iRacing partners with Microsoft Research to benefit on-track AI performance

Artificial intelligence is currently all the rage, and machinations about its role in society are ongoing.

But AI has been used in some form or another in racing games for decades, with gamers generally taking it for granted.

The current advances, however, are slowly seeping into sim racing with hopefully positive effects. Take Sony AI’s Gran Turismo Sophy project, for example, which uses a learning agent to deliver realistically reactive opponents within Gran Turismo 7.

With that in mind, popular PC simulation iRacing has announced a partnership with Microsoft Research to formulate a new project.

Its aim is to create Large Action Models (LAMs), deep learning processes designed to understand and interpret human intent and turn them into autonomous actions.

Whether or not such research can then be translated into iRacing sim improvements sounds like it could be much further down the road, although the duo states an intent to at least publish the findings, presumably as a paper.

The membership-based racing service theorises that it “can provide a foundation to power advanced functionality in the iRacing sim, including further improvements to AI-based drivers, AI-based coaching, crew chiefs, and many other possibilities.”

iRacing’s AI rival performance is already up there with the very best, and it is hopeful that the conclusion of this collaboration is models that its developers can use to enhance the virtual racing experience.

It is claimed that former IndyCar driver Oriol Servia initiated the collaboration and will stay on the project as a “collaborator and domain expert.”

“Our goal is to leverage the realism and quality of our existing product, build upon our market-leading existing single-player opponent AI, and learn from the collaboration with Microsoft Research to create new and improved features that will benefit all iRacers,” said iRacing president Tony Gardner.