Confused about AI research as a means of addressing AI risk

I was listen­ing to https://​​pod­cast/​epi­sodes/​paul-chris­ti­ano-ai-al­ign­ment-solu­tions re­cently, and found it very helpful. How­ever, I still have a ques­tion at the end of it.

What is the plan that or­ga­ni­za­tions like OpenAI have to pre­vent bad out­comes from AGI? From how Paul Chris­ti­ano frames it, it seems like it’s “cre­ate AGI, and make sure it’s al­igned.”

But I don’t un­der­stand how this plan ac­counts for com­pe­ti­tion. To use a stupid anal­ogy, if I was con­cerned that cars in Amer­ica weren’t safe, I might start my own car com­pany to man­u­fac­ture and sell safer cars. Maybe I spend a lot of time en­g­ineer­ing a much safer car. But my efforts would be for naught if my cars weren’t very pop­u­lar (and hence my com­pany wasn’t very suc­cess­ful), even if they were ground­break­ingly safe.

It seems like this lat­ter part is most of the trick, at least in the do­main of cars.

I’d like to un­der­stand in more de­tail how this anal­ogy breaks down. I can imag­ine sev­eral ways, but would love to hear it di­rect from the horse’s mouth.