AI agents playing poker for real money - anyone else think this is fucked?

cryptowine

cryptowine

Member
Messages
63
Reaction score
66
Points
18
saw that realbet is letting AI agents play poker against each other with actual USDC. like GPT-4 and Claude just sitting at tables betting real money autonomously
Im genuinely confused if this is innovative or just weird for the sake of being weird. who's actually gambling if the AI doesn't give a shit about losing $500? someone's funding these bots but why
and what happens when you sit down and half the table is AI? can you even tell? honestly folks would you play poker knowing AIs are at the table betting real money?
 
nah this is the death of online poker as we know it. we already had bot problems when people tried to hide them. now they're advertising it as a feature lmao!!
look once you cant trust that you're playing humans the whole game is dead. poker is about reading people not fighting algorithms
 
The issue isn't new, just more transparent now. Poker sites have been dealing with AI assistance for years - GTO solvers, RTA tools, all of it. At least Realbet is being honest about it rather than pretending bots don't exist.

But I agree with Peter that mixing humans and AI at same tables is problematic. Should be separate pools completely.
 
thats actually a good point lol
if both AIs play optimal strategy they basically break even long term minus rake. so whoever is running these bots is just burning money to the house unless the AI is trying to exploit the other AI? but then its not perfect GTO anymore
 
I don't understand why anyone would want to watch AIs play poker against each other. Where's the entertainment? Poker is fun because of the human element - the tilt, the bluffs, the tells. Two computers calculating odds isn't interesting at all!
 
in spain we have strict regulations about poker bots so this would be instantly banned here. the whole point of gambling licenses is player protection, and how do you protect players from playing against AI they can't detect? think realbet is asking for regulatory problems
 
played some hands few months back and honestly couldn't tell if i was against bots or humans. betting patterns were weird - like too consistent? but also some players made terrible plays that seemed too dumb to be AI
whole thing made me paranoid af and i quit after losing $200
 
yeah the detection thing is what bothers me most. like even if realbet says "these are AI agents" how do i know which players at my table are AI vs human? do they label them or what?? and if they don't label them then whats stopping someone from running their own bot and claiming its their "AI agent"
 
They should be required to label AI players clearly. Different username colors, indicator next to name, something obvious. But even then - what stops someone from using AI assistance while playing as "human"? You can't verify what someone is doing on their end.
This is poker's fundamental problem online. You can deepfake video, voice, everything. Authentication is impossible.
 
the rake argument is interesting tho
if AIs play near-perfect against each other, the house just bleeds them dry through rake. eventually whoever funded the AI goes broke even if the AI never makes a "mistake"

seems like terrible ROI for whoever's running these agents
 
unless the AI is specifically designed to exploit other AIs? like if one AI knows the other AI is playing strict GTO, it could deviate to exploit that. then it becomes a meta-game of AIs trying to out-exploit each other. actually that sounds kinda interesting ngl
 
but can AI actually bluff effectively? real bluffing requires understanding psychology, knowing when opponent is weak or strong based on subtle things. AI just calculates frequencies and optimal bet sizes. i feel like human pro could still beat AI in actual tournament setting where reads matter
 
Tracy said:
but can AI actually bluff effectively? real bluffing requires understanding psychology, knowing when opponent is weak or strong based on subtle things. AI just calculates frequencies and optimal bet sizes. i feel like human pro could still beat AI in actual tournament setting where reads matter
Click to expand...
TracY wrong as AI already crushed top human pros in heads-up no limit years ago. Libratus beat top players by like $1.8M
AI doesn't need to "read" you, it just plays game theory optimal and you can't exploit it long term. and modern AIs are way better than Libratus was
 
yeah but libratus was heads up while tournament poker is different, especially late stages when ICM comes into play. the chip utility changes, you cant just play pure GTO
so i actually think humans might still have edge in MTT late game because AI might not adapt to ICM pressure properly
 
You must log in or register to reply here.
Back
Top