Data | Philosophy Tube

The direction you took it in was really engaging! V good content

This video is insanely good. I love the format, since it doesn't ever stop and become Olly talking about things at the camera, it's a great intro to Breadtube/video essays. Several uninitiated friends have already enjoyed this video!

Format is brilliant. Love this. You make philosophy and critical thinking so accessible and relevant! You're my favorite parasocial relationship. :D

Great video and such a good song choice at the end!

While most of the things you have said are true, I'd like to add that not all models are black boxes. In fact, traditional models are easily interpretable, because they represent our understanding of reality (i.e. what we expect to happen under certain conditions). In other words, traditional models are hypothesis-driven. Most modern AI models are black boxes, because they are not created to embody a hypothesis (or a set of hypotheses).

That was *brilliant*.

I was considering for quite a while but this video pushed me over the edge

crustybatteryacid

I pledged because of this video

Markus Kitsinger (SwooshyCueb)

Excellent video! Hard to beat a good Socratic dialogue! Nice subtle dig with the €50 note ;)

John D Ruddy

Yeah so my name in the credits made me realize that Patreon has my full name... whoops

Olly, I love all of your videos, but I'm so impressed by the creative risk you took with this one. Also much love from a fellow depression buddy - hope you get feeling a bit better soon.

“I used to be a police officer!” ~ Greatness.

Sascha Erni

Hey Olly, this video was great! Sorry to learn from the credits captions that you're going through depression, hope you can find support and get through it. But the video was amazing, it really works as a fiction short, it's cool, the people waiting in line in the background really add something. People are great props, I guess? Also loved seeing Ellis back too, his ominously creacking leather jacket is a great sidekick. Video was great, it's an ongoing joy to support you work. Love.

Alien6981

Great work with the research on this one! I work in AI and I'm impressed how well you understand how it works. I and some of my colleagues have been thinking about the societal impact of the technology that we help build for a while now and some of the discussions in your video sound very familiar. But I still got several new perspectives and arguments out of it that I hadn't thought of before. I'll share the video with my colleagues, so we can discuss it internally. Minor nit: The things that you call "algorithms" that recognize things in pictures or make decisions or predictions are called "models" in the AI community. I know, they're usually called "algorithms" in literature targetted at mainstream audiences (and using that same terminology probably makes your video more accessible to such audiences). But I did want to clarify the difference here because it's somewhat relevant to one of the points you're making. An algorithm is a set of instructions written by a human programmer, so that programmer would be able to tell you exactly how the algorithm works and what each part does. A model is more like a black box because it's essentially a very complex mathematical function that has been automatically constructed by a computer (through machine learning), so it's often difficult or impossible for humans to tell which part of it serves what purpose. (There are models that are easier to interpret for humans, but they typically don't do as well on the predictions or decisions they're supposed to make, so they're not used a lot.) Your argument still holds though because regardless of what you call these black boxes, the fact that they're black boxes is a problem.