There's also a lot of libs/services he can use
And if he needs help in understanding or implementing something, he can always ask for some help from someone @ FB...
Everything after that is a matter of typing correctly and adjusting design choices to unexpected constraints.
Having the knowledge of FB's engineering at his disposal will certainly make things easier.
The guys of wit.ai are now in the Facebook M team, I've read somewhere last month (can't remember the source though)
If he can achieve his goal will mean that a lot of people could do that and there is nothing special in that project. You can argue that he has more access to data than you and me but several people working at FB have the same access as him. His simple AI will be... yes, simple.
I think that's kind of the point of his personal challenges. He's trying to set an example of being intellectually curious, not show off that he's a genius. Wasn't one of his previous projects for the year to learn Chinese? A difficult challenge, but not a superhuman one. Plenty of other people are capable of it.
There are things to dislike about Zuckerberg, but I don't think these challenges are one of them. They're pretty admirable imo.
He says "simple", so I would say being able to understand and execute "turn off the lights in the second floor bathroom" probably counts. I'm sure he can do something like that in a year (even though there are programmers who can't).
I think a lot of programmers are under the impression that he is this idiot who knows a little PHP and got lucky building a website. But I'm sure he is probably better than the vast majority of working programmers out there, on account of building the initial versions of Facebook (and I say this as a working programmer). People just don't want to believe that because he seems douchey in many ways.