Hope Returns to the Machine Learning Universe

If you’re not living under a rock, then you’ve surely encountered the Heroes of Deep Learning, an inspiring, diverse band of Deep Learning all-stars whose sheer grit, determination, and—[dare we say?]—genius, catalyzed the earth-shaking revolution that has brought to market such technological marvels as DeepFakes, GPT-7, and Gary Marcus.

But these are no ordinary times. And as the world contends with a rampaging virus, incendiary wildfires, and smouldering social unrest, no ordinary heroes will suffice. However, you needn’t fear. Hope has returned to the Machine Learning Universe, and boy, oh boy the timing couldn’t be better.

As confirmed to us by several independent witnesses, the sun, moon, and stars have been joined in the night’s sky by new, supernatural, sights. After a months-long meticulous investigation, including consultations with NASA, MI6, and Singularity University, we can confirm the presence, on Earth, of the Superheroes of Deep Learning!

Superheroes of Deep Learning

Who are these superheroes, you ask? What makes them so super? Let’s get one thing clear. These ain’t your run-of-the-mill GPU jockeys. The Superheroes of Deep Learning physically manifest the marvels of AI through psychokinetic powers. Can any obstacle stand in their way?

MOOC — Educator by day, vigilante by night.
MOOC — Educator by day, vigilante by night.

With legions of students and quad-rotors at the ready, MOOC never goes into battle alone! Legend has it that in his last appearance on Earth, MOOC narrowly escaped an ambush by The Syndicate of Stanford Statisticians. With the dreaded Lasso constricting, MOOC was abruptly snatched from the jaws of defeat and carried to safety by a swarm of drones. Then, following his expert demonstrations, the drones acquired paranormal fighting skills via imitation learning. The Statistics Department has never recovered from the smashing defeat that followed!

Continue reading “Hope Returns to the Machine Learning Universe”

Macro-causality and social science

Consider a little science experiment we’ve all done, to find out if a switch controls a light. How many data points does it usually take to convince you? Not many! Even if you didn’t do a randomized trial yourself, and observed somebody else manipulating the switch you’d figure it out pretty quickly. This type of science is easy!

One thing that makes this easy is that you already know the right level of abstraction for the problem: what a switch is, and what a bulb is. You also have some prior knowledge, e.g. that switches typically have two states, and that it often controls things like lights. What if the data you had was actually a million variables, representing the state of every atom in the switch, or in the room?

Continue reading “Macro-causality and social science”