Wasted potential of The Turing Test

Oct. 13, 2019

Gaming Pontification 

I’ve been playing the Turing Test recently. No, not the test of machine intelligence, the game named after it. Basically, the name is kinda fun, but the rest of it is not brilliant. It’s a Portal-esque puzzle game (wander rooms with tool, manipulate stuff and get out of room, even down to the “activate door with powerlines” thing), minus the humour.

Now the plot (such as I’ve seen so far) is that your character (Ava Turing) gets woken up by the base AI saying “I can’t contact the rest of the crew, please go and look” and you have to get through the puzzles to get to them. The dialog then later describes the rooms as “Turing Tests” and claims they’ve been made by the crew for reasons unknown (ominous, etc), and you should get through them to find them. Now, barring the obvious double meanings (tests for someone called Turing, as opposed to machine intelligence tests), they’re not Turing Tests. Not even slightly, and that’s one of the reasons I stopped playing. I’ve however realised since that there’s a much better game idea buried in there…

So, here’s the idea: you have again a sort of puzzle room game. Except, there’s a dumbed-down solver AI for these being applied to them, which after you’ve gone through a few tutorial levels gets activated, and then the two of you solve puzzles together. Basic idea is the solver does all the easy stuff, and then it gets stumped at some point, and you have to start messing around with things until it can see a path through. This could be lots of fun for several reasons:

  1. It gets around the puzzle room problem where you know how to solve a room but still have to go through the motions anyways which is a bit boring.
  2. It improves accessibility, in that you can have precision jumping bits and tight timing, without needing the human to be any good at any of that, as that’s a thing the computer could do.
  3. You can then do fun things like improving the AI over time, so that the computer does more of the basic stuff for you, and you get to focus on the actually hard and new problems. Maybe even a mechanism where the only things it’ll do are stuff you’ve shown it before, which helps with the believability.
  4. It’s actually a new idea in a somewhat overused field (Portal was brilliant, the clones not so much)

I’d probably implement this as some sort of dual-control of a robot possibly where the computer shows ghost patterns in the view of options it’s thinking about, and you can swap over to it having control at any point. Maybe also a visual indicator of “computer thinks it can solve the room now”. Of course, you can always do a bit more if you want, or stop the computer at any time if you disagree with it’s choices, etc. Feels kinda like Centaur Chess but for video games.

Previously: Parochial: working around limits on DLNA players Next: Hacktoberfest 2019