Unpacking Square Enix's 'Aggressive' Pursuit of AI
It's Far Less Sensational Than it Sounds
AI and Games Plus is my second YouTube channel that is dedicated to expanding on topics from AI and Games in more detail. It’s made possible thanks to crowdfunding on Patreon as well as right here with paid subscriptions on Substack.
Support the show to have your name in video credits, contribute to future episode topics, watch content in early access and receive exclusive supporters-only content.
Happy New Year, and welcome back to AI and Games. I’ll be continuing my work to dig into the intricacies of how artificial intelligence is utilised across the games industry, and as soon as I was back at my desk, we had another story that caught some headlines.
Game developer and publisher Square Enix have a tradition whereby their acting President publishes a letter largely aimed at business executives and investors. In this years entry, naturally dated January 1st, President Takashi Kiryu details the companies desire to be focus on artificial intelligence applications in the coming year. This subsequently gained headlines across gaming media, given the language adopted and the underlying sentiment.
“We also intend to be aggressive in applying AI and other cutting-edge technologies to both our content development and our publishing functions. In the short term, our goal will be to enhance our development productivity and achieve greater sophistication in our marketing efforts. In the longer term, we hope to leverage those technologies to create new forms of content for consumers, as we believe that technological innovation represents business opportunities.”
Takashi Kiryu, Square Enix President
Naturally it’s easy to read this statement and approach it with, highly warranted and completely justifiable scepticism. There is a lot of hype surrounding artificial intelligence in the past 12 months courtesy of the recent improvements to OpenAI’s GPT, alongside a surge of new products and services that have emerged promising to democratise game development and increase quality and productivity of titles. Despite this, we’ve yet to see much of this work make its way into gaming products, but a lot of money is being thrown around for a variety of products of varying scope.
The language adopted, combined with their track record, is what raises eyebrows across gaming media. Square Enix has been known in recent years to exploit any and all opportunity where they think there’s an opportunity to make a bit of money, often with it resulting in the opposite. The very same newsletter published in 2021 and 2022 by former President Yosuke Matsuda discussed a desire to aggressively invest in blockchain technologies and sell NFTs as part of its gaming portfolio. This has largely been met with disdain from consumers, all the while Square Enix has struggled to land any major success outside of their Final Fantasy franchise as they continued to explore ‘live service’ provisions with titles such as Babylon’s Fall and Marvel’s Avengers.
But despite this backdrop, I do not feel that this news is as sensational as it may seem on first reading. From the outside looking in, Square Enix have already invested significantly in useful and practical applications of artificial intelligence for existing game productions. And perhaps their recent efforts have highlighted their awareness of how ready this technology is for wider consumption, and how quickly it should be shared more broadly.
It’s worth highlighting that these New Year letters are very much aimed at investors and those operating at C-suite level of the games industry. As the new calendar year is underway, its important that executives at companies of this magnitude are able to provide some expectation of their plans for the coming year. Note that this is seldom digging into the minutiae of titles being released, such as the next entry of Final Fantasy VII Remake. But rather to highlight key strategy and areas of technology that the company is exploring.
So it’s not unsurprising when big technology booms are happening, that they receive some sort of mention in these newsletters. It’s an effort to sweet talk investors to put their money into the company. Hence the repeated emphasis of Blockchain in previous years (which led to a surge in their share price), as well as the likes of cloud compute and virtual and augmented reality (VR/AR). It’s not surprising, given the current technology and investment climate, that Blockchain is mentioned but once in the entirety of the letter, meanwhile AI receives such prominent billing.
“Artificial intelligence (AI) and its potential implications had for some time largely been subjects of academic debate. However, the introduction of ChatGPT, which allows anyone to easily produce writing or translations or to engage in text-based dialogue, sparked the rapid spread of generative AIs. Its release made it apparent that the applicability of generative AI was by no means limited to text, and the subsequent months saw a quick succession of launches of new services and content that expanded generative AI into a variety of domains with close ties to digital entertainment, including images, video, and music. I believe that generative AI has the potential not only to reshape what we create, but also to fundamentally change the processes by which we create, including programming.”
Takashi Kiryu, Square Enix President
But even when we rationalise this as simply to lure in investors, there is still a lot of concern about the subsequent development and applications of this technology. While Blockchain gaming is on the wane, generative AI will continue to become more prominent across a variety of business sectors in 2024. Given there is still many issues surrounding this field - notably copyright, legality of data sources, workers and artists rights - its natural to be concerned by it.
While I think this is certainly warranted, I wanted to take a moment to highlight that this is not some new foray for the company. Square Enix have been heavily involved in AI research and development for some time now, with their work bearing fruit in a myriad of projects.
Previous AI Endeavours
The Square Enix AI Division which has existed in its current form since 2020 works to explore possible advancements for artificial intelligence that can later be used in actual game productions. This sits alongside the myriad of other AI-focussed teams and projects that exist across many of the larger video game production companies, be it Ubisoft’s La Forge, Electronic Art’s SEED or the likes of Sony AI as well. While there is justifiable concern about the adoption of generative AI in games, the continued evolution of both traditional (symbolic) AI and deep learning in games productions is an ongoing one, and a process that Square Enix has very much been a part of for some time now.
Like many a large company invested in these areas, the goal is often to find new ways in which AI can be applied and what gains it can have for either game developers or the end user. Whether it’s AI for attempting to improve game balancing for upcoming strategy games, or finding new ways to expand on FEAR’s GOAP technology (which are talks we’ve had showcased at the GDC AI Summit in recent years), or their efforts to improve combat AI in Final Fantasy XV or using deep learning for facial animations in Final Fantasy VII Remake, Square Enix has been involved in this space for some time now.
Though it’s worth highlighting that quite often this technology is only adopted in a game once it becomes evident that it carries some practical use in production. Otherwise it runs risk of impacting the games reputation and success overall. Arguably the most high-profile of Square Enix’s investments into generative AI that’s been shown publicly thus far, was also its most disappointing. “The Portopia Serial Murder Case” released in 2023 was a low-budget demo that was also a remake of the text adventure game from 1983 of the same name. Released for free on Steam as an indicator of how far Natural Language Processing technology has improved since the original game 20 years prior it has failed to leave a positive impression, with a tally of over 450 reviews on Steam summarised as very negative.
Sadly, for a technical demonstration of the use of Natural Language Processing, it’s not doing the best to paint it in a positive light. The ability for characters to respond to text input is incredibly limited, and more often than not they fail to respond in any meaningful way to the players input. The actual big innovation of that demo is that the language models runs entirely locally on the users device. The likes of GPT as well as tools I’ve previously explored such as Inworld or Convai rely on an internet connection given they host their language models online (and then charge you for their usage). Portopia is really a demo to highlight that applications using large language models for NLP is attainable, with a smaller language model somewhere within the ~7GB install size of the game. Sadly, the quality of the model is still an open challenge.
While the research in this space will continue to evolve, and no doubt Square Enix will take that investment money where they can in order to continue, I am less concerned about the subsequent output at this time. Portopia aside, the work that has emerged from the Square Enix AI Division in recent years has been rather impressive. With many of them employed in recent games where players were simply none the wiser.
But their recent public displays may lead them to err on the side of caution. Given if this is the response received from end customers on this kind of generative AI adoption (even for a free to download title), then there future efforts will no doubt aim to better appease their investors. Given frequent misfires like Portopia would ultimately speak ill of their collective efforts.