Japanese gaming giant Square Enix wants the majority of video game quality assurance and debugging to be handled by some form of generative AI instead of human workers.

Square Enix, the company behind Final Fantasy, Dragon Quest and more, confirmed that it wants to remove the majority of human workers from its QA workflow and replace them with automated processes.

Via VGC, the company has partnered with the Matsuo-Iwasawa Laboratory at the University of Tokyo specifically to bring AI into video game development. The partnership is “aimed at improving the efficiency of game development processes through AI technologies”.

The main push for this is the “Joint Development of Game QA Automation Technology Using Generative AI” project. Currently helmed by a team of ten researchers and Square Enix engineers, the project aims to “automate 70% of QA and debugging tasks in game development by the end of 2027”.

Square Enix says that this initiative will “improve the efficiency of QA operations and establish a competitive advantage in game development”. However, the push to move human workers away from QA work in the games industry has been met with some controversy over the past few years.

QA and debugging is notoriously known as one of the most repetitive and often gruelling parts of game development. However, with generative AI prone to hallucinations and other issues, there’s also the worry that replacing human workers with automated processes will cause major issues to be missed, resulting in buggier games.

Additionally, many QA workers in the games industry eventually become game designers. (For example, Battlefield and Marvel Rivals designer Thaddeus Sasser started working in QA before becoming a level designer and then a game designer.) With 70% of this process removed from human workers, the entry level of the game development pipeline faces being eliminated.

This worry is not wholly tied to Square Enix. In fact, as soon as generative AI tools started to become mainstream head of Xbox Game Studios Matt Booty claimed that replacing QA with AI would increase “how quickly we make content”.

There’s also the issue of how QA features with creative choices. Sometimes, major issues in game flow or gameplay mechanics simply don’t get caught until QA which helps games become better. How can a robot tell if a level is simply a little bit too long?