I am immediately suspicious of any scientist (or expert in a subject) who doesn't say "well, actually, it's sort of complicated..." because it always is. Everything is complicated and confounded by innumerable factors, and requires years of study to even see the full shape of.
But, politics and media do not currently thrive on presenting complexity, so if you're going to run the risk of asking a scientist their opinion, you either select scientists who are willing to dumb down the science, or you ignore everything they say that doesn't match what you already believed.
I have a cognitive bias poster on my wall that categorizes about 1/4 of researched biases as "Too Much Information" [1].
In the context of Democracy, the trouble with "actually, it's sort of complicated", is that there is absolutely no way that all citizens can approach everything this way and still have time for, e.g. the Pursuit of Happiness.
In other words, trust (and trustworthiness) is key. We must delegate to someone, whether expert or not.
Traditionally, organizations with cult-like properties envelop people in a kind of "information bubble" that makes deciding who to trust a tractable problem.
The culture of science does this to some extent, but is unable to compete well for a number of reasons.
Yes, but delegation is always accompanied by principal-agent problems (like moral hazard). Lobbyists ensure that all but the most morally unimpeachable representatives will inevitably place moneyed interests before their electorate.
But, politics and media do not currently thrive on presenting complexity, so if you're going to run the risk of asking a scientist their opinion, you either select scientists who are willing to dumb down the science, or you ignore everything they say that doesn't match what you already believed.