We’re not thinking about AI too much - it’s too far away on the horizon to seriously consider while we’re busy focusing on getting the game into alpha, beta, and doing our cert submission prep. Some of the more senior leadership have talked about it in that kind of long-term planning sort of way and there’s been some preliminary evaluation on that front, but the general attitude from my studio leadership (and the leadership of the franchise I’m currently working on) is that AI is interesting but the potential legal pitfalls from using AI trained on large internet-scraped data sets for things like asset generation have yet to be addressed… and we’re far too well-known and visible a franchise to risk that kind of legal trouble knowingly.
Our primary goal in evaluating AI as a development tool is for the same reason we develop tools and automated assistance to help artists create assets - anything that helps automate the annoying and busy work that artists have to do frees up more dev time for them to do the creative stuff that we want them to do. Instead of having to create normal and spec maps by hand, maybe we can automate and improve the process so that the artists can spend their time creating more cool assets and not have to handle the fiddly busy work. If an AI can help by pre-filling in values and setting things up for them to review and adjust, that’s a lot faster than having to do it by hand. A large portion of our tool engineering is targeted at those kinds of problems.
Got a burning question you want answered?