Sorry, you asked "why one might." I gave two reasons: (a) actual direct evidence of threat in one case vs absence in another, and (b) incentives to lie. There are certainly reasons in favor in the AI takeoff threat, but that was not your question :). I think you are trying to have an argument with me I did not come here to have.
In case I was not clear, regardless of the actual state of probabilities on the ground, the difference between asteroids and takeoff AI is PR. Think of it from a typical person's point of view. Tyson is a respected physicist with no direct financial stake in how threats are evaluated, taking seriously a known existing threat which had already reshaped our biosphere more than once. EY is some sort of internet cult leader? Whose claim to fame is a fan fic? And who relies on people taking his pet threat seriously for his livelihood? And it's not clear the threat is even real?
Who do you think people will believe?
I think I replied before reading your edit, sorry about that.
I'd say that Tyson does have incentives for popularizing a threat that's right up his alley as an astrophysicist, though maybe not to the same degree as MIRIans. However, assuming the latter may be uncharitable, since people joined MIRI before they had that incentive. If the financial incentive played a crucial part, that dedicating-their-professional-life-to-AI-as-an-x-risk wouldn't have happened.
As for "(AI takeoff) likely isn't possible", even if you throw that into your probability ...
If it's worth saying, but not worth its own post, even in Discussion, it goes here.