How We Teach Machines to Think (And Should We Be Afraid?)
In 2012, Google spent millions of dollars to teach a computer how to recognize… cats. It sounds absurd, but that experiment became a turning point for modern machine learning.
In this video, I explain in simple terms:
what machine learning actually is (and how it’s different from traditional programming)
why this idea has existed for over 70 years, but only started working recently
how neural networks learn in a way that’s surprisingly similar to the human brain
why scale (data + compute) suddenly changed everything
what cats, YouTube, and Nvidia have to do with it
No formulas.
No technical jargon.
Just stories, examples, and visual metaphors.
At the end a bit of Star Trek philosophy and a question that doesn’t yet have an answer: Are we building incredibly powerful tools or something more?
⚠️ Note:
This video is intentionally simplified to make the concepts accessible. If you want slightly deeper dive with more technical detail and nuance, read the extended article: / how-we-teach-machines-to-think
Keep in mind: machine learning and neuroscience are both complex fields, and any comparison between them is necessarily incomplete.
_________
About the channel:
Hi! I’m a data scientist, and on this channel I explain complex ideas from technology, business, and psychology.
Only the essence — drawn by hand.
“Let’s live. Let’s search. Let’s become better.
Write books, draw posters, be knowledgeable and meaningful.”
— Oleksandr Polozhynskyi, Tartak (Ukrainian band)
________
Research mentioned in the video:
Google (2012) — Unsupervised learning of visual representations
https://static.googleusercontent.com/...
The “Jennifer Aniston neuron” study — neurons responding to specific people and concepts
https://www.nature.com/articles/natur...