Are American police biased to use lethal force against African-Americans?
Probably yes on average as of the mid-2010s.
A study of one undisclosed, high-crime large American city in the early 2010s used "the as-good-as-random variation in officer assignment to 911 calls" to determine that White officers were disproportionately much more likely than minority officers to use their guns when dispatched to predominately minority neighborhoods, but this study may not apply to other cities, had some minor methodological problems, and didn't control for the possibility that minority locals might respond more forcefully towards White officers.
A 2015 meta-analysis of experimental simulations found that officers were equally likely to shoot unarmed Black and White targets, but more likely to shoot Blacks in other situations, with the racial bias estimated to be 19% as large as individual differences in readiness to shoot.
Other studies have attempted to systematically control for context to look for nationwide racial bias. One study of ten cities found no racial bias in officer-involved shootings; another looking at 2015-2018 nationwide data found that Blacks were 15-25% more likely than Whites to be fatally shot in similar situations. Two older studies of nationwide datasets argued for anti-Black bias, but relied on relatively crude attempts to control for context. A study with multinomial regression on nationwide statistics claimed that officer race did not independently predict the race of fatally shot civilians, but had a critical methodological flaw.
Simple benchmarking tests suggest no overall anti-Black bias in fatal shootings, with some tests suggesting anti-White bias. However, this methodology is particularly dubious.