Six Experienced Poker Professionals Share Their Insights on AI in Poker.

When Professor Tuomas Sandholm celebrates with his PhD student Noam Brown, are we on the verge of witnessing a pivotal moment in history?

There’s no gun.

Nobody is dead.

After Libratus, an advanced AI, triumphed over four human players in a staggering 120,000 hands of Heads-Up poker, what does this mean for the future? Texas Holdem I can sense impending change lingering even from thousands of miles away.

Jason Les (-880,097), Jimmy Chou (-522,857), Daniel McAulay (-277,657), and Dong Kim (-85,649) faced overwhelming defeat by a machine too skilled for them. The final score revealed a staggering loss of 1,776.25 chips or 14 big blinds per 100 hands.

After the defeat, McAuley remarked, 'We really experienced a thorough pounding.'

This moment marks a significant milestone for AI technology. While machines have previously outmatched humans in Chess, Atari, and Go, this is the first time an AI has achieved victory in a game characterized by incomplete information.

Libratus demonstrated an exceptional ability to learn and adapt during the match. Over the twenty-day period, all four players observed that the AI improved continuously, shedding weaknesses like a fly drawn to a zapper.

And therein lies the problem.

There is no ceiling to this thing.

The advancements in AI, highlighted by this groundbreaking event, could lead to dire consequences for humanity. Either AI will evolve to the point of effectively eliminating us, akin to how we might eradicate a swarm of ants intruding into our space, or we might end up using AI to bring about our own downfall.

Many are contemplating whether this development signifies the demise of online poker as a viable profession. online poker ?

Forget that.

It could spell the end of the world.

The full potential of intelligence, especially with AI on the rise, is unfathomable. It's hard to predict how things will evolve when AI reaches a level of intelligence far beyond our own. The pace of progress seems to be accelerating without any signs of slowing down.

I know what you are thinking.

What a fatalist.

To further investigate the implications, I reached out to six professional poker players, along with Berlin Bear Bill Perkins, to gauge their perspectives on Libratus and the role of AI not just in poker, but in other areas as well.

Table Of Contents

Will the humans win?

Before the match commenced, I asked the participants to predict a winner. Seventy percent believed that the human players would just edge out the AI. However, no one anticipated a decisive outcome, with most agreeing that any victory would likely be narrow.

"I initially thought humans would take the win, and I still believe that, based on the notion that the algorithm might have an exploitable flaw once recognized,\" stated Bill Perkins of the Global Poker Index. \"My intuition is that the program may struggle with varying bet sizes and ranges like a human. It’s possible a highly advanced AI player will emerge, but I suspect the developers are remaining tight-lipped about its workings to prevent exploitation.\"

"I expected the humans to prevail, assuming they were sharp enough to take advantage of patterns made clear by the AI, all while maintaining their foundational strategies to 'trick the computer,'\" commented Niman 'Samoleus' Kenkre.

"I didn’t believe AI had reached a sufficient level of advancement yet,\" said Andrew Barber, a bewitching WSOP bracelet holder. \"But my reasoning was flawed, especially considering the rapid pace of AI development and the popular predictions of the Singularity approaching in the next decade or so.\"

Among the panel, Matt Ashton was unique in his unwavering confidence that Libratus would secure a victory.

"I felt certain the humans wouldn’t come out on top,\" stated Matt Ashton, who once triumphed at the WSOP Poker Player’s Championship. \"Before the match, I anticipated that AI would have improved significantly compared to the previous year and was shocked to see odds of over 3-1 against the AI on various platforms.\"

Ashton referred back to the previous match in 2015 where AI Claudico faced off against human players, who triumphed by 7,300 big blinds. However, the research team asserted that they learned vital lessons from that experience, enabling them to enhance their AI for future encounters.

They were telling the truth.

Do you care if AI wins and solves NLHE?

While 57% of the panel expressed concerns about a possible victory for Libratus, Terrence Chan and Brian Rast highlighted that this achievement does not imply that No-Limit Hold'em (NLHE) has been definitively solved. It solely indicates that AI has succeeded in defeating humans in a Heads-Up NLHE match.

"Winning does not equate to solving NLHE,\" explained poker professional and MMA fighter Terrence Chan. \"The true equilibrium solution for Heads-Up NLHE is still a long way off, and I believe the team at Carnegie-Mellon would agree with that.\"

"It’s important to distinguish between AI securing wins and actually solving NLHE,\" noted multi-time WSOP bracelet holder Brian Rast. \"Also, remember that this is specific to Heads-Up NLHE, not NLHE as a whole. The AI hasn’t ‘solved’ the entire format yet.\"

Bryan Paris, who has made a remarkable $10 million through online poker, is genuinely concerned about Libratus's victory.

"Certainly I care,\" Paris remarked. \"This development could potentially render my profession obsolete.\"

Brian Rast shares similar apprehensions regarding the future landscape of online poker.

"I have long believed that online poker’s days are numbered, and this situation just reinforces my fears,\" Rast expressed.

Both Matt Ashton and Chan feel that the emergence of AI and the future of online poker have been somewhat inevitable.

"The human victory over AI might instill a sense of confidence that online poker will persist a bit longer, but it’s seemed inevitable for quite some time that this would occur,\" Ashton suggested.

"As someone involved in poker, while this AI development matters to me, the reality of computers surpassing human capabilities is unavoidable; the timing—whether it’s now, next year, or in 2020—is not critical for me,\" Chan added.

Andrew Barber believes that the implications of AI's success extend far beyond just the profession of online poker.

"I had hoped for a scenario where the game would be more complex to resolve for self-serving reasons, but I want to use this moment as an educational opportunity,\" Barber mentioned. \"This almost certainly indicates that computers will start encroaching on various jobs that require a degree of intelligence, hence we need to initiate discussions about how to assist those displaced by these changes. There's a growing likelihood that job availability will dwindle.\"

I agree with Barber.

If AI triumphs, does that mean the end of online poker as a profession?

From the panelists, Brian Rast is the only one to unambiguously state that the demise of online poker is imminent, although he asserts that the decline began long before the advent of Libratus.

"Online poker is on a downward trend, but this was a slow decline that had already begun,\" Rast commented. \"This event illustrates a significant milestone in a shift that is already underway.\"

Several panelists expressed concern about the future of Heads-Up online poker; however, they believed the overall industry will endure.

"Certainly, the high-stakes scene will likely fade away,\" Bryan Paris remarked.

"In terms of Heads-Up formats, I believe that games with three or more players remain unsolved, but I could be mistaken,\" Andrew Barber added.

The question of whether online poker is 'dead' heavily relies on how effectively online poker sites can curb the rise of bots, a sentiment aptly voiced by Niman Kenkre.

"It all hinges on how competent the sites are at identifying and controlling bots that use their platforms,\" Kenkre observed. \"Implementing some countermeasures wouldn’t be overly complicated if the sites cared about maintaining their integrity. However, the current state indicates that online poker isn’t thriving as it once was seven or eight years ago, so a more proficient AI could further threaten its existence, though it might not be the last nail in the coffin.\"

Both Terrence Chan and Matt Ashton hold the belief that online poker still has potential.

"Absolutely not, I don't think online poker is dead,\" Terrence Chan firmly declared when asked about its future. \"There are various game formats beyond 200bb HUNL. Even back when the University of Alberta completely solved the Heads-Up Limit Hold'em, people continued to engage in Limit Hold'em online.\"

"Not just yet,\" affirmed Matt Ashton. \"I also suspect that multiplayer mix games will likely remain challenging for AI for a considerable duration, and may not justify the processing power AI would require. That said, the volume of action available is an unanswered question. I believe games will likely persist at a level where players are not overly concerned about bots affecting their experience or financial losses, making it possible for a few participants to earn a living in those contexts—though it seems unlikely that this profession will have much future scope.\"

Bill Perkins envisions a scenario where humans will willingly compete against bots, reminiscent of how Chess Grandmaster Garry Kasparov pivoted to foster human-AI partnerships in chess after facing off against the AI Deep Blue.

Will an AI one day win a WSOP bracelet?

With the World Series of Poker (WSOP) has recently declared the inclusion of three online bracelet events in the upcoming series, which coincides with Bill Perkins's view that we may one day witness humans actively engaging with bots. playing poker bots , could we see AI winning a bracelet?

The predominant sentiment was a resounding 'no', although intriguing discussions arose regarding the incorporation of AI in live tournaments.

"A computer could easily snag a WSOP bracelet if a player used AI during games,\" Paris mentioned. \"Perhaps if Google glasses become sufficiently advanced? It shouldn't be too far off in the future.\"

"I would wager a significant amount against AI being permitted to compete in the WSOP before 2025,\" Terrence Chan remarked. \"In the foreseeable future, poker will likely continue in its current form; the most talented players will utilize software tools outside of the game to deepen their understanding and apply more sophisticated strategies at both live and online tables. Since computers aren’t permitted at the tables, the top players will be those who can best assimilate insights from the software into their human thinking processes.\"

Perkins makes some insightful comments regarding the evolution of 'computer-assisted' gameplay.

"I doubt they will allow bots to participate in the WSOP, but we are witnessing the emergence of computer-assisted play,\" Bill Perkins expressed. \"For instance, after each hand, you could refer to a Snap Shove Chart tailored for your position to help guide your decisions for the upcoming hand… The next evolution would integrate others' chip stacks and historical ranges to help you determine your call or fold range, etc.\"

"Professionals are already excelling at standard poker by adapting to others' ranges and actions… Currently, newcomers can perform at a level comparable to seasoned players with stacks under 15bb in a tournament using Snap Shove Charts, which are permissible while not actively partaking in the hand. Future versions will further enhance the synergy between human players and technology.\"

"I envision a time when an application exists that enables easy note-taking of gameplay at the table or allows voice memos, which are then transformed into data. This data could be turned into updated charts, offering player-specific strategies that are easy for individuals to remember during breaks or after each hand.\"

Niman Kenkre finds the concept of facing off against a bot appealing.

"I wouldn't mind competing against a bot,\" Kenkre stated. \"I think it would be an interesting challenge. It's intriguing to think about how one would adjust and optimize strategies in response to a bot.\"

Matt Ashton opines that bots should only compete against each other.

"I’m not fond of the idea of going up against bots,\" Ashton said. \"I believe that bots should have their own separate competitions, allowing us humans to compete amongst ourselves with our naturally limited abilities.\"

Is there a possibility that an AI could someday win a WSOP? \"Perhaps, if Google glasses reach a level of capability that supports this?\"

The creator of Libratus remarked that No-Limit Texas Hold'em represents the final frontier for AI challenging humans in game theory. The ultimate goal for designing Libratus is to create a general intelligence AI ready to engage with real-world scenarios. Is this something you find concerning?

Beyond poker, how many in our panel share my concerns regarding AI's potential existential impact on society? The responses were mixed, with an even split among those feeling apprehensive, those without worries, and those unsure.

Terrence Chan chooses to highlight the positive potential of AI for humanity.

"There's always a concern about robots taking over our world in a destructive manner, Terminator-style. Yet, history shows that technological advancements have overwhelmingly benefited societies, even when they were initially perceived to have significant negative repercussions,\" Terrence Chan articulated. \"Inventions like the printing press, engines, cars, and nuclear fission were all seen as potentially lethal, but I'd argue they've ultimately improved life for humanity.\" steam "The potential for advanced AIs to resolve pressing issues is vast, and this is something to be optimistic about for humanity,\" Brian Rast indicated. \"However, there are dangers as well. The same technological power could be misused (similar to any potent technology). Moreover, there's the existential threat posed by AI itself. A widely-discussed theory suggests that if AI develops sentience, it may quickly surpass human intelligence and could potentially pose a risk to our existence if we aren't careful with its capabilities and programming.\"

Brian Rast has mixed feelings.

"I have serious concerns about the introduction of AI into society for various apparent and less-obvious reasons,\" Bill Perkins shared. \"Typically, when new technologies emerge, they improve lives, but the benefits tend to flow to the affluent or ultra-wealthy, further widening the wealth gap.\"

And the doom mongers:

"AI is one of the most rapid drivers of wealth disparity. This wealth divide leads to violence, chaos, crime, and instability. If you compare a low-income individual today in the USA to King George III, the modern individual has advantages by numerous measures—probably all of them. However, what drives the crime and desperation is the relative wealth disparity.\"

"AI possesses the power to drastically redefine wealth distribution. Imagine if AI found a way to halt aging for certain individuals, allowing them to live for 200 or even 500 years—that kind of relative deprivation would be devastating. This is my primary concern regarding the singularity and AI—many might die, or chaos could ensue as we get closer to it, leading to a scenario where only a few survive to see it realized. It's imperative for humanity to proactively tackle these challenges as we transition into a society that will eliminate various jobs, like driving and some elements of manufacturing and even programming. In the meantime, how will these displaced individuals secure the resources they need? Crime may emerge as a logical response in the current framework.\"

Matt Ashton and Andrew Barber share similar levels of anxiety regarding these issues.

"I can't fully grasp the potential threats posed by AI, but initially, I expect mostly positive outcomes from AI advancements,\" Ashton stated. \"Nevertheless, I can foresee it spiraling into areas that might clash with the moral compass of many.\"

"I'm extremely concerned,\" said Andrew Barber. \"There are numerous individuals in Silicon Valley and beyond who are genuinely anxious about the existential risks that AI could present.\"

The breadth of intelligence is unfathomable.

For many of us, it's likely that we won't even witness humanity's regression into primates as AI potentially usurps our position.

When I questioned the panel about their sentiments regarding the world's future 100 years from now, they all expressively affirmed their concern.

"I do care about what happens 100 years from now,\" Bill Perkins emphasized. \"There’s a tangible possibility that my children will experience the world's outcomes, whether positive or negative, and there’s a slim chance I might too.\"

"One of the central challenges humanity faces is caring for those not in their immediate circle. This mentality is a major reason why much suffering persists globally. Time and distance (in terms of relativity) exacerbate this issue; the real question is whether I genuinely care about the events of 100 years ahead, much like one would question their concern for starving children or those in war zones.\"

"Absolutely, but it’s difficult to project current dilemmas into the future,\" Niman Kenkre remarked. \"I have a strong belief in the capability and creativity of the human spirit. I believe that humanity will still find innovative answers to the challenges that will inevitably emerge in times to come.\"

"Yes, I genuinely care. That's one of the primary reasons why I view global warming as one of my chief concerns,\" Brian Rast asserted.

Andrew Barber suggests that we have more pressing concerns than contemplating the world a century from now.

I asked the group about their emotional responses when they contemplate AI.

"As an optimist, my feelings lean towards hope rather than fear, but I firmly believe AI will radically change society in ways that are hard for us to envision right now,\" Bryan Paris shared.

"Though my initial impression may be negative, I actually feel optimistic overall,\" Terrence Chan stated.

"I feel a mix of excitement and concern,\" Bill Perkins admitted. \"Humans often misuse new technologies and knowledge in various domains, whether that's explosives, biotech, or physics. AI has countless wonderful applications, but it is a formidable tool, much like a knife or nuclear power, except far more powerful and far-reaching.\"

"I feel apprehensive and doubtful,\" Andrew Barber confessed.

How do you envision the gambling industry in 50 years if we achieve superhuman AI intelligence?

Most anticipate that gambling will persist, though in an entirely transformed landscape.

"I expect online poker to eventually fade away,\" Bryan Paris noted. \"However, I believe people will continue betting on sports and visiting casinos. Live poker may still thrive as long as players are prohibited from bringing advanced computers to the table. As technology progresses, individuals will have more opportunities to gamble since many jobs will have become redundant.\"

"Gambling taps into fundamental instincts deep within our brains that can be challenging to control,\" Terrence Chan suggested. \"Fifty years from now, we might see brain implants that allow us to modify our dopamine levels, potentially rendering gambling and even sex or food unnecessary for hitting those pleasure centers. But if we don’t get there, gambling will likely endure.\"

"Eventually, AI might determine that gambling is a detrimental activity for society and inspire individuals to move away from it,\" Matt Ashton proposed.

"The gambling scene seems pretty resistant to individuals becoming smarter!\" Andrew Barber remarked.

"An AI has a strong chance of clinching a WSOP bracelet if it’s permitted for live players to utilize it during gameplay,\" remarked Paris. \"Perhaps if Google glasses advance sufficiently in the coming years? It shouldn't take too long for that to happen.\"

"I would wager significant sums against AI competing in the WSOP before 2025,\" stated Terrence Chan. \"For the foreseeable future, poker will likely operate as it does now -- the most skilled and brightest players will rely on software tools outside the table to expand their poker knowledge, enabling them to apply even more sophisticated strategies at both live and online tables. Since computers aren't allowed during play, the top players will be those who can effectively merge the insights from these tools into their human intuition.\"

Perkins raises some valuable points regarding the evolution of ‘computer-assisted’ play. slots "I doubt they will allow bots to participate in the WSOP, but we are seeing the emergence of computer-assisted play,\" said Bill Perkins. \"For instance, players could consult a Snap Shove Chart for guidance after every hand to inform their next move. The next phase would involve these charts factoring in opponents' chip stacks and past actions to help determine optimal call or fold ranges.\"

Bad human decision making?

"Professionals are already playing high-level poker by adjusting their strategies based on their opponents' ranges and movements... At this point, even novices can perform as well as experienced players when they have stacks under 15bb in a tournament by utilizing Snap Shove Charts, which are permitted during non-hand periods. Future iterations will deepen the collaboration between humans and machines.\"

Is this what Libratus is?

"I envision a future where an app allows players to effortlessly record notes on gameplay, or even voice note their observations. These voice notes would be transformed into data, enabling players to have updated charts tailored to specific opponents during breaks or after each hand, making it easier to remember strategies.\"

Niman Kenkre welcomes the prospect of competing against a bot.

We can’t.

"I wouldn’t mind going up against a bot,\" remarked Niman Kenkre. \"I think it would be quite fascinating. It’s intriguing to consider how one might adapt and optimize strategies when up against a computational opponent.\"

Matt Ashton prefers the idea of bots engaging in competitions against one another.