Episode Summary
Executive Summary: Noah Gift traces an unconventional path from teenage TV/film editor to AI practitioner, then explains how machine learning was used at Score Sports to identify influential athletes, predict page views, and power growth, marketing, and product decisions. The conversation expands into sports analytics, emphasizing that real-world ML success depends more on data collection and labeling than model training, and highlights promising future uses in athlete scouting, UFC strategy, nutrition, and personal AI assistants.
Main Topics: Noah Gift’s unconventional path into AI: He describes working in television and film as a teenager, then moving through Caltech, startups, and sports tech, with AI interest sparked by people and problems he encountered along the way. Machine learning at Score Sports: Gift explains how Score Sports used data from social platforms and Wikipedia to identify influential athletes, predict page views, and drive user growth and monetization. Real-world ML is mostly data engineering: He stresses that collecting, cleaning, matching, and validating real-world data consumed far more time than model training itself, especially with noisy social identity data and API issues. Sports analytics beyond marketing: The discussion extends from influencer marketing to predicting player quality, wins, and undervalued talent using social engagement and sports performance metrics. Untapped ML opportunities in combat sports, training, and nutrition: Gift points to UFC/BJJ submission sequencing, periodization for track athletes, and nutrition forecasting as areas ripe for experimentation and predictive modeling. Pragmatic AI and applied cloud ML: He closes by describing his upcoming book, Pragmatic AI, which focuses on hands-on, production-oriented AI examples in Python and cloud platforms.
Key Arguments: The hardest part of applied machine learning is usually defining the problem and building reliable data pipelines, not choosing the algorithm. Median engagement metrics are more useful than raw follower counts because outliers and bots distort averages. Wikipedia page views can be a surprisingly strong proxy for public attention and correlate with social engagement. Influencer signals from athletes can be used to predict page views, acquisition cost, and campaign value. Fans may collectively identify high-value players earlier than conventional stats do, making social engagement potentially useful as a leading indicator. Real sports organizations already use machine learning on motion-tracking and performance data to gain tactical advantages. There is substantial opportunity to apply ML in athlete training, nutrition, and health monitoring because these domains are complex and non-linear. Future AI assistants should help with decisions humans are bad at, such as sleep, health, driving, and nutrition planning.
Data Points: Age when Noah Gift started in TV/film: 10 years old - He said he worked in television and film starting around age 10. Year period of ABC News editing work: 1992–1993 - He recalled editing national news stories for ABC around that time. Pay as teenage editor: $500/day - He described his salary as a teenager working in TV editing. University degree: Bachelor’s in nutritional science from Cal Poly San Luis Obispo - He said this was his undergraduate major. Traffic spike from Brett Favre post: ~5 million page views in an hour - A Brett Favre NFL draft prediction drove a massive traffic surge on Score Sports. Organic reach decline for athletes: 1%–3% - He cited low organic reach on Facebook as a reason for Score Sports to build its own channel. Conor McGregor engagement growth: ~7,000x in four years - He described McGregor’s engagement growth as evidence that follower counts are misleading. Average likes per Conor McGregor post at partnership start: ~50 likes - He referenced McGregor’s early engagement level in 2014. Current likes per Conor McGregor post: ~4,000–5,000 likes - He contrasted current engagement with early numbers to show growth. Monthly active users on platform: tens of millions - He said the platform reached sustained tens of millions of MAUs for a period. Cost per user acquisition: ~$0.80 per user - He said the company acquired users at roughly eighty cents each. LeBron James wins impact: ~23–25 games - He cited LeBron as an outlier in wins added to a team over a season. NBA salary heuristic: ~$1 million per point - He described a rough market tendency to pay players by scoring average. Two most valued NBA teams: Los Angeles Lakers and New York Knicks - He said both are worth several billion dollars despite poor recent records.
Pivotal Quotes: "The whole 80-20 thing of cleaning the data and collecting it takes all, for me, is absolutely the truth in a real world." — Noah Gift: He was explaining that data collection and cleaning dominated the machine learning workflow. "Follower counts are really meaningless because of the bots and people buying things." — Noah Gift: He argued that engagement is a better signal than raw follower totals when identifying influential athletes. "I do think that is the future of AI." — Noah Gift: He was responding to the idea of AI assistants that forecast health, sleep, and behavior outcomes.
Implications: The episode argues that practical AI wins by solving messy data problems and extracting actionable signals from real behavior. For sports, that means better scouting, marketing, and strategy; for consumers, it points toward personalized forecasting tools for health, sleep, and daily decisions.