Welch Labs — searchable YouTube transcripts & AI Q&A

The most cited paper of the century is a brilliant hack

This video explains how the ResNet paper introduced residual connections, a simple hack that allowed for much deeper neural networks and solved the vanishing gradient problem.

Can LLMs make significant new discoveries?

This video explores if LLMs can make new discoveries by examining how their knowledge is limited by training data cutoff dates.

Help us pick t-shirt designs!

Welch Labs asks viewers to vote on t-shirt designs featuring AI, math, and computer science concepts.

Do AI models really need to be generative?

This video questions if AI models truly need to be generative, arguing that internal representations learned from tasks like next token prediction are key for intelligence.

Imaginary Numbers are Real.

Explains how imaginary numbers arise from the equation x^2 + 1 = 0 and the fundamental theorem of algebra.