white-list or disable AboveTopSecret.com in your ad-blocking tool.
Thank you.
Some features of ATS will be disabled while you continue to use an ad-blocker.
...the group will work toward developing standards and ethics around the development and implementation of AI.
originally posted by: Aazadan
a reply to: soficrow
A Utopian society is hundreds of years off, but it's still what everything is building towards.
Automation... is done to increase productivity and reduce working hours.
Also, people mostly make good choices. On the individual level a lot of people get a lot of things wrong, but crowds are mostly capable of getting it right.
Poker now joins chess, Jeopardy, go, and many other games at which programs outplay people. But poker is different from all the others in one big way: players have to guess based on partial, or "imperfect" information.
...
...no matter how impressive your card skills, your power of prediction is limited by Texas Hold 'em because only certain cards are visible. An ace in your hand and an ace on the table might look like good news, unless your opponent is holding a pair of aces, or worse. No amount of computational power can answer that question.
...
“The typical approach to addressing perfect-information games [like chess] was to search through a game tree to find an optimal path,” he continues. But that’s no good in poker, because without knowing what cards are where, you can’t even figure out where on the tree you are.
Libratus is a whole new kind of machine.
Rather than try to sort through an unknowable tree, it focuses on finding a favorable move that represents a so-called “Nash equilibrium” solution...
You won’t always end up in the best possible situation, but you may be able to avoid the worst outcome.
...
Libratus plays poker in a similar way, never guaranteed to win any particular hand but likely to stay in the black over the long run.
...
People tend to bet and bluff in certain increments, and noticing those patterns helps professionals find an edge, but the computer was hard to read.
...
As Libratus was analyzing the day’s games each night, the pros did the same. While they weren’t able to find a consistent winning strategy, they suggest the experience made them better poker players. “Once you face Libratus, there's nothing worse any human could ever do to you. Every human is going to seem like a walk in the park,” said Jason Les, another one of the players.
...
The imperfect information nature of poker makes the win a huge achievement for the AI community, with far-reaching real world applications that Brown says include negotiations, auctions, and security interactions, to name a few. “In truth, most real-world scenarios involve hidden information. In the real world, not all the information is laid out neatly for all sides to see like pieces on a chessboard. There is uncertainty and deception,” he explains.
One remaining enclave of human superiority is multiplayer situations, however. Liberatus prefers bilateral dealing to group negotiation. It took on each poker player one-at-a-time, and couldn't have won in an ensemble game. Still, two-sided negotiations are common in the real world, and Libratus lays the foundations for computer programs that help negotiators elevate the art of the deal to a science.
originally posted by: soficrow
So you're not a big believer in the mob mentality. Nor obviously, in a fully informed populace.
originally posted by: AceWombat04
Pertinent:
When we finally do crack - or accidentally end up creating, or allowing to emerge - more generalized AI, it's likely something like all of these different capabilities will end up rolled into it. And said capabilities are advancing much faster already than anyone thought they would.
originally posted by: AceWombat04
I'm fairly certain this still counts as a breakthrough ...
originally posted by: Ohanka
This thread just makes me wish I'd done software development at university instead of history. Bah.
originally posted by: BrianFlanders
As I always say....what could possibly go wrong? A code of ethics is going to be developed by people who have none.
As I always say....what could possibly go wrong? A code of ethics is going to be developed by people who have none.
originally posted by: soficrow
a reply to: LogicalGraphitti
Ai can already out-bluff humans. ...What do you think the implications of that are?