I would put it this way: apart from really impressive experience and knowing very well what goes on in people's minds (of course in a different way than the neurologists do), my mother has never paid for a speeding ticket, because she knows how to talk to a policeman to avoid the fine, discovering all his weaknesses, fears, frustrations in 5 seconds - honestly, I'm not exaggerating! She get's paid for two hours of work what many people earn by working the entire month.
She usually deals with CEOs and other top-management guys who, for instance, are extremelly good and efficient in what they do, but their subordinates cannot stand the pressure the boss is creating, or to put it straight - the managers are practically monsters. And so she comes in and turns a Steve Jobs into a Wozniak, if you know what I mean.
Think for a second of a police negotiator who persuades a terrorist not to blow himself up, or a suicidal person not to jump out of the window.
I know I have used the examples remotely related with therapy, but I wanted to give you a broader picture of what this artful science (yes!) is coupable of.
There are of course certain limits to what can be achieved and an experienced psychologist is able to define if a person needs to consult a psychiatrist to get medication, or if recovery is at all possible - the goal of the therapy can be differently defined, sometimes it is just to make the patient "stable", but healing is not possible.
To give you an example, while meeting my mother at the mental institution where she worked, I once stumbled upon the head of the drug addictions department whom I asked what is their "success rate" in therapy. She asked me: "What do you mean?" - "How sure you can be that a person is not going back to the addition", I replied.
"We are sure that almost all our patients sooner or later would return to drug abuse". I was shocked, but this is the realistic, frank approach.
To sum it up let me just mention that when my dear mother was doing PhD she had a very distinguished supervisor, one of the best specialists in the country, for many years the chairman of the National Psychologists Association (not in US/UK obviously), and one of his most important books was titled something like "The therapeutic aspects of therapy", and his conclusion was "talking to people change them, but it is yet to be discovered how and why".
A psychiatrist (in NL) first goes through regular medical school (5 years) and then another 4 years of study + on the job study for psychiatrist. (maybe there's one year overlap because I seem to remember it's 8 years total)
While maybe an intelligent person can figure out a lot of things for themselves--even if we ignore the fact that it's harder to figure out without a bit of distance--surely somebody (also intelligent) who's studied for ~8 years in this very field, will be better at spotting and making such assessment than a very intelligent person that trained their intellect in some other field.
If I had a big and important carpenting job, I'd hire a skilled and trained carpenter as well, even if I could probably do and/or figure out most of it myself. Over some level of complexity, I wouldn't even try, I'd hire a carpenter. Add to that, "figuring it out for yourself" becomes much harder if the subject is "yourself" instead of "my house/furniture".
From what I understand, psychiatrists in the US are a bit more like psychologists here. But with 4-5 years of study and who knows how many years of experience, they're still pretty good at it.
One thing though, a very important red flag (IMO), if after a while with a therapist you really get the idea "I could do this better, this person doesn't really get me", then get a new one. Therapists come in many shapes and sizes, and often it's important that you get one that can at least match your intellect to some extent, or it may become hard to trust them (if you can't help but second-guessing them). There will be therapists that can deal with this, talk about particular skeptic feelings and explain it in a way that is acceptable to you. But if they keep setting off all those alarm bells, you're not seeing one of those.
I'm sure they work for some people, but there are so many bad and underqualified ones who basically just sit there and listen to you without offering much in the way of insights (but sometimes offering up insults, in my experience). I think a bad therapist can be really damaging, which is unfortunate. People in these sorts of situations don't really need any more things going against them.
edit: I'm in the US, so that's where I'm coming from. When I reread this I realized you're referencing a system outside of the US.
To use a programming related metaphor, it's a bit like rubber ducking (http://c2.com/cgi/wiki?RubberDucking), where a deep and difficult abstract problem can suddenly cease to be such during the mere process of explaining it to a stranger. Suddenly the faults in your own reasoning become obvious, and the very act of sharing what's troubling you makes it more concrete and approachable.
As for this:
> Surely an intelligent person can see what their questions are getting at and influence the therapist's inferences accordingly, in the same way as you can when taking those silly multiple choice psychological assessments?
If you're seriously seeking help, why would you bother doing this? You're going to a therapist because you've got problems you think might be helped by talking them through with someone. If you're just interested in proving how clever you are or showing the superiority of "training in rigorous disciplines", there's cheaper ways to do it that don't waste other people's time.