one thing he said was that he didn't consider belief a binary as in that you either believe something or don't. He viewed all beliefs as a continuum. You can believe one thing 10% and another thing 90%, but he wouldn't let me pin him down as to whether he "believed" any particular thing or not.
That seems pretty reasonable. The only thing I believe 100% is that my consciousness exists in some way. I'm about 99.9% certain that reality is roughly as I experience it (I have a physical body, the things I witness correlate to an external world, I'm not a brain in a box or in some kind of simulation, etc.). Every other belief carries some higher degree of uncertainty.
I think of how much evidence I'd need to believe something. If someone told me their dad was childhood friends with Bill Clinton, I probably wouldn't believe them, but all it would take to convince me would be a yearbook and a couple old photos. If someone told me they had a tea party with Sasquatch, they could show me a video and I'd still assume it was faked.
This seems like a healthy perspective, to me. The problem pops up when you start assigning high confidence levels to unlikely claims, or spend too long obsessing over low confidence claims. I suppose aliens could run the government, but even 10% confidence is way too high.