Artificial intelligence (AI) in gaming isn’t a recent innovation. As early as 1949, mathematician and cryptographer Claude Shannon pondered a one-player chess game, in which humans would compete against a computer.
Indeed, gaming has been a key engine of AI, and a proving ground for the simulations, constructed environments and tests of realism that are the foundation of virtual experiences.
AI for the gaming experience
In 1989’s Sim City, for example, players controlled complex simulations, and rudimentary gaming AI was deployed to simulate something close to realism – i.e. deeply human characteristics like unpredictability. The shoot-‘em-up genre has also been spruced up with realism. In 2000’s Total War, virtual fighters had human-like emotions — just like soldiers in real-life battlefields.
AI is particularly valuable to gaming, because the gaming experience is uniquely dependent on quality. A practically Neanderthal visual experience is fine (nobody ever complained about the ‘realism’ of Pacman). A supremely polished visual experience is also fine. But a nearly-perfect-but-not-quite experience is awful to the point of disorientation and even revulsion. Game designers call it the “Uncanny Valley” – and you can see some revolting examples here. Put AI to work, and games can achieve the realism required to avoid the Uncanny Valley – the following video shows a gorgeous example of realistic animation.
Gaming website GamaSutra notes the many ways that AI techniques are contributing to the gaming experience: “There have already been successful implementations of AI in commercial games… There’s Black & White (machine learning), F.E.A.R (context-sensitive behavior), Façade (natural language parsing), Spore (data-driven life form simulations)… to name a few.”
All of these discrete techniques amount to two fundamental threads: more realism in artificial environments and/or more naturalistic interfaces between players and those environments. Allied to this is a further evolution in which environments will be spontaneous – instead of pre-scripted plots, developers will create just the environment and its mechanics, allowing AI to generate personalised scenarios and spontaneous challenges.
AI is creating the next generation of the whole leisure industry
Today, a raft of improvements in technology (consoles, cloud/connectedness, ultra-powerful graphics cards, VR/headsets, rendering algorithms) are powering artificial intelligence which in turn delivers ever more impressive environments in which virtual characters exhibit human behaviors and intelligence.
“Thanks to the modern gaming industry, we can now spend our evenings wandering around photorealistic game worlds, like the post-apocalyptic Boston of Fallout 4 or Grand Theft Auto V’s Los Santos, instead of doing things like ‘seeing people’ and ‘engaging in human interaction of any kind'”, says Jordan Pearson, perhaps less charitably, writing for Motherboard at Vice.
AI for the games industry
Then, there is AI’s contribution to the gaming business itself, rather than the gaming experience. Investors realise that the gaming industry is rapidly blending with real-world experiences (Disney World), films and other media (The Lego Movie), or merchandise (Crazy Birds) and that the monetization opportunities of this blended world will continue to grow with our increased leisure time and the addition of virtual and immersive experiences – powered by AI.
In this respect, artificial intelligence and machine learning are creating the next generation of the whole leisure industry.
That’s why Cambridge-based Prowler.io has attracted seed funding from two of the most prestigious venture capital firms; Amadeus Capital Partners and Passion Capital. Prowler makes autonomous, self-learning AI agents and is targeting the games industry first. Its technology will allow virtual entities to learn much more rapidly in their game-play environments.
Yes, that will certainly generate highly realistic experiences; but it will also allow those experiences to be generated more rapidly and more reliably. That means further automation in the process of producing viable games, plus easy tweaking of game methodologies; for example for producing sequels, spinoffs and sub-brands.
Another essential function of artificial intelligence in gaming is the interpretation of user data; and again this is a result of convergence in the industry. Traditional games were downloaded ‘whole’ and then played in a linear, unchanging fashion. But the rise of mobile casual gaming giants like Rovio, Supercell and Zynga has taught the games industry that gameplay itself is a marketable concept which can be flexed to perfectly meet user expectations.
The casual gaming business pioneered the freemium model, augmented with ‘add-on purchases’, by which the profitability of a title is dependent on its audience relevance, and the mobile gaming community became expert at deploying user data (plus feedback, ratings etc.) to optimise the product experience. Today, the same AI techniques used by marketing professionals to assess user sentiment is being deployed to maximise the relevance and enjoyability of games.
Putting it all together, artificial intelligence and gaming are rapidly becoming symbiotic. While it has always operated at the cutting edge of tech to make better games, game theory is also contributing to better AI practice. A glimpse of the future comes from Michigan State University, where researchers have deployed AI into a game specifically to learn from each player’s behaviour.
“We use Darwinian evolution to optimize the AI while the game is being played, which hopefully leads to arms races between players and AI, which will present players with new challenges all the time”, the researchers say.