They have? That's the approach they are using? Because that doesn't mesh well with practical reality. Where AGI use Deep RL, it's to improve on vision tasks like object classification, none of them are making any driving decisions - that seems to remain the domain of I guess you could call it discrete logic.
B. R. Kiran et al., "Deep Reinforcement Learning for Autonomous Driving: A Survey," in IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 6, pp. 4909-4926, June 2022, doi: 10.1109/TITS.2021.3054625.
I haven't read the paper, so this is not a reading recommendation. Just posting as evidence that there is work in the area.
