@Volume
Unfortunately human factors is not yet an independent science, it is covered by a lot of different disciplines these days (ergonomy, neurology, psychatry...), they do not produce a lot of output useful for aircraft designers or training programme providers yet.
My impression is that human factors is not a science as such, and wonder if it should be. It is an umbrella label that covers a lot of different subjects (just like 'structures engineering' could cover loads, stress, stiffness, fatigue, etc).
When you dig deep into specific aspects of accidents you run into multiple factors. So you have to study them all and use the published knowledge.
Human factors indeed do not tell you what and how to design, or who to listen to. But you can and shall certainly make use of knowledge of all these aspects under that label while working on designs in design teams. Design teams consisting of people with a lot of different specialisms. I would expect that multiple people in these teams have knowledge of multiple human factors aspects.
A good designer tries to read as much as possible on as many aspects as possible. And uses and applies that while designing. I would say that modern aircraft are not and can not be designed without serious knowledge of a range of especially these human factors aspects. I have certainly seen a lot of that knowledge being applied to designs that happily fly around today.
To further improve on that you would need a lot of detailed information. And that may well be the real challenge.
What could be very useful to engineers to go beyond what they can do now, is information that is not directly available today. In such cases the engineer would not be interested in the specific airline or pilot as such. He just would like to see if pilots (new, bad, average, expert) act and react in the way the design (by lack of deeper knowledge) assumes. What he can use is information based on pilots interviews, and he of course uses test pilot inputs. The engineer would also like to have data that supports pilots statements and is accurate enough for design input (see Volume's post). In quite a few cases you have data that is available to airlines but not to design engineers. In this AirAsia case we see that KNKT investigators withhold CVR information that would certainly have been studied under human factors for years to come. Which is shocking.
There are multiple reasons why the ideal flow and type (think about video of instruments or even pilots) of information is not directly available. Every party that delivers information would wants to be sure that the information is only used for the intended purpose, and cannot be misused or even used against them. Which is hard nut to crack indeed.