It cannot be much lower. The only 'data on vacuum' we have is from Diamond Princess. The mortality rate is 1%, with all the victims being over 70 years old. Half of the sick people did not show symptoms. Half, not 10 times as many seem to think.
Now the good part is that the age distribution is skewed. There are many old people there than normally in the world (around 17% of the infected people there were over 70 years old, compared to around 11% in the UK; only 10% were under 30 compared to 30.4% in the UK). So the most vulnerable category are overrepresented there, while the least vulnerable category is massively under-represented there. What is the real mortality rate is hard to know, but for sure it is lower. In South Korea, it is 0.7% and they are kings of the testing. In Bahrain is 0, and they have done even more testings. It is safe to assume that some people without symptoms (and so not detected) in South Korea exist. If I have to make a guess, I would say that the mortality rate in the wild is probably around 0.5%, maybe slightly lower.
Now the bad part, is that this assumes everyone who needs intensive care gets intensive care. Which won't happen unless there is a containment. If even 10% of the population gets infected within the year (in whatever distribution you model), the medical system totally collapses. It is near collapsing in Lombardy with only 15k cases or so. 10% of the population getting infected means that the number of cases increases 100 times. Let's assume uniform (totally impossible, but I want to be optimistic best case scenario), that is 160k cases per month. People are struggling now, in this best-case scenario, there is 1 ventilator for roughly 11 people. 1 might survive, the other 10 certainly die. This scenario is totally unrealistic, if there is no containment, it will get worse. Much worse. With Gaussian modelling, it is significantly worse (1 or 2 months when we need mass graves, followed by relatively few deaths).
What is the true mortality rate (no intensive care for everyone who needs it)? We cannot know now, but it is certainly over 0.5%, it is over 1%, and it might be over 3% estimated from WHO (in Italy currently it is 8% but that is overestimated, there are not enough tests going on). If the medical system collapses, the mortality rate converges to the number of people who needs ventilators (estimated to be around 5%). If we assume that half of the people do not show symptoms, then that number converges to around the number of WHO. Even if we assume that some people who might have it and show symptoms don't get tested, that number goes down, but not to 1%.
So, while it is true that in an ideal world when everyone who needs treatment gets it, the mortality rate is 0.5% (or lower), without medical support, it will be 1-5%. I think that WHO's estimate might be quite correct.
Data from Diamond Princess*:
https://cmmid.github.io/topics/covid19/severity/diamond_cruise_cfr_estimates.html
* Since then, the number of infected has increased, but not the number of fatalities. There are still 15 people in critical conditions, so the mortality can increase.