Four tech billionaires walk into a mansion. It sounds like the setup for a punch line, but it also forms nearly the entire conceit behind "Mountainhead," a savagely entertaining but somewhat shallow new satire written and directed by Jesse Armstrong, the creator of "Succession." The film, which is streaming on HBO's Max, is a sort of chamber play, its stage a modernist castle in Utah--the Mountainhead of the title--overlooking snowy peaks. The players are a quartet of friends, or, more accurately, frenemies, who resemble a mishmash of real-world Silicon Valley founders. Steve Carell plays Randall Garrett, the group's Peter Thiel-esque mentor who, not unlike the late Steve Jobs, has cancer that his doctor tells him is incurable.

For the quartet of tech billionaires in Jesse Armstrong's Mountainhead, ideas are so powerful that nothing else seems real. Holed up in a resplendent snowy retreat built by meditation-app developer Hugo Van Yalk (Jason Schwartzman), they're glued to their phones as the outside world is erupting into chaos, thanks in no small part to the wildfire spread of A.I. deepfakes on the social media platform owned by the world's richest man, Venis Parish (Cory Michael Smith). People in Gujarat are being burned alive after being falsely accused of desecrating religious symbols, and Midwestern Americans are machine-gunning each other over minor disagreements, but for these four men, the widespread devastation is in some ways proof of concept that they're as important as they believe themselves to be. And besides, those bodies going up in flames are just images on a tiny screen, so distant they might as well be theoretical. As he trudges through the snow with Randall (Steve Carell), the venture capitalist who serves as the group's self-appointed philosopher king, Venis asks him, "Do you … believe in other people?"

We describe an unconventional line of attack in our quest to teach machines how to rap battle by improvising hip hop lyrics on the fly, in which a novel recursive bilingual neural network, TRAAM, implicitly learns soft, context-dependent generalizations over the structural relationships between associated parts of challenge and response raps, while avoiding the exponential complexity costs that symbolic models would require. TRAAM learns feature vectors simultaneously using context from both the challenge and the response, such that challenge-response association patterns with similar structure tend to have similar vectors. Improvisation is modeled as a quasi-translation learning problem, where TRAAM is trained to improvise fluent and rhyming responses to challenge lyrics. The soft structural relationships learned by our TRAAM model are used to improve the probabilistic responses generated by our improvisational response component.