Chris's AI Deep Dive

Ever notice how some people have that magical ability to make you feel heard, seen, and genuinely interesting, whether you’re pitching in a boardroom or flirting on a date? These freaks of human interaction are what Charles Duhigg calls “Supercommunicators.” (I’m pretty sure he coined it, but let’s be real—Tony Robbins probably trademarked it in the ’90s.) What's their secret to being world-cl...

This is the paper that kicked it all off. The Genesis chapter of AI. Today, we’re diving into Transformer architecture. Sure, we’ve all heard about tokens and vectors—yawn—but did you really get what multi-head attention does? Spoiler: it’s not just an AI with commitment issues. #AI #google https://arxiv.org/abs/1706.03762

What's that name for that crazy mathematical theory where a drop of water on Laura Dern's hand causes genetically engineered dinosaurs in Costa Rica to go on a killing rampage? In this episode, we’re talking about a final tool in the data scientist’s toolkit—a tool so mystical, every recent PhD attempts to apply it to show off how well rounded they are, but few can actually convince a VP with an M...

Panel Data is basically like having a time machine for statistics, but minus the DeLorean.

Did you know cigar smokers have a higher mortality rate than cigarette smokers? And people driving a Subaru WRX with tinted windows have a 100% chance of vaping? It’s science. In this podcast, we tackle the issues confounding causal inference and offer a simple solution: matching. It’s my go-to tool because we can crank out quick and dirty directional results in 24 hours and hand them to the CEO—w...

I went to Jack-in-the-Box after the latest minimum wage hike, and let me tell you, the franchise fast food business model is falling apart. It took over 20 minutes to get my Jumbo Jack® Cheeseburger because they had one poor employee running both the drive-thru and the counter. Multitasking, sure, but not in a good way. This podcast dives into a data science technique called Difference in Differen...

Your eyes aren’t playing tricks on you. One of these men is the Rock, the actor. The other? A wax version of the Rock, aka a “Synthetic Control,” crafted to look like… well, the first guy. Personally, I prefer the real Rock—especially in that disaster movie where California plunges into the ocean thanks to a massive earthquake—but I digress. This podcast dives into an equally obscure data science...

Who would have thought Andrej Karpathy, the mastermind behind Tesla’s “it’s not advanced cruise control, it’s autopilot,” wrote his PhD dissertation on Barack Obama? Check out the podcast to learn more. #Tesla https://cs.stanford.edu/people/karpathy/main.pdf

Instrumental variables are a fancy technique data scientists use to prove causality—I think I spend about 50% of my time in meetings reminding executives "correlation isn't causation" (I'm VERY popular). But let’s be real, you’ve got about as much chance of finding a naturally occurring “instrumental variable” as you do of spotting Bigfoot. That’s why A/B testing is so powerful—i...

MrBeast has racked up over 300M subscribers on Youtube and is known for pushing the boundaries on innovation, as well as kids' waistlines with his overly fatty "Feastables" chocolate bars. Today, we deep dive into his recently leaked internal document, a top-secret recipe for going viral that’s spicier than the time he gave away an island. Who would have thought he was obsessed with analytics?

Let's talk about my favorite topic: myself, in the third person https://www.proquest.com/docview/1237283080

Dmitri A. Dolgov is the CEO of Waymo. This podcast explains his PhD dissertation from 2006 ( Computer Science Department, University of Michigan ). https://ai.stanford.edu/~ddolgov/dolgov06phd.html