Todays Players

There is a cultural tension around video games that remains to be resolved. The problem goes beyond the adhesion or aversion that we feel for this medium that, without a doubt, has changed how we conceive of the cultural industry and its scope within the woven social network. The issue has to do with the formative 'humanizing' role, to a large extent, that this industry must necessarily occupy in the decades that follow. The concern about a disruptive medium is very old[^1]. However, a couple of years ago, an even more radical dimension was added to the discussion. What was once the problem of the impact of videogames within the young, is now of how it affects artificial childhoods and their upbringing. If previously what was sought to be regulated was a generational passing of the torch, today it is an ontological passing of the torch. What do we really represent through our games? What exactly do these machines learn and ignore with their abysmal neural networks? What does a multimodal intelligence perceive when it has to set up a house and defend itself from zombies inside Minecraft? What happens to our games—and our human intelligence— when these new players are introduced to the map of the imagination? Regarding the existential weight of raising children, we have—perhaps in vain—plenty of references. Much less so on educating new ways of thinking and, even further, possibly new ways of life.

A couple of months ago, OpenAI, the company behind ChatGPT, made a resource called Universe, a massive library of games with which they hope to start the path toward artificial general intelligence. Today, the so-called artificial intelligences are still dedicated to learning a single and exclusive activity. Bombe was dedicated to deciphering the Enigma machine, AlphaGo plays Go; ChatGPT writes, Midjourney creates images, The Terminator wants to exterminate John Connor (solving millions of human tasks in the process, but you get the point). This is not surprising if we consider that a single behavior is very complex to code and expensive to train. Furthermore, there are rarely suitable scenarios to generate or obtain data with which to educate the neural network on the multiple sensory levels of each possible task. In the words of OpenAI: “If we are to make progress towards generally intelligent agents, we must allow them to experience a wide repertoire of tasks so they can develop world knowledge and problem-solving strategies that can be efficiently reused in a new task”. Let’s set aside the concept of world knowledge for later.

There are several reasons to choose video games for this task. First of all, these are made to be learned. Each game is, in itself, a self-contained learning machine with feedback systems designed so that, ideally, any user can independently learn to master the mechanics and dynamics necessary to overcome the challenges and tasks. It is essential to think that, from the first moment, the main objective of a well-executed game is that any player can independently learn to play. Secondly, and aiming towards this, the games have systems of rules that are sufficiently clear and defined to put into action the feedback systems that reinforce the user’s learning. Third, the levels of difficulty and complexity in a game can increase or vary, enhancing the player’s learning and, ideally, consistently pushing their capabilities to new limits. Finally, they have the convenient possibility of being repeated billions of times in millions of digital instances to train intelligent agents across thousands of computing centers in the world.

These are the mechanical and tangible points that make digital games an ideal environment. However, I would like to focus on two other points that are equally fundamental although much less tangible. On the one hand—and this has been commonplace in the West¹—playing shows the true face of the player. A popular saying in Chile says, “At the table or at play, you’ll get to know their face.”² On the other hand, the ancient analogical relationship—inaugurated by fragment 52 of Heraclitus—where game and world intertwine. Be it time, destiny, or the world as a complete and abstract whole, these concepts fit into a simple and continuous game of dice.

Probably one of the richest examples of this is found in the passage of the funeral games held in book V of the Aeneid. In this description, Virgil presents, with customary mastery, the impact and effect that the game generates on humans and the world. In the story, the game surrounds and shapes the rite. It reconstitutes the cycle of the year of the death of Anchises, father of Aeneas, displays life and society through the interaction of Aeneas and his companions, and prefigures the eventual death of Nisus and Euryalus in a dramatical scene. The first of the games is the regatta. The rowers, “with tense muscles, wait for the signal” (V, 136). In the beginning, “the sailor’s shout hurts the sky” (V, 140) and, once in the water, “to the turning arms backward the beaten sea bubbling foam. // In rhythm they cut furrows and the entire surface of the liquid plain opens // torn by the oars and by the three-toothed waits” (V, 141-143).

The players stretch their capabilities to maximum tension and burst into the surrounding physical and social nature. “Instantly, the entire forest resounds with the applause // and the shouts of the spectators, // who ardently encourage their people, and their voice rolls across the shell of the beach // and hurts the hills and the echo bounces against them their cry (V, 150-153).”

There is a patent violence in Virgil’s description. The player, seeking and pushing his limit, moves towards a world that reacts terrifyingly. Somehow, an emotion similar to that of the choir in Antigone by Sophocles³ resonates with this description. The player’s violence is shown as power in the face of matter. As a harmony between the intellectual and the material that results in technique, in human labor, in the intrepid journey across the sea. We should ask ourselves, at this point, about the physicality of these AIs. What statistical muscle is strained when neural networks evolve and reorganize their ideas? What is the sensible knowledge behind an algorithm? What does a computer vision algorithm ‘see’ when it reduces to (qu)bits the light that enters through the cameras and travels through the circuits?

We know that there comes a point when that digital muscle grows. We have seen it act in unexpected ways through machinces that learn to box with a marvelous biological simile—it is charming to see it fight like a gorilla—and unexpected discoveries. The same could be said of the famous move 37 that AlphaGo used against Lee Sedoll in his second game of Go. The wonder of the spectators in Korea was no less than that described by Virgil at the funeral games. A piece of our reality was wounded that day.

However, who do we play against when playing against AlphaGo? Is it against the engineering team behind the technology? Is it the centuries of science and technology on top of which it is built? Is it against a very complex system of interconnected rules that we could hardly understand, limiting ourselves to monitor? In Lee Sedol’s game against AlphaGo, the Korean player’s need for an opponent is evident. Aja Huang, although close to representing one, fails to play the full part.

As we continue Virgil’s story, we find the development of the first point that we discussed previously. The regatta continues with the intrepid movement of Cloanthus, who, skimming the rocks, surpasses Gyas, who loses the lead. Gyas was already accusing his helmsman of cowardice for leaving too much space between the ship and the stones, and eventually ended up throwing him into the sea after losing the lead. However, it is in the foot race where the characters’ profiles are best demonstrated. Their passions, stratagems, beliefs, and motives are drawn through the description of a dramatic race.

The second game, the foot race, is described with the same violence and efficiency as the regatta. However, the scene reveals greater complexity on a human level, since this race builds a more complex bridge between the external world and the artificial, closed, allegorical world of the game. The player here acquires a role that escapes the minimum rules and transcends even to the point of prefiguring, in the case of Nisus and Euryalus, their destiny.

The race is led by Nisus. “Close to him, yes, but close at a great distance Salius follows him. // Then comes a space and Euryalus comes” (V, 319-320). Almost at the end of the race “Nisus slips, unfortunate, in a pool of blood[…]”. However, “he does not forget Euryalus, // the love of his soul, and rising from the slippery mud // he blocks the way for Salius with his body, who rolling over him // remains lying among the thick sand” (V, 335-337). With Salius left out of the race, Euryalus arrives in first place while the latter breaks into complaints. Aeneas, as befits his piety, rewards all the competitors with splendid prizes and consoles them in their reconciliation.

The scene highlights things that we generally identify as eminently human behaviors: forgetfulness and carelessness in the pool of blood, the love of runners, and direct cheating. ¿Would an AI choose to exploit the rules of a game for a significant other (AI)? If the game were collaborative, it would undoubtedly be a possibility. But would it do it just to see the glory of its loved one? Could we consider a cheater a machine that finds bugs or nooks and crannies its environment to achieve a goal in unusual ways? To what extent and with what particularity will they carry out their dawning notions of style? How did AlphaGo feel when playing a move that some called a glimmer of humanity?

The challenge that today’s designers assume when creating games for AI will define multiple dimensions of how our lives will be articulated with this technology. Technology that in turn—especially if considered in conjunction with the other great scientific advances of recent decades—is taking leaps and bounds towards questioning many of our fundamental categories. Lee Sedol sacrificed a masterful life so we could glance at how human activities may be affected. What is happening in the world of chess continues to attract attention, where after years of new ‘machine plays’, multiple accusations of cheating arise, calling into question whether or not we understand the nature of a game played for centuries.

Luckily, we have spent years of studies understanding the complex multimodal rhetoric involved in games. However, these new players are a black box of surprises. In a context where programs can learn to play without even knowing in advance the rules of a game or simply watching videos and visual records, and quickly begin to adopt and generalize their infinite multimodal experiences, it is essential to start analyzing their reactions to more complex systems in rhetorical terms. Considering that these intelligences will be the ones that will eventually emerge from their play to our daily activities and start mediating the very essence of our interactions, it is urgent to review and advance the digital learning ecosystems we are generating for the development of the AIs around us.