This seems to be a major threat to online casinos.
Is there any study on how they plan on fighting the AI's that will likely be flooding the market soon ?
The market has been flooded with them for years. They have various mechanisms for fighting them like mouse detection and scanning your computer for various programs and confiscating your funds if caught. In the end, people will always find a way around the detection methods though.
Like using a chess computer, it seems that there would be no way to protect against someone who relays the cards to a separate machine, and then enters the bids by hand on their own computer.
There are ways to detect cheating on chess websites, though these methods are not perfect. The two main signatures are:
1) Playing moves that the engine recommends consistently. This looks for people who are inhumanly good, even the top chess players in the world regularly deviate from engine suggestions.
2) Time between moves always takes a minimum threshold, the time needed to relay the position to the engine. Humans vary wildly in the time thought based on the position, in a way that cheaters do not.
Neither of these detection methods are perfect, especially for small numbers of games. I don't think we will see large prizes for on-line chess tournaments because of possible cheating.
Somewhat easy to detect a human using a computer, by comparing to what the computer would do, the rating of the player and the usual play of the player.
However, a chess professional would be able to use a chess computer almost undetected, as they know what would look like "computer moves", but could use it to just choose the best of their own suggested moves.
Heads-up is a contrivedly simple scenario compared to solving poker in general. There have been bots for a long time that have co-existed with real players, and we're likely not close to a scenario where bots can reliably beat real players in general for real world scenarios.
However, in theory even a break-even bot could be a gold mine because it can earn loyalty points. That's how it was the last time I was playing online, I don't know if loyalty programs still work the way they used to. Of those programs would just go away if abused enough.
Online poker has the concept of "bonus whoring" where one takes advantage of various loss-leader promotions offered by different sites.
A typical promotion would be offering a 20% bonus on a $1000 deposit, with the caveat that you must play in, say, 1000 raked games before cashing out. It's possible to play in games where the total expected rake if the user breaks even is, say, $100, leaving $100 profit.
Why do you say we're not close to solving Poker in real-world scenarios? The multiplayer version is likely to be harder, but is it so hard, that it'll take decades more?
Depends what you're trying to achieve. If it's a bot that beats most humans, that probably already can be done. If it's a bot that earns more from a cash ring game than the world's best would against all levels of players...
It took many years for a computer to bridge the gap between better at chess than almost all humans and better than the best human.
When they released the initial paper a few months ago I looked into it and it sounds like this wont be profitable anywhere except the highest levels (although obviously the highest costs are in the training phase).
This in combination with casinos already doing a lot for bot-detection makes me think that we are at least a few years away until this is an issue.
