Taking a peak behind Hidden Door's AI

30.08.2023 - 14:57 / gamesindustry.biz / Ai

Generative AI is the speculatively disruptive trend of the moment, inheriting the title from the metaverse and blockchain gaming before it. And while narrative gaming start-up Hidden Door just surfaced last year with a pre-seed funding round, CEO Hilary Mason had years of experience working in machine learning and AI beforehand.

And while she clearly believes in the technology's potential – as evidenced by co-founding Hidden Door in the first place – she finds some of the hype around the field of late overblown.

"I'm a bit of a pragmatist in this space," she tells GamesIndustry.biz. "I do believe the technology can be incredibly useful as a tool, and a lot of the applications I see being developed right now are essentially productivity tools around, 'Can we more efficiently do the things around games and entertainment that we're already paying somebody to do?'"

That approach can work very well if the model in question already contains every possible version of the content you're looking to create, she says. As an example, she suggests AI-powered writing of weather reports could work because whatever the day's weather is, it has likely happened before somewhere at some point.

"But when you do see weather that is novel, which unfortunately we're seeing too much of right now, it has no idea what to do," she says. "That as a principle can be applied to the kinds of content work where this technology can actually be useful.

"If you're writing something that has been written before and there's no really unique human insight that you need to pull out of it, it can be a useful tool for that. But if you're doing something that requires some sort of insight in a broader context, some sort of extrapolation as to what could be useful or reasonable, you're going to hit certain limits.

"That's a long way of saying I think there's a lot of hype right now, and a lot of angst derived from hype that is not matching the reality of what the systems are actually capable of, or the work you have to do to make them useful practically."

That angst she alludes to is the concern about AI costing people's jobs. We suggest that the angst is coming not from a fear that AI could do these jobs better than humans so much as it is coming from how eager some employers seems to be to replace humans with algorithms and large learning models.

"Yes, and that's really the thing," Mason says. "They are eager to replace you with something they have never actually seen work in that way. Because it's a lot of energy based on hype and sales pitches that I don't think we've seen substantiated yet… Even if those systems catch up – and there's a very fair question in terms of where they will have impact on certain kinds of work and labor