Topic Name: Voices of VR
Bio: Voices of VR is a podcast about VR started by Kent Bye. Bye interviews game developers, enthusiasts, and technologists, especially about virtual reality. Episodes are usually around 30 to 45 minutes long, but some interviews can go upwards of an hour plus. Bye travels to different events all across the country to see what's going on in the industry and interview people and talk about what he sees.
Name of Episode: "#623: Training AI & Robots in VR with NVIDIA's Project Holodeck"
Date of Episode: February 2018
Describe Artwork, Article or Topic: Kent Bye talks with Omer Shapira about robots, artificial intelligence, and the project at SIGGRAPH 2017 where a robot is demonstrating its learning from the Holodeck by playing dominoes with people.
Opinion of Artwork, Article or Topic:
For the technology, most is old news to me - this is all pretty basic ideas in computer science regarding artificial intelligence and machine learning. The technology is not what fascinates me: it is what the robot is a proof-of-concept OF that intrigues me. Shapira talks about AI as a tool, and this robot learning as a kind of tool to understand how it learns given certain inputs. As is referenced multiple times throughout the podcast, "garbage in, garbage out" (which is even funnier given low-level programming language about the idea of "garbage collection," basically getting rid of unneeded data). So, the idea that in building a tool, if it isn't being abused or people aren't trying to break it, then it's not a good tool (~10 min) is really interesting to me. Shapira talks about this also in context of trying to find weird behaviours in the robots, you don't want something to go awry, particularly for something that is "mission-critical."
Shapira also talks a lot about what a game engine is, which as a developer myself I've also tried to explain the role to other people as to what a game engine does for me. He describes it as "somewhere between a media framework and an operating system." (~11:40) What VR is is a simulation - AI and ML are also simulations, and game engines run these things beautifully. They have the thinking power like an operating system would, but they're also an interface for media and I think I'm going to have to think a lot more about what a game engine does for me before I subscribe to what Shapira says, but it's certainly a different way to view a game engine other than its role as exclusively something where one builds games.
Finally something else I got really excited about in this interview that Shapira brought up was what will make tech more human, and that is the recognition of hands. As is described, it is a more delicate form of input, since the input we have today are exaggerated (exaggerated motions, simplified movements, etc) - interaction design for hands in game engines have never been dealt with. Shapira speculates that "the current generation of game engines has to go through a transition of becoming gentler, softer, more precise and more delicate." It is my hope that technology will not require bulky hardware that simplifies so well what humans do.
Linked below is a TED talk I watched many years ago which completely changed the way I viewed technology, and why much of what I do and build now is designed and built for the human body. Let's start with hands.
Links:
https://www.ted.com/talks/sean_follmer_shape_shifting_tech_will_change_work_as_we_know_it