If this is interesting to you, kinda related, but more theory.
The Markov Property, Chain, Reward Process and Decision Process

And this is why I only spout complete gibberish. Take that, machines!

This works exactly how you predict it would work. At the root of this computing question, there is only one correct answer.

This is not a new problem, or at least the logic for it is not. This is actually a problem given to students in the UTD Comp Sci masters degree courses (because the guy who owns Language Computer teaches at that school and recruits promising students, I assume - I mean, that dude did give my brother his first job so it's not a far stretch).

Back in 05 or 06, my brother (a fellow SEer) gave me this problem when I was a freshman at UTD, before I ran away to join the circus (my brother is rich in money, I'm rich in job satisfaction, we'll see who was right when I die in a gutter).

I told him that I would use the internet (a LARGE data sample) to search for all instances of that word and what words followed it to create a predictive set. In my answer, it wasn't just individual words, but strings of multiple words that were being used as the grounds to find the top, most probable word or phrases that might follow. This, combined with hard-coded information about syntax and grammar, along with learned information about syntax and grammar (this is why the strings of words mattered), could lead to something pretty fluent.

I suspect most of you here could have come up with this logic, the challenge is in the implementation.

I guess I'm a little insulted he was surprised I got the question right, but flattered he even asked.

This is markov models, and some clustering based on friends, shared location, interests, etc. Not as complicated as one might assume, but it requires lots of data, which is why they want all your data, why companies by android keyboard apps, and setup free messaging platforms and don't monetize them, etc.

Maybe he set you up thinking you would fail and he could mock you? No, wait, that is how guys treat other guys, not their sisters.

My brother doesn't treat anyone that way lol He has smart people motives, which are innately atypical from the average person.

Smart people are still people, and, thus, a bunch of bastards.

And usually with exaggerated manifestations of bastard ( -ity? -ness?)

... you bastards know I love my brother, right? lol

Smart people do more of a keeping up with the Joneses type of deal. They don't care if you fail, only that they are relatively more successful than you. How smart they are matters less than how much smarter they are than the next person.

My brother wouldn't be a fair comparison because he's like three and a half years older and was always four grades higher in school because I fell on the other side of the cutoff for kindergarten, so there's an established childhood pattern of him being more capable due to age and grade level difference, an advantage which dissipates in adulthood.

And besides, sibling rivalry is its own special thing. I know my brother respects my intelligence and I respect his. We have different areas of strength but we still beat Portal 2 co-op together, talked about how some of it was hard, and wondered how much harder it would be for stupid people. We also had a great talk about how I use bullets to zero out my transactions in Fallout and bullets are the obvious currency of the post apocalyptic world, not bottle caps.

Whatever our differences, we can agree we're better than the rest of you and that's the true meaning of family ;P

The true meaning of family is getting into fights at Thanksgiving and taking sides in the divorce.