In Iceland they tested 10% of the population randomly (pulled names from phone book.. who still has a phone book?). One of the takeaways was that of the people who tested positive, 50% were asymptomatic at time of test.
We can use this to extrapolate that "confirmed cases" in Iceland should be doubled to get actual cases.
Using that... we learn that in the US instead of 1.6m confirmed cases we probably have 3.2m actual cases. Thus our actual death rate is 3%.
This is the first widespread sampling done in the world. (10% is a large sample). Not sure how Iceland compares to the US. But what we do know is that in Asia and Europe the 3% rate holds fairly constant. In the US, due to better care we might see that closer to 1% when all the actual data comes out... but we won't know for certain until years down the line. But right now, the data points to a 1% or greater fatality rate