I focus on this bit because I think it makes a great example. Even about specific topic there could be million different things one could find interesting and useful, especially if the context is something that involves a lot of things that were improvised on the spot. This will colossally be influenced by what the GM thinks is 'cool story element' and whatever 'interesting and useful' the GM makes up will greatly influence the direction of the 'story.'
1) Of course the GM is going to influence the the direction of play. It’s not only impossible for that NOT to happen, it’s undesirable.
Influencing play and curating outcomes to achieve
story outcome x are not remotely the same thing. Your reducing the space between the two to nothing. There is a giant, gaping chasm between the two and there is architecture that ensures the space between the two sides.
2) There are not a million different responses to a question like:
“I believe I’ve heard in my studies that there is an ancient dwarven forge around here.”
And there certainly aren’t a million answers that are interesting. And there certainly aren't a million ways that are both interesting and useful. And there certainly aren't a millions ways that are all 3 of interesting, useful, and thematically relevant to the characters. And there certainly aren't a million ways that are all 4 of interesting, useful, thematically relevant to the characters, and following the GM's principles and rules of the game!
And better still, if the player in question feels that it isn't interesting or useful, they have recourse!
"That isn't useful...I don't see how to use that in the situation."
or
"That isn't particularly interesting or relevant?"
To which the GM can then say..."ok what about y or z...are either of those two things interesting?"
To which the player can then say..."I like z...that is very interesting and relates to thing n about my character. But it doesn't appear particularly useful..."
To which the GM can then say..."ok what if these other things a and b are also true...you should be able to use those."
To which the player can then say "yeah, I can definitely use b. Alright so here is what I think we're doing guys."
Play ensues.
3) Your interpretation of the kobold situation is 100 % the sort of adversarial GMing that isn't honoring the player's decision-point + action resolution loop. If they got an outright success, to create that sort of complication is a move that is absolutely hostile to a faithful rendering of play...and absolutely hostile to Skilled Play priorities.
Now, consider an alternative:
If they got a Success With Complications on their move and you did what you did (gave them what they want and attracted unwanted attention in the process), THAT is the appropriate move to make. If you ignore that complication, THAT would be a hostile move by the GM (and a violation of Skilled Play priorities).
Finally, if they got an outright Success on their move and you (a) don't give them what they want (actionable intelligence on the dungeon) and (b) ALSO give them the complication you're envisioning...
...well, that is such a profound Calvinballing of play that you'd be lucky to not endure a complete walk-out on the spot.