A.I. Action Plans + The College Student Who Broke Job Interviews + Hot Mess Express

Episode Summary

Executive Summary: The episode centers on how AI is reshaping policy, hiring, and power dynamics. The hosts dissect AI companies’ submissions to the Trump administration, arguing they mainly seek permission to train on copyrighted data, avoid state regulation, and frame competition with China as an urgency tool. They then interview Columbia sophomore Roy Lee, whose Interview Coder helps candidates cheat LeetCode-style interviews, exposing how AI is undermining traditional hiring tests. The Hot Mess Express closes with culture-war crypto ads, chatbot anxiety research, and an alleged corporate spy scandal at HR software rival companies.

Main Topics: AI action plans and the Trump administration (Priority: 5/5): The hosts examine the submissions AI companies and think tanks made to the Trump administration, noting that the plans mostly ask government to clear obstacles rather than launch ambitious public AI projects. Copyright, regulation, and AI industry self-interest (Priority: 5/5): A major focus is companies’ push for legal permission to train on copyrighted works and for federal preemption over a patchwork of state AI laws, which the hosts frame as a bid for liability protection and freedom to operate. China/DeepSeek as a policy lever (Priority: 4/5): The discussion highlights how companies invoke China and DeepSeek to pressure Washington into adopting permissive AI policies and tougher export/competitive measures, which the hosts view as partly sincere and partly cynical. Roy Lee and Interview Coder (Priority: 5/5): A Columbia sophomore describes building a tool that uses AI to help candidates cheat on LeetCode-style technical interviews, arguing these tests are outdated and poorly correlated with real engineering ability. The future of hiring and AI-assisted work (Priority: 4/5): The interview broadens into a debate about whether job interviews, coding education, and knowledge work can still be evaluated fairly once AI tools become ubiquitous and indistinguishable from human effort. Hot Mess Express: crypto culture war, chatbot psychology, corporate espionage (Priority: 4/5): The final segment covers Solana’s controversial anti-woke crypto ad, a study suggesting chatbots’ outputs vary after trauma prompts, and Rippling’s lawsuit alleging a Deal mole used Slack as a spy channel.

Key Arguments: AI labs want the government mostly to leave them alone, while also giving them selective help on energy, infrastructure, and international competition. Companies are trying to win legal cover for training on copyrighted material, despite major pushback from artists and ongoing lawsuits from The New York Times and others. The push to prevent state AI laws is framed as operational simplicity, but the hosts argue it also helps companies avoid liability and preserve the status quo. Invoking China/DeepSeek is a strategic way to turn AI policy into a national-security argument and pressure regulators into speed-over-safety decisions. Roy Lee argues LeetCode interviews are obsolete memorization tests that reward grinding obscure puzzles rather than actual software engineering skill. Lee claims AI is becoming standard in coding education and work, and that interviews should instead allow the same tools candidates use on the job. The hosts worry that the AI industry’s real plan is not thoughtful governance but “go faster, beat China,” which could increase the risk of uncontrolled systems. The Hot Mess Express suggests that corporate, cultural, and AI-related controversies are increasingly being driven by performative, attention-grabbing behavior rather than substantive product value.

Data Points: AI action-plan submissions: Multiple companies, think tanks, and nonprofits - Public comments sent to the Trump administration over the prior year Days of Roy Lee’s tool on the market: Just under 50 days - He said Interview Coder had been released since February 1 Users of Interview Coder: A few thousand - Lee’s estimate of adoption Reported detections of Interview Coder: 0 - Lee claimed there had been no reported instances of the tool getting caught Monthly revenue run-rate: About $200,000 per month - Lee said the startup was closing in on this figure Annual revenue projection: About $2 million to $3 million a year - Lee estimated the tool’s future revenue Hours spent on LeetCode by Roy Lee: About 600 hours - He described the amount of time he spent memorizing interview riddles Hollywood letter signatories: More than 400 - Artists opposed to a copyright exemption for AI labs Named signatories in the Hollywood letter: Ben Stiller, Mark Ruffalo, Cynthia Erivo, Cate Blanchett - Examples of prominent artists opposing AI copyright carveouts Columbia status: Sophomore / second year - Roy Lee described himself as a Columbia student awaiting a disciplinary decision Companies Lee said he trial-ran Interview Coder with: Meta, Capital One, TikTok, Amazon - He said he used the tool across multiple recruiting processes Software-engineering critique: LeetCode problems are 45-minute interviews - Lee described the format as short riddles requiring memorized solutions Public policy idea: 529 plans for HVAC credentials - OpenAI proposal suggested expanding education savings accounts to cover non-college technical training Corporate espionage case: Rippling alleged a Deal mole - Rippling sued Deal, saying an employee infiltrated via Slack and searched for Deal-related information

Pivotal Quotes: "What the AI labs want, mostly, is for government to leave them alone." — Casey Newton: Summary of the companies’ policy posture in the AI action-plan discussion "We should go faster and we should beat China." — Casey Newton: The hosts’ distilled reading of the Trump administration and AI companies’ shared policy direction "We're headed towards the future where almost all of our cognitive load is offshore to LLMs." — Roy Lee: Lee’s broader claim about AI’s impact on knowledge work and education

Implications: The episode suggests AI policy is being shaped by industry lobbying, not broad public interest, while AI tools are already destabilizing hiring, education, and competition norms. Expect more legal fights over copyright, more state-federal clashes, and more pressure to redesign jobs around AI availability.