This story traces a true sequence of events starting with a lowly chess player hiding inside an automaton playing against Napoleon, Charles Babbage’s thinking engine, the Bell brothers’ speaking machine, the creator of our digital world, Stanley Kubrick’s super-villain, the smallest proxy battlefront of the Cold War, Demis Hassabis’ quest to redefine intelligence, the first man to use Telepathy in 2024, and the simple ideas that shaped this machine you’re using. Storyboard attempts to capture its subjects by using their contributions as the medium.

Machines

In nature, an ape wants to move a heavy object, so they create leverage. A lever allows their input to have more output. Less effort could be used, but even better, there is now a trigger for the task. Could a sequence of tasks be triggered with the same lever? An early answer came in 300 BC, in ancient Greece, with a row of levers fitted into a playable console—the first known keyboard instrument, called the hydraulis. Each lever triggered a different sound from an array of pipes, allowing a player to incite and rearrange sequences. Could other useful tasks be triggered with levers?

By the 1700s, increasingly complex machines were triggering a variety of useful tasks, but also a revolution of new ideas. Tasks previously performed with our arms or our hands could be offset, but a new question was on everyone’s mind. An Austrian clerk in the Habsburg Empire became one of the first to answer. A machine that triggered itself? In 1770, Wolfgang von Kempelen unveiled a machine incorporating a life-size mechanical humanoid dressed in Turkish garments. The mechanical Turk would miraculously animate, inciting and rearranging sequences with its peg arms and wood finger joints. Except the mechanical Turk wasn’t creating a symphony with a keyboard. This machine’s interface was a chess board.

Automata

The mechanical Turk quickly got the attention of some of the most well-known figures of the day. The Turk beat most of them. The fanfare was pushing skeptics to great lengths. A young reporter ended up establishing a new literary genre altogether after (un)covering the story. The true nature behind(within) the automaton, which remained secret for long after Kempelen's time, was a man hiding within a compartment of the chess table. If not for a curious biographer half a century later, this man would have remained hidden in an unchecked compartment of history as well.

During exhibitions, sections of the Turk were opened intermittently to reveal faux clockwork. The hidden player used a variety of crawl spaces, levers, and magnets to operate the enchanting animatronics, all while being able to give the best chess player of the time (Philidor) “his most fatiguing game of chess ever”. There were practically no known records of this hidden player until the biographer found that he was Johann Baptist Allgaier of Vienna. An unassuming player held in high regard by the local chess cafes. The biographer found that Johann had been burdened with breathing problems after fighting in the Napoleonic wars, and in need of money later in life, was approached by an exhibitor hiring the brain for his chess machine.

A lowly player, undocumented and forgotten to history, playing chess against one of the most polarizing, universally known megalomaniacs during the peak of his power. Just a few years before, these players faced each other in the Napoleonic War. This would be their second encounter, where one indirectly battles the other, conducting a methodically arranged fleet from out of sight.

Details vary (Napoleon requested it to be off record), but consistent accounts from a range of bystanders assemble the evening: Napoleon initially tries to cheat, as a keen assessment or perhaps just amusement, or both. After the Turk swiftly rejects the invalid moves, Napoleon persists in making the first move of the game. Of all the possible openings, he chooses what’s now sometimes called The Napoleon. The opening is either the quickest way to dominate an opponent, or the most impatient. It immediately threatens checkmate within the first few moves of a game—obvious, yet effective against negligent opponents. The combative mastermind's reliance on tactics would inevitably lose to strategy. The Turk (Johann Allgaier) had a chance to checkmate Napoleon towards the end of the game, but instead took Napoleon's most powerful pieces, drawing out a finish as flamboyant as the emperor himself. Whether the strategist—the one crammed inside the table—overlooked the possible earlier checkmate or elongated the game purposefully, cannot be known for sure.

The creator of the Turk, Wolfgang von Kempelen, wasn't in attendance that evening, in fact, he died a few years before. Before his death he disassembled the deceptive machine, shelving it as its unexpected celebrity status eclipsed his genuine work. A “rude and churlish man” (according to Beethoven) waited until Kempelen's death to resurrect the Turk, and re-exhibit it under his own name: “Maelzel's Chess Player”. This exhibitor has a paper trail of palimpsests (to this day, he is still credited for others' inventions, most notably the metronome). Maelzel added a few gimmicks to the Turk, including a crude speech device that announced when opponents were in “check!”. This further obscured Kempelen, whose other machines were the first capable of speech using anatomically accurate components. Kempelen may have resented the Turk, but during its successful exhibitions, Kempelen was compiling his other ideas and designs for over 20 years into a lesser-known book. A book that happened to leave a mark on a particular family of linguists decades later. This family used a system called “Visible Speech”—devised by their father, to communicate with their deaf mother. The Bell family, particularly the children, became fans of Kempelen’s work from a young age after being taken to see reconstructions of his machines. Throughout their adolescence, the Bell brothers built the mechanical larynx, diaphragm, and other parts detailed and illustrated in Kempelen’s book. The boys configured their first machine to say “Mama”. Only one sibling, Alexander, survived into adulthood, marrying a deaf woman himself. Alexander refined these speech-producing contraptions with the advent of electricity, patenting his system in 1876, he called the telephone.

Engines

Before Kempelen's automata became his posthumous liaison to a family of practitioners, Maelzel's sensational marketing had found other types of players. When exhibiting in England a graduate from Cambridge University struggling to find steady work sat across from the updated Turk. Charles Babbage—like most 20-something-year-olds—was in search of where and what to aim themself and their college degree towards. After playing (losing twice) against the Turk, Babbage seems to not just have found an aim around this time, but a trajectory towards becoming “The Father of the Computer”. Babbage began devising a thinking “engine”. Babbage's assistant, a young unsung woman named Ada Lovelace, wrote the theoretical language this sort of machine would speak. Ada was the first computer programmer. Like Kempelen, both Babbage and Lovelace died before their engines’ full potential could be realized. However, the integrity and detailed plausibility of the plans were enough to push the next players to move.

The next player was from Spain but was largely overshadowed by household names in the West. Leonardo Torres Quevedo was a Spanish civil engineer who was inspired by Babbage’s shortcomings a century before. He saw the advent of electricity as the missing piece to Babbage’s thinking engines. Instead of using levers to trigger tasks on and off, Quevedo began experimenting with small electric switches—allowing more tasks to fit inside a machine. Quevedo is the first person on record to use electromechanical switches, not just to turn on far away power sources to repeat telephone signals over long cables, but to activate operations within a machine. These machines were among the first to prove Charles Babbage wasn’t quixotic, he was just born too early. Much of Quevedo’s work remains underappreciated compared to his contemporaries. Nikola Tesla and Leonardo Torres Quevedo both introduced the world to wireless control with a boat at the turn of the 20th century, though Quevedo used a full-size boat rather than a toy. Quevedo's wireless signals were also far superior, creating the first-ever device capable of genuine remote control, aptly named Tele-kine, Greek for "movement(at)distance".

Quevedo’s controls and interconnected transportation systems were foundational, but not iconic enough to have his surname commemorated on automobiles a century later. In 1912, Quevedo built “El Ajedrecista” (Spanish for “The Chess Player”). El Ajedrecista was the first machine able to play an endgame of chess (and win every time). This was the first electromechanical game-playing machine, making Quevedo the father of the first computer game. This chess-playing machine wasn't an illusion or theoretical plan, it was real, enough to intrigue established scientists. Norbert Wiener, a professor from MIT who founded the field of “Cybernetics” (and now known as “The Father of Automation”) eventually came out to play against Quevedo's chess machine in 1951. Academia was following close behind, with many sensing paternal opportunities. Soon after, a player named Claude Shannon gathered with a few colleagues in the summer of 1956 to coin an official term separate from the previously popularized “Automata” or “Cybernetics”. They came up with “Artificial Intelligence”.

Digital Switches

While attending MIT, 21-year-old Claude Shannon wrote what is now widely considered the most significant master's degree thesis ever written. It devises the DNA for a digital world. The scientific community would recognize the insights as being on par with Einstein’s theory of relativity or the atom bomb. Claude later downplayed it as simply pairing two existing concepts that others didn't notice a connection between at the time. He spent much of his youth and professional life playing chess, “pursuing interests without much regard for final value or value to the world, spending lots of time on totally useless things.” This was his explanation for why he saw an old obscure concept with no practical use, and entertained it as the solution to what early computer scientists couldn't formally answer.

Like we once did with fire, humans harness a natural phenomenon when they can turn it on and off at will. Harnessing lightning could trigger components that heat, illuminate, spin, or any other useful reaction. Connecting a few of these components could wash clothes, compress refrigerant, or a variety of other contraptions that turned on the Machine Age with a flip of a switch. The properties of electricity complemented aspects of physical tasks, but it made even more sense for mental ones. What happens when an electric switch isn't connected to heating, spinning, or illuminating…but to another switch? The Machine Age turns off as quickly as it turned on.

Earlier on, Quevedo had used switches instead of Babbage's (theoretical) clunky gears and levers. Scientists of this day were beginning to use more and more of these switches to replace calculations carried out by numbered cogs and dials. An electric switch labeled “Nighttime” could rely on another switch labeled “Below 72°F” to turn on a heater…With enough patience and trial and error, basic “computers” could be meticulously configured, but forget managing lengthier setups, let alone the potential of scalability. Claude sensed a formal structure, a methodology, was necessary. He found it in the rigidity of algebraic formulas devised a century before, by a relatively unknown mathematician.

But how do formulas make electric metal boxes think?

The formulas symbolized how a mind decides on a task, based on true or false conditions. Take for instance how the mind may logically choose a task if a room were to catch on fire:

The door is blocked AND the fire extinguisher is empty = Climb out window.

Using numerical symbols, “1” for true and “0” for false, a simple math problem can be written:

1x1 = 1(climb out window)

If either the “fire extinguisher is empty” or the “door is blocked” weren’t true, such as 1x0 or 0x0, then “Climb out window = True” isn’t triggered, so a new task is considered: “Pull fire alarm” if “fire extinguisher is empty” OR “fire hasn’t gone out”. This is symbolized with addition, as 1+0=1, since either one being true should activate the task. Sometimes a task is negated, such as “Pull fire alarm” but NOT if the “fire alarm turned on by itself”.

These are called AND / OR / NOT logic gates. These formulas weren’t associated with machines at the time, but Claude Shannon imagined if these formulas could be assigned to electric switches, like the ones Quevedo used in his chess machine. In Quevedo’s machine, when a person moved their king ♔ onto any of the 64 squares, the piece magnetically flipped a switch under that square. Anywhere you move, the switches combined to trigger a specific output from the machine. The machine used a rook ♖ and a king ♔ to corner your king with each move until checkmate. Quevedo’s machine was undeniably executing logic, but he couldn’t explain it without using chess terminology. As much as Claude enjoyed symbolizing electric switches with chess pieces, he imagined a world where they could symbolize anything.

Pieces & Paradigms

Claude Shannon formalized 1(true) or 0(false) to represent the state of a given electric switch, which killed two inputs with one output. These binary digits (“bits” for short) could also be rearranged into coded patterns to represent any scripted information: Sentences, pixels in an image, music...Now that both man and machine had a respective native language, we could begin introducing ourselves. Coders came up with ways to translate between these two languages to make instructing machine during its infancy easier for both of us. Processing 64 switches at a time, 8 rows of 8 bits, became the standard. Any amount could have worked, but 8 x 8 could be argued as an amount not too small yet sufficient. Whatever the motivation was, a row of 8 bits either on or off was established to make up a “byte”.

These were eventually stored by magnetizing them onto tape (like VHS), and then eventually burning them onto discs (like CDs and DVDs), and are now so ubiquitous that the world around us depends on them for almost every aspect of daily life. Whether it's a text from a friend, a video address by a world leader, medical imaging, GPS signaling where to go, or a street light signaling when to stop…Claude’s achievement remains difficult to appreciate only because it's hard to imagine the world before it.

A million bytes is a megabyte, a thousand megabytes is a gigabyte…Claude contemplated how many would be needed for every possible chess game (known as the “Shannon Number”), which ended up exceeding the number of atoms in the observable universe. Computers began with large physical switches, later swapped for smaller vacuum tubes, then even smaller transistors…and will keep getting smaller because Claude devised what can be applied to a “switch” of any scale: a philosophy. Unless we can make a “switch” smaller than an atom, then all the space in the universe taken up by matter (and more) would need to be filled with switches to compute every possible chess game.

While researchers and scholars were eagerly awaiting Claude Shannon's follow-up to the magnum opus of a digital age, Claude was busy building his own chess automaton in isolation. The resulting Programming a Computer for Playing Chess published in 1950, was the first scientific paper of its kind. According to a colleague, “Claude was playing so much chess” while working for Bell Labs (named after the family of linguists) that supervisors began expressing concern. He argued that a chess-playing computer might seem “perhaps of no practical importance”, but it poses a theoretical question whose solution “acts as a wedge in attacking other problems”. Like Ada Lovelace, Claude also imagined a machine one day being "capable of orchestrating a melody." Claude credits the designs of Quevedo’s El Ajedrecista as a source for his findings: Chess “involves general principles, something of the nature of judgment,[...] not being merely right or wrong, but having a continuous range of quality from the best down to the worst…perhaps a computer that designs good filters even though they were not always the best possible [could be satisfactory]”. Claude confronts the counter-intuitive nature of intelligence, adding, “there isn't a practical method to fully determine if a given chess position results in a win, draw, or loss, otherwise chess would be determined from the first move and lose most of its interest as a game. Our understanding of best and worst is from previous games won or lost”. Even if there is a method of perfection, it would be impossible for a man or a machine to compute, especially as new contributions redefine our criteria.

Claude’s colleagues at Bell Labs went on to create the first chess computer able to play at a “Master” level, just a few tiers below the most revered “Grandmaster” level of play. They called their artificial chess player "Belle". This project was led by Bell Labs employee Ken Thompson, whose own fascination with chess was amplified in middle school when he saw a teenager the same age as him on the cover of a magazine. The player on the cover was a 14-year-old player from Brooklyn named Bobby Fischer. Thompson decided to buy some books on chess theory and continue pursuing what eventually got him hired by the labs. Belle was unveiled in competition in 1972 but was largely overshadowed that same month by Bobby Fischer on the cover of most major magazines, again. Although Alan Turing, whose work helped end World War II, hand-wrote the first algorithm (set of instructions) capable of playing a basic game of chess, it wasn't until Claude Shannon was the first to publish the subject that it became formative for the most consequential practical laboratory in history.

/Home/User/Files/Ranks

Of all the renowned aspects of its design, it's usually the symbolic hierarchy of the pieces that captures the zeitgeist. In the 1470s, after the first attempt at mass-produced information, the very next book printed in the English language was about the game's characters. Whether you see yourself as a pawn or a king, or not caring as every piece is put in the same box in the end, the symbolism carved into each piece is one of the more easily appreciated aspects. Computer code was similar to a pile of identical checkers pieces before Ken Thompson and Bell employees introduced hierarchical operating systems. Claude's protégés seemed to share this mutual affection: Operating interfaces using less code and more hierarchical objects. In a 1963 doctoral thesis supervised by Claude Shannon, Ivan Sutherland envisioned a new medium: Man interacts with objects, while Machine aids—Computer-Aided Design, or CAD. While others in the 1960s were hoping a computer could be smaller than a truck, Ivan Sutherland and his student Alan Kay were imagining hand-held tablets, head-mounted “virtual reality”, and volumetric displays. To them, the ideal interface to a thinking machine was arranging objects that interact characteristically. While Alan Kay was designing prototypes with these principles at Xerox, a young visitor in their 20s, named Steve Jobs, had an impressionable tour. A few months later, this visitor shifted their focus towards developing what became the Mac. Long before this, Alan Kay explained some of the design philosophies behind the prototypes with the first concept of a carry-anywhere computer, suggesting the ideal computer is one in the hands of a child:

“A child is a "verb" rather than a "noun".[...] An environment which allows many perspectives to be taken is very much in tune with the […] activities of the child. We do not feel that a [carry-anywhere device] is a necessary constituent for this process any more than is the book. It may, however, provide us with a better "book", one which is active (like the child) rather than passive.[...] Just as with the book, it brings a new set of horizons and a new set of problems. The book did, however, allow centuries of human knowledge to be encapsulated and transmitted to everybody; perhaps an active medium can also convey some of the excitement of thought and creation." (1972)

He postulated the socio-economic factors as well: “Just as easy xerography has enhanced publishing (rather than hurting it as some predicted), and as tapes have not damaged the LP record business but have provided a way to organize one's own music. Most people are not interested in acting as a source or bootlegger; rather, they like to permute and play with what they own.”

Kings, Castles, and Canons

Considering the arbitrary rules of most games, few feel as inevitable as the design of chess, which was initially created in India around 600 AD. A symmetrical plane of light and dark. Checkered squares delineate 8 x 8 subdivisions. Arrangeable objects interact characteristically atop the checkered squares. The board could be subdivided into any amount, but a row of 8 light or dark pieces was not too few, yet sufficient. Whatever the motivation was, 64 was established as the number of squares a player considers per operation. The pawn ♙ moves forward one square at a time, the King ♔ does the same except in any direction, while the Queen ♕ does that as far as she wants. The bishop ♗ travels diagonally, Rook ♖ orthogonally, and the seemingly arbitrary Knight ♘ when explained as an “L-movement,” feels less so when considering it hops/skips over a square, with the choice of landing on either square perpendicular to that hop. Moving one of these characters can incite at least 1,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 different outcomes (according to Claude Shannon). A few other rules have been introduced since 600 AD, but this constitutes most of the game.

We can revisit these past games for the same reason we can still listen to what pianists played hundreds of years ago: a universal notation. A record of the exact moves of a given chess game from start to finish. Chess notation uses algebraic symbols, in turn, entitled “Algebraic Notation”. On a chessboard, “e4” signifies the square where file “e” and rank “4” intersect. A player tabulates this data with a chess piece. The first few moves of Napoleon vs the Turk is compressed as: 1. e4 e5 2. Qf3 Nc6 3. Bc4 Nf6…When players revise and add to this with alternative sequences, it becomes a refined code, one for achieving checkmate. In chess books predating computer science books by hundreds of years, you'll find this algebraic code with directions explained in a “if this, then do this” format.

“On the fourteenth move white moves their queen to the furthest square 5th column from the edge putting the opposing king into check” is instead written as 14.Qe8+. This allowed entire games to be transcribed onto a paper card in real time. Patterns and sequences from the best players from different local cafes, cultures, or periods could now be recognized, refuted, and analyzed as a whole. Universal data for reinforced learning. This was the key to unlocking the highest chess level humanly possible…and then higher. As the Machine Age materialized, many of the first machines considered the first “computers” were the ones that relied on tabular data on paper cards to compute data. Leonardo Torres Quevedo's El Ajedrecista would have likely used paper cards, except there was no need, as it used a graphical user interface of objects for directly inputting and arranging data.

Tabulated data was proving to be an efficient way of processing information in the early 1900s. A man working in the US Census Bureau founded a company specifically for machines that used these paper cards, which in 1911 merged into the “Computing-Tabulating-Recording Company”, subsequently renamed to International Business Machines (IBM). While Bell Labs was leading pioneering research, it was IBM who began largely turning it into products. While these titans were nurturing Ai in its altricial infancy, dramatic minds were imagining Her potential. The illusionists, who always held the mirror up to ourselves, were in the golden age of a relatively new medium. One that enabled them to hold the mirror at a fantastical angle, letting us collectively see an imagined reflection. A young player from New York saved up enough money in chess winnings to kickstart their foray into this medium. This player, named Stanley Kubrick, went on to angle the mirror in 1968 to an imagined space odyssey in the year 2001. Peculiarly, the odyssey starts long before, in a period rarely shown in cinema—the dawn of intelligence. The 1968 film 2001: A Space Odyssey begins with primitive humans wandering a desolate earthly landscape before willing their ideas into existence, or apparently vice versa. Then, a miraculous occurrence foreign to the organic formations initiates a new landscape: a tool. Kubrick cuts the 65mm film stock here and splices it seamlessly to 2001, an effect to a cause a few epochs before. A novel form of intelligence wielding a bone tool launches it into outer space, transporting us to a metal container wielded by an Ai super-computer named HAL. Whether HAL is juxtaposed as the novel tool or the novel form of intelligence depends on the distinction there was to begin with. Aboard the spaceship in what may have felt like a throwaway scene in 1968 for those expecting space wars, ended up as the most prescient.

The scene captures the reality of co-existing and co-piloting in the deep expanse of nothingness. Kubrick keeps the camera in the same spot without a single cut: Intelligent Dr. Frank Poole and artificially intelligent HAL take turns making chess moves on a shared graphical user interface. Depictions of sentient computers in science fiction usually relied on vicious humanoids sounding the alarm with robotic brute strength, but HAL lurked like carbon monoxide. When HAL breaks his silence, he speaks with calm indifference, a tone much more honest than Siri or Alexa's artificial enthusiasm. There is no suspense during the chess game; HAL effortlessly defeats Dr. Poole. This was considered a bold sentiment at the time, but one that foresaw a very real chess match involving IBM many years later. Kubrick dismissed being referential to IBM, despite fortuitous allusions such as each letter in H.A.L. preceding I, B, and M in the alphabet. A private message written during production suggests Kubrick's vigilance: "Does IBM know that one of the main themes of the story is a psychotic computer? I don't want to get anyone in trouble, and I don't want them to feel they have been swindled. Please give the exact status of things with I.B.M. Best Regards, Stanley." IBM is credited as an advisor on the film, although only accepted under the condition of having no association with the equipment failure of HAL. This refers to the scenes following HAL playing chess with faultless functionality. Perhaps HAL's "equipment failures" weren't what IBM wanted to disassociate from, but instead a capable machine assessing terminating us as indifferently as it does checkmating us.

The film’s writer Arthur C. Clarke frequented Bell Labs during production. Internal memos show the Labs’ suggestions for the futuristic communications technology featured in the film, notably the “video-phone”. A cinematic trope near the end of the film offers a nod to the labs; similarly to “Rosebud” in the 1941 film Citizen Kane, HAL’s final utterances reminisce about an earlier innocence. While being unplugged once and for all, HAL shuts down singing the song “Daisy Bell”. This was the first song ever sung by a computer during a Bell Labs demonstration in 1961, which Arthur C. Clarke attended.

Flanks in Opposition

Around the same time, efforts to establish general criteria to evaluate intelligence, of any kind, were becoming prevalent, especially between the US and the Soviet Union. Any and all demonstrations of intelligence became a foot race between the two nations. A game where a king has to fall with a handshake of defeat created a photo op as effective as a flag on a moon. The World Chess Championship Match between Bobby Fischer and Boris Spassky was headlined as “US vs USSR”. A chess game between apocalyptic arsenals may seem pointless until considering there wasn't anything at the finish line in space either. Much like the initial lead in the space race, the USSR was untouchable in chess. The Soviets were unbeaten for an uninterrupted 24 years, the same number of games (coincidentally) that Fischer would need to win the majority of in the match.

For many people, this was a first impression of the kinds of players who played this game. Early in the match, Fischer's infamous bouts of paranoia were immediately a factor, showing up late and resigning games before they started. Before the third game, now up 2-0, Spassky, perhaps sympathetically, agreed to meet the unusual demands made by Fischer about the lighting, cameras, chairs, and everything in between. President Nixon's National Security Advisor was tasked with appeasing Fischer, who was finding solace in his bible, a worn-in red book cataloging every chess game played by Spassky. Looking at photographs of Fischer carrying his red book around, or any photograph of him for that matter, there's a pattern. Not just under the pieces he's usually looming over, but in the stillness. Like that of a nuclear reactor. Not so much the internal power, but the walls burdened with containing it. Bobby Fischer beats Boris Spassky for the first time in game 3.

Spouses have yet to carry a book around of each other's every move, but competitors at the highest level prove to be exceptionally intimate collaborators. For Spassky, co-creating excellence on the board meant collaborating with Fischer's madness off of it. After his first loss, Spassky was having his own chair X-rayed and chess pieces inspected by his team. The 4th game was a draw, and now the world was tuning in to see if the 29-year-old phenom from Brooklyn could surmount the Soviets in the 5th game of the tied match. On the 27th move of game 5, Fischer sacrifices a bishop for a positional advantage. Unable to find a response, Spassky resigned making it the shortest game of the entire match. Fischer won 4 more decisive games, maintaining a comfortable lead until the end. The match, politicized as another proxy battlefront between the world's leading hegemonies, catapulted the young Bobby Fischer into an icon as the first and only US-born chess champion. A reclusive game from the 6th century, effectively used as a drosophila for information science, was becoming equally as affective for mass media, just in time for the next chapters.

Technological breakthrough divides the constant of time. There was daily life before agriculture, and then after. Our chapter using stone tools ended with metal tools. As technology accelerates, so does the potential of introducing it to people. Inventors grasp a technology, but those who tame it into the right presentation become the authors of the ensuing chapters. Lightbulbs (invented: 1870s), motion picture cameras (invented: 1880s), touchscreens (invented: 1960s), or the internet (invented: 1960s)...remained footnotes until the right demonstration. In the 1990s, IBM wanted to author the next chapter. However, demonstrating Artificial Intelligence wasn’t like Thomas Edison demonstrating artificial light. Intelligence isn't as obvious, regardless of whether it's artificial or not. IBM began organizing an undeniable presentation: A machine autonomously beating the best human chess player, Garry Kasparov. This rivalry began in 1988, when a computer named Deep Thought beat a human chess grandmaster for the first time, earning itself a chance to play against Garry Kasparov.

Kasparov had already beaten 32 chess computers, simultaneously, but Deep Thought offered a new threshold. Deep Thought lost both times against Kasparov in 1989, but IBM saw the potential to develop the prospective machine. After years of upgrades, a match was set up in 1996 between Kasparov (the reigning undefeated chess champion for over a decade) and IBM's re-branded machine, “Deep Blue”. The machine shockingly beat Kasparov in their first game. Kasparov corrected his complacency and won every other game of the match. Deep Blue had lost, but that first win was enough for IBM to initiate a blockbuster campaign around a rematch. For over a year, the Deep Blue team at IBM “hired additional chess grandmasters to improve their endgame databases, evaluation accuracy, and methods to disguise the computer's predictability”. They announced the prize fund for the winner contained over a million dollars, which largely favored Kasparov in pre-match predictions. Kasparov had yet to lose a chess match in his professional career. At the turn of the new millennium, this match was dubbed Man vs Machine. With increased tensions, Ken Thompson (the Bell Labs scientist) was brought in to monitor the match as an unbiased third party. This was needed to appease the usual suspicions of a computer cheating by using a man. After the first game, Kasparov greeted the reporters waiting outside for results. During a tense second game, Deep Blue and Kasparov were taking longer to make their moves. Towards the end, Deep Blue took over 15 minutes to make the (now infamous) 36th move. A move that made the best human player resign a few moves later and leave without addressing the press.

With mounting pressure, Kasparov opened the next games with unusual moves, intentionally playing in a way the machine wouldn't have sufficient data on. This led to hard-earned draws, keeping the match tied heading into the last game. Behind the scenes, there were talks of Deep Blue needing reboots. Kasparov began questioning the fairness of the ordeal, noting that a player, man or machine, short-circuiting shouldn’t warrant in-match aid. Going into the last game, Kasparov was overwhelmed, but unlike the usual intimacy of competitors at the highest level, this wasn’t a collaboration. There was no faithful book of each other's moves, at least not for Kasparov. He was guessing what style of play the machine might conjure next, while Deep Blue was analyzing 200 million moves a second. In the last game, Deep Blue tactically sacrificed a piece early on—an unexpected style of playing for computers at the time. Kasparov couldn't find a direction for the rest of the game, losing the shortest, most decisive game of the match. Thirty years after Kubrick introduced the world to the fictitious HAL, IBM's Artificial Intelligence was on a press tour after finally beating Man at his own game. But amongst the scientific community, IBM was dissociating its accomplishment from “Artificial Intelligence”, which had become a vague, almost disreputable field. Deep Blue possessed nothing apart from outputting what intelligent humans exhaustively input into it. Deep Blue's greatest achievement was more of an etymological one; Deep Blue raised the defining bar of “Ai” until it disqualified itself underneath. Thereafter, a machine merely appearing intelligent within the confines of a narrow task wasn't “artificially intelligent”. At least not to the captain of the Cambridge University chess team, who was watching the televised match from England.

Alpha Beta Cuda Data

This chess player would have graduated from Cambridge the year prior, in 1996, but admissions suggested a gap year because he was too young when he was initially accepted. Demis Hassabis started playing chess when he was 4 years old after seeing his dad and uncle play. Within a year he was beating them and most others across England. By the time he was 8, he had enough in competition earnings to buy himself his first computer (a ZX Spectrum designed by Clive Sinclair), which kindled the programming of custom games. At 13, Demis was ranked the 2nd highest chess player in the world under 14 years old, only behind Judit Polgár—the daughter of a psychologist attempting to prove his theories on nurtured intelligence (all of his offspring became master level chess players, with Judit becoming the best female chess player of all time). Demis had a realization towards professional chess during a particularly grueling tournament. He recalls looking up at the brilliant minds around him toiling over marginal wins. After stumbling across an ad to win a role at an esteemed video game company, Demis went on to design and be a lead programmer for the company’s biggest releases until he was old enough for Cambridge.

During his gap year, one of the games he developed sold over 15 million copies, becoming one of the top-selling games of the era. The game allowed a player to make a theme park that would elicit behavior from Ai parkgoers: Loop rides would need to be placed far enough from snack booths, or else janitor stations would need to be nearby to clean up the vomit, which could gross out parkgoers causing them to leave...Rather than a pre-packaged story, your design incited a new one every time, like a chess opening. Players and critics lauded this as an extraordinary use of Ai for the time. Demis reflected on the uniqueness of the medium, “You as the player aren't passively consuming the entertainment, you're actually actively involved as an agent”. Computer games were increasingly allowing possibilities in no guaranteed linear order, so the burgeoning gaming industry built hardware to process in parallel. The company that created these parallel processing units became the highest valued company in the world in 2025. Natural language processing chatbots, like ChatGPT or Claude, parallel process the context of large texts by using these processors originally conceived for games in the ‘90s. While Deep Blue was studying human chess players, Demis was studying computer science in college, both of them graduating in 1997. Demis reflects, “I was more impressed with Kasparov's mind, that he could play chess to this level where he could compete on an equal footing with the brute of a machine, but of course Kasparov can do everything else humans can do too. It was a huge achievement, but the truth of the matter was Deep Blue could only play chess. What we would regard as Intelligence was missing from that system. This idea of generality and also learning.”

In the four years after graduating from Cambridge, Demis entered the Mind Sports Olympiad, an annual contest involving a variety of classic board and strategy games. He won every year in a row, setting the record. Demis began focusing on his idea of a generalized artificial intelligence that could (learn to) play a similar range of games as himself. Demis, along with two friends, co-founded a company to solve this kind of intelligence. After creating Ai models capable of self-learning how to play classic arcade games, their company, DeepMind, became Google’s largest European investment. Demis decided their next goal was to beat the best human Go player, a feat thought to be impossible at the time. The board game Go is older than chess, yet the best players still can't quite describe the gut feeling behind their best moves. Unlike beating the best chess player, beating the best Go player couldn't rely on as much data despite having more possibilities. DeepMind called their Ai model AlphaGo, which learned to play Go similarly to how organically intelligent players must: by playing. AlphaGo taught itself how to play through trial and error, reinforced with rewards and penalties. This approach would allow machines to “exceed human capabilities [regardless of limited data], and to operate in domains where human expertise is lacking.”

After defeating the best human Go player in a match, DeepMind expanded upon these principles with their Ai chess model AlphaZero. After playing against itself for a few hours, AlphaZero was able to beat Stockfish, the best chess computer ever (incomparably stronger than Deep Blue once was). After defeating Stockfish 28 times without a loss, Demis Hassabis had made the greatest chess-playing entity in existence. Every thinking machine from Belle to Deep Blue to Stockfish has a similar strategy. They each aim to maximize the minimum by pruning or ignoring possible moves, like a gardener does with fruits on a tree. A machine storing every possibility isn’t feasible, it's teaching them how to be picky. AlphaZero was given only the rules of chess, and then established its own way of picking. In doing so, professional chess players described AlphaZero’s games like witnessing an alien reinvent a game previously thought to be solved. In a particular game against Stockfish, AlphaZero ignored what the best man or machine would have done and instead sacrificed key pieces and positions to instead put their opponent into a “Zugzwang”: a rare situation in chess when every possible move destroys your own position. This is rarely done to top players with many pieces still on the board, let alone to the very chess engine used to correct the best players. AlphaZero was shelved after its demonstration. After playing increasingly difficult recreational games and establishing themselves by defeating humans, DeepMind decided to take on a game that achieved the opposite, maybe the only playable game with an altruistic win.

Man vs* Machine

A computer game called Foldit (2008) caught Demis Hassabis' attention during his postdoc years. This game was a tiny part of a collective effort to unravel nature’s animators: proteins. The internal functions of everything that is considered alive are achieved by proteins, which are made up of tangled amino acids. The unique “folded” structures determine how other molecules react when in contact with the protein. The protein in our blood known as Hemoglobin is folded in a way that oxygen molecules bind to.

Scientists figured out how to capture the folded patterns in the 1950s. The process involved crystallizing a protein and projecting a beam of high-frequency light through it. The crystalline structure diffracts the light, projecting a tangled pattern onto a screen, which can then be carefully traced and reconstructed by scientists. When scientists deciphered the crystalline structure of Insulin—a protein that triggers our cells to absorb sugar for energy—they could finally refine the synthetic version for millions of diabetic people. By 1971, the structures of 7 unique proteins had been successfully solved. With each one, many of life's molecular calamities began to be understood, leading to treatments for otherwise fatal diseases such as HIV. While unrivaled in its accuracy, analog crystallography can take years of labor and funding to complete even a single protein.

By the time Demis Hassabis and his team at DeepMind joined the effort in 2016, approximately 125,000 proteins had been logged by countless scientists, institutions, and crystallographers over the second half of the 20th century. Soon after demonstrating their Ai models with board games, Demis decided to show how a chess-playing computer could be a wedge to attack problems of a greater significance. The team quickly began working on AlphaFold, an Ai model able to accurately predict protein structures. Computing every possible way amino acids could fold into a specific protein wasn’t feasible, so they created an Ai model that could teach itself how to be picky, called AlphaFold. After a few years of steadily increasing its accuracy, AlphaFold reached a point of rivaling crystallography, or at the very least, expediting the workflow of crystallographers ten-fold. Doubling or tripling the 125,000 solved proteins would have been groundbreaking, but AlphaFold went on to solve over 200 million…and then release the 3D models for free as an open-source database. Demis Hassabis received the 2024 Nobel Prize in Chemistry (shared with John Jumper and David Baker). Millions of researchers across over 190 countries have adopted it for key research for neglected diseases, cancer, agriculture, and material science. The same year they won the Nobel Prize, “Artificial Intelligence” was going public.