Game On, Skynet! How AI's Chessboard Became Our 'Ready Player One' Reality
Game On, Skynet! How AI's Chessboard Became Our 'Ready Player One' Reality
"[the] cool thing about games. You can think of them as a gym for the mind. In life, big decision moments happen only rarely and often they are extremely critical. Where do you get to practise your decision-making under pressure? ... Games provide a safe training ground where you can get immediate feedback on your performance and then use that information to improve. Then you can draw on all that experience and training when it's time to face the real event." Demis Hassabis
Sadler, Matthew; Regan, Natasha. Game Changer (pp. 102-103).
Throughout human history, we have consistently crafted games that challenge and expand the limits of our intellect. Previous discussions have highlighted how technology and media significantly enhance our cognitive abilities. The skill of reading and writing, for instance, bolsters our short-term memory, thereby boosting our overall intelligence—an increase in working memory correlates with enhanced fluid intelligence. Additionally, the Flynn Effect illustrates how advancements in education and media over the 20th century have elevated our IQ levels; interestingly, an individual with an average IQ of 100 today would have been considered a genius a century ago!
This analogy is profound and timely as we navigate the intersection of artificial intelligence and human experience, transforming our 'Ready Player One' dream into a tangible reality.
In the chess world, figures like Magnus Carlsen emerged after the AI milestone of Deep Blue's victory over Kasparov, leveraging computer-assisted training to reach new heights of strategic mastery. Carlsen has even developed video games to share his training techniques.
MultiDimensional Games
Our fascination with multidimensional chess as a symbol of advanced intelligence has permeated popular culture, notably in TV shows like "Star Trek," where Spock, an alien of superior intellect, is depicted playing 3D chess—a testament to his brilliance despite an early episode where Kirk outsmarts him.
3D Chess on Star Trek
Similarly, "The Big Bang Theory" features Sheldon, a socially awkward but genius physicist, creating a three-person chess game, further exploring the theme of complex strategic thinking in entertainment.
The Big Bang Theory - Three Person Chess
Introducing additional dimensions to chess, whether by incorporating multiple boards or new pieces and players, significantly increases the complexity of games. Reflecting on last week's tech tonic, the scene from the MCU where Doctor Strange battles Spider-Man and manipulates the mirror world vividly illustrates this concept.
Tech Tonic: The AI Thought Show with Kes & Scott: Fri, Feb 2, 2024,
This scenario offers a glimpse into what a multidimensional video game could look like, reminiscent of a Rubik's cube combined with 'Portal.' In 'Portal 2', the cooperative gameplay further enriches this concept, and a captivating video demonstrates how bots might navigate such a complex environment.
Portal 2 Co-op
The metaphorical nature of multidimensional chess hints at the complexity and depth awaiting us. Recently, over the last decade, we've been bringing AI together with real-time strategy (RTS) games.
"Real-time strategy (RTS) is a subgenre of strategy video games where players control units and structures to secure areas of a map and/or destroy their opponents' assets. Unlike turn-based strategy (TBS) games, where players take turns, RTS games operate in "real-time," meaning all players are making decisions and taking actions simultaneously. The term "real-time strategy" was coined by Brett Sperry for the marketing of Dune II in the early 1990s"
Source: https://en.wikipedia.org/wiki/Real-time_strategy
Spatial dimensions aren't the only factor when considering adding complexity to a game. Real-time strategy (RTS) games have expanded the boundaries of complexity by incorporating elements like simultaneous actions in real-time, uncertainty (such as "fog of war"), elements of chance (non-deterministic factors), and increased spatial complexity. For instance, the state space and number of possible actions in Starcraft place it on an entirely different scale of complexity: 10^50 for chess, 10^170 for Go, and an astounding 10^1685 for Starcraft. Beyond just state space, incorporating branching factors (b) and depth (d) introduces a whole new dimension of complexity, further illustrating the intricate challenges these games present. For example:
Starcraft, however, pushes these boundaries to an unprecedented level, effectively reaching a complexity that can be considered infinite:
Source: Santiago Ontañon, Gabriel Synnaeve, Alberto Uriarte, Florian Richoux, David Churchill, et al.. A Survey of Real-Time Strategy Game AI Research and Competition in StarCraft. IEEE Transactions on Computational Intelligence and AI in games, 2013, 5 (4), pp.1-19. hal-00871001
StarCraft represents a quantum leap in game complexity, with state spaces and action possibilities dwarfing other games by several orders of magnitude.
AI researchers have dedicated efforts to developing bots capable of competing in Starcraft competitions for about a decade. Unfortunately, with traditional AI methods, the best AI bots only achieved rankings of D or D+, while top human players were rated A and A+. That changed when DeepMind introduced AlphaStar. Utilizing deep learning and their reinforcement learning strategy, AlphaStar began outperforming the world's best human players, marking Starcraft's definitive "Alpha moment."
AlphaStar: The Inside Story
Following this breakthrough, OpenAI turned its attention to DOTA—a highly popular eSport—employing a similar technique. This approach led to significant victories over some of the top DOTA players in a team-based format, where teams of five compete against each other.
Two Minute Papers: OpenAI Five Beats World Champion DOTA2 Team 2-0!
Move over MBB, RTS is the Future of Business Strategy
AI's prowess in mastering real-time strategy (RTS) games is an excellent metaphor for the complexities of business strategy. While some speculate that AI could become the future CEO, the integration of digital technologies, AI, and big data suggests that future business strategies resemble RTS games. However, I remain skeptical about AI solely assuming the role of CEO. Instead, I envision a future dominated by AI-enhanced human CEOs and strategists who leverage AI capabilities to make informed decisions.
Humanity's Alpha Zero Moment?
Economist Daniel Susskind has proposed that humans might be sidelined in the economy, much like horses were with the advent of modern technology in the 20th century. He references a perspective from the 1970s, highlighting the parallel between the obsolescence of horses in economic roles and the potential future for human workers in the face of advancing AI technologies.
"Wassily Leontief, the Russian-American economist who won the Nobel Prize in 1973, the same events suggested a more unsettling conclusion. What he saw instead was how a new technology, the combustion engine, had taken a creature that, for millennia, had played a central role in economic life—not only in cities but on farms and fields—and, in only a matter of decades, had banished it to the sidelines. .. Leontief made one of the most infamous claims in modern economic thought. What technological progress had done to horses... it would eventually do to human beings... What cars and tractors were to [horses]... computers and robots would be to us."
Susskind, Daniel. A World Without Work (pp. 2-3). Henry Holt and Co.. Kindle Edition.
Viewing the world through the lens of traditional economic models, the economist may not fully grasp the complexity of human behavior. Despite evolving from the concept of Homo Economicus—the notion that humans make rational economic decisions—we acknowledge that human rationality is far from perfect. Yet, the expectation for rational reactions persists.
The Race Against the Machines
Just as the invention of the automobile didn't render Olympic track and field races obsolete—Usain Bolt doesn't compete against Ferraris—and we continue to enjoy watching Magnus Carlsen play chess, despite AI's ability to defeat him, I foresee a future where AI and humans coexist and collaborate. Indeed, chess and running are sports, but when it comes to business and economics, leveraging AI for its speed, cost-efficiency, and superiority in certain tasks seems logical, much like opting for the fast food model for efficiency and scale. This might lead to a competitive race to the bottom, with AI-driven companies providing low-cost products and services. However, with our unique and sometimes irrational preferences, I also believe that humans will continue to cherish and seek out products and services offered by exclusively human-operated companies and those that combine human and machine efforts. This appreciation for the human touch, augmented by AI, will enable achievements previously unimaginable.
The Centaurian Interface
To achieve a future where humans and AI work symbiotically, we must delve deeper than merely outsourcing cognitive tasks to AI. It requires a harmonious blend of synchronous and asynchronous collaboration between human thought and machine intelligence, akin to a finely choreographed ballet. This journey necessitates transcending our narrow perspective on user experience (UX) to embrace how augmented reality (AR) and spatial computing, in conjunction with AI, can elevate human capabilities in the most humane and enriching way possible.
Next week: Through The Looking Glass: Navigating the Future with AI and AR in the Age of The Matrix:
"There are certain things that machines do that you can’t learn from. It depends on why they are better at doing them. Let’s take chess computers as an example. The top programs are all stronger than the best human chess player. ...they are stronger because they calculate more lines and don’t make tactical blunders.
However, what we are finding with AlphaZero is that some of the improvements or the advancements are actually strategic in nature, and that is something we can potentially incorporate in our own play.
... we are also trying to build analytical and visualisation tools that offer an insight into how these systems make their decisions and allow us to better understand what factors they are weighing up." - Demis Hassabis
Sadler, Matthew; Regan, Natasha. Game Changer (pp. 102-103)
As we look ahead, our exploration will not stop here. Next week, we will dive into the future possibilities of AI and AR, examining their potential to revolutionize how we interact with technology and, by extension, with each other. This exploration is about technological advancement and crafting a future where technology amplifies our human experience, making us more capable, connected, and compassionate.
Join us as we continue to envision and shape a world where AI and humans not only coexist but thrive together, unlocking new realms of possibility and pushing the boundaries of what it means to be truly human in the age of artificial intelligence.
We stand on the brink of a new era where technology elevates the human experience to unprecedented heights, transforming how we think, interact, and innovate—game on, indeed!
I've been thinking beyond the one human+one machine paradigm to human-machine teaming in these knowledge work domains, and since you've been convincing me more and more about the power of games, I was trying to figure out whether there are any games that demonstrate collaboration among multiple individuals that we could use as a model. Someone recommended Pandemic...are there others?