For the people viewing this thread, I had a long email exchange with Mieszko the other day. There’s a bunch of stuff changing in the AI codebase so I wanted to know where was safe to start doing tutorials and where I might want to hold back a bit (or do something myself). The short answer is that EQS is definitely fair game for tutorials. APPerception is likely to change a bit, but probably worth doing now. Hotspots aren’t ready to check into main yet, so no point in starting to think about that yet.
You also might not know, but Epic recently hired John Abercrombie (AI guy and lead from Bioshock Infinite), so they now have some more AI experience, although I suspect John will be more involved as a lead rather than an AI guy.
As a community we really should all be giving Epic a hand to help evolve the AI toolset a bit more. I feel really bad for Mieszko because its hard being the one guy doing what a small team should be working on. Especially given his involvement with Fortnite and all.
I’ve got plans to incorporate some of my own research into the engine for everyone to use. Either as plugins or as a pull request. But I feel like there should be some more community effort involved in actual AI “tools” rather than specific technologies (i.e. whenever anyone talks about AI they usually talk about decision making systems of some sort, rather than design tools for making better AI).
Which also reminds me! Did you know that John Abercrombie did the AI for Swat4? My game Police State is sort of mass effect meets Swat 4 So I asked him how the Swat4 AI was created and he told me it was a lot of designer donkey work!
So yes, a lot of AI tutorials incoming. But I’m going to try and get some time to work on a set of designer-oriented tools to extend the engine a bit. Maybe test them out as plugins then submit them as a pull request into the main engine. Just not quite enough time in the day (plus I’m coming down with another bloody cold right now).