primerlearning
patreon
Preview clip of the next video on neural networks
🕑 Added 2025-07-09 22:52:56 +0000 UTCHey everybody,
I wanted to share the current version of a scene from the next video. It's one of the first scenes and shows a sort of proto training loop for a single artificial neuron, which is being used to predict how far a projectile will go when shot from a cannon with a given energy. Once the rest of the scenes are sketched out like this, I'll do the voiceover, time the animations to it, and publish.
As a whole, the video will focus on how neural networks work. Along the way, we'll build a network that can control a cannon with the goal of playing golf with mangoes (with the first step being the single neuron in the clip). There will also be some simple classification examples. The script structure and the mango golf cannon are already built, so it's now just a matter of putting the actual video together.
As I'm sure you've noticed, there's a lot of chatter about AI. There are some good explainers out there already, but I think it's a big enough deal where it's worth a slot in the schedule to increase the number and diversity of resources that help people actually understand the technology.
While the focus will be on the nuts and bolts, I also plan to have a section where I address some questions that touch more on philosophy, culture, and economics. I won't have the answers on these, but my perception is that I can make a net contribution to the discourse on the topic. A throughline will be that it's wise to allow yourself to hold uncertainty.
So, if there's a particular question you'd like to me to discuss in the video, whether relating to copyright or jobs or doom, or anything else. Please let me know!
I also want to reference lots videos or other resources where people can learn more. I have a few in mind already that have been helpful for me, but I'm curious to know about your favorite videos on this topic from others, especially from creators I might not already know about.
Thanks as always for your support.
Justin
Comments
Jacob O'Neill
I’m sure you are aware of it, but 3blue1brown’s series on llms is one of my favorites. I think a lot of deep learning based AI feels pretty non-intuitive, so I like a balance of high level enough to draw meaningful conclusions while also low level enough to talk about some meaningful math.Welch Labs has good AI videos in a similar vein as well. I also quite liked the video ‘Things AI Will Never Understand’, though that one’s more of a video essay from a philosophy/linguistic perspective.
Kenn
This is a great preview, and I think it absolutely fits the idea of "Primer" - well-explained, intuitively animated demonstrations of scientific principles and fields through simulation. IOW, this is what I'm here for! As to the question on philosophy, etc., I think everyone's been thinking about things like recent research about AI doom, whether AI is actually thinking, and "wow AI is getting way too real with video now, what's next?" I'd love to see a brief section with your thoughts on how you get from A--the basic explanation the functioning of a single neuron, to Z--what people are seeing as the ultimate evolution to AGI or ASI, and whether/why is it that this "software algorithm" is fundamentally different from every other software we've had for decades.
