#201: Anthropic vs. Pentagon Round 2, AI Job Impact Study, Services as the New Software & GPT-5.4 (1h 25m)
ai-autonomous-weapons ai-driven-innovation-economy ai-governance-laws ai-in-workforce-disruption ai-literacy-public-awareness ai-monetization-strategies ai-surveillance-privacy post-work-ai-society transhumanism-ai-enhancement
- Release date: 2026-03-10
- Listen on Spotify: Open episode
- Episode description:
The data is in, and it's harder to ignore. Anthropic's new "observed exposure" study reveals AI can handle 94% of knowledge work tasks in theory and the gap between theory and reality is narrowing fast. Paul and Mike unpack what that means for your career, your company, and the broader social contract around work. This week: the Anthropic vs. Pentagon saga escalates, a Sequoia partner predicts AI will replace entire service industries (not just software), GPT-5.4 drops and outperforms professionals on economic benchmarks, a Polish mathematician has his "personal singularity" moment — plus rapid fire on AI journalism, OpenClaw security risks, AI copyright law, and Meta's smart glasses privacy scandal. Show Notes: Access the show notes and show links here Click here to take this week's AI Pulse. Timestamps: 00:00:00 — Intro 00:04:32 — AI Pulse Survey Results 00:07:00 — Anthropic vs. US Government Round 2 00:19:43 — Anthropic Analyzes AI Job Impact 00:35:30 — Services as the New Software 00:49:19 — Barriers to Enterprise AI Adoption 00:54:54 — GPT-5.4 01:00:55 — The Move 37 Moment for Math 01:05:39 — AI and Journalism Update 01:10:03 — NVIDIA CEO Calls OpenClaw “Most Important Software Release Ever” 01:13:35 — AI Art Can’t Be Copyrighted 01:16:08 — Meta Sued Over Smartglasses Privacy Concerns 01:19:21 — Microsoft Copilot Cowork This week’s episode is sponsored by our 2026 State of AI Report. This year, we’re going beyond marketing-specific research to uncover how AI is being adopted and utilized across the organization, and we need your help to create the most comprehensive report yet. It’s a quick seven-minute lift. In return, you’ll get the full report for free when it drops, plus a chance to win or extend a 12-month SmarterX AI Mastery Membership. Go to smarterx.ai/survey to share your input. That’s smarterx.ai/survey Visit our website Receive our weekly newsletter Join our community: Slack Community LinkedIn Twitter Instagram Facebook YouTube Looking for content and resources? Register for a free webinar Come to our next Marketing AI Conference Enroll in our AI Academy
Summary
- ⚔️ Anthropic-Pentagon Clash: Escalating dispute labels Anthropic a supply chain risk for rejecting military AI uses, yet Claude powers combat ops, blending ethics, dependency, and politics amid court challenges.
- 📊 White-Collar Job Exposure: Anthropic data shows AI poised for 94% of knowledge tasks but at 33% now, hitting high-paid grads hardest; Gen Z hiring slows as reality nears theory.
- 🚀 Autopilots Eat Services TAM: Sequoia foresees AI replacing $6T services (vs $500B software) in accounting, consulting; finance openings crash 75%, outsourcing first as models gain judgment.
- 📉 Negative Sentiment & Uneven Adoption: AI polls at -20, worse than politicians; firms split by leadership—some race ahead, others blocked by risk aversion despite consulting pushes.
- 🧠 GPT-5.4 Superhuman Leap: Crushes benchmarks, solves elite math creatively; Microsoft Cowork agents follow Claude, urging paid tests as exponentials hit all knowledge work.
Insights
Will ethical stances against AI in autonomous weapons and surveillance lead to government blacklisting of leading AI firms like Anthropic?
Time: 6:42 – 18:38
Category: AI Governance & Laws, AI Surveillance & Privacy, AI & Autonomous WeaponsAnswer: Anthropic faced a unprecedented ‘supply chain risk’ designation from the Pentagon for refusing military uses of Claude, despite its active deployment in combat operations, escalating tensions with federal agencies while highlighting dependency on single providers. This conflict underscores the clash between AI safety principles and national security needs, potentially forcing companies into legal battles or compromises. Both sides appear open to deals, but political elements and public fallout complicate resolution. (Start at 6:42)
Is public sentiment turning against AI faster than its benefits can be realized?
Time: 15:34 – 19:43
Category: AI Literacy & Public Awareness, AI in Workforce DisruptionAnswer: An NBC poll showed AI with a -20 net sentiment score among 1,000 voters, worse than Trump or the Republican Party and only better than the Democratic Party or Iran, amid growing negativity on data centers. Hosts predict negative effects like job loss will hit visibly first, while positives like scientific discovery lag, fueling a ‘negative narrative wave’. This could influence midterms and policy as politicians gauge AI’s electoral role. (Start at 15:34)
Are highly educated, well-paid white-collar workers the first to face massive AI exposure in jobs?
Time: 19:44 – 27:06
Category: AI in Workforce Disruption, AI-Driven Innovation EconomyAnswer: Anthropic’s study using real Claude usage data reveals a gap: AI could theoretically automate 94% of knowledge worker tasks but currently does 33%, hitting programmers (75%), customer service, and data entry hardest, with exposed workers earning 47% more and holding more degrees. Early signs include slowed Gen Z hiring in exposed fields, signaling workforce shifts despite no broad unemployment yet. This task-level analysis shifts focus from IQ benchmarks to real labor impacts. (Start at 19:44)
What happens when AI breaks the traditional social contract of work, education, and prosperity?
Time: 27:21 – 35:25
Category: Post-Work AI Society, AI in Workforce Disruption, AI Governance & LawsAnswer: Hosts discuss how AI disrupts the implicit deal where education leads to stable jobs and fulfillment, questioning who captures productivity gains, retraining obligations, employer duties, and if jobs remain central to security. Ideas like automation taxes, universal basic services, or billionaire-funded UBI emerge, but no solutions exist amid fears of trillionaires controlling society. This calls for policy, shared gains, and human oversight to maintain dignity. (Start at 27:21)
Can AI ‘autopilots’ turn software companies into trillion-dollar services giants by replacing human labor markets?
Time: 35:30 – 48:07
Category: AI-Driven Innovation Economy, AI in Workforce Disruption, AI Monetization StrategiesAnswer: Sequoia partner predicts AI shifts from copilots to autopilots handling intelligence tasks autonomously, targeting $6T services spend (10x software) in insurance ($140-200B), accounting ($50-80B), consulting ($300-400B), starting with outsourced work for easy swaps. As models gain domain judgment via data, they commoditize professions; finance job openings already collapsed 75% since 2022 peak. This expands TAM to all knowledge labor, accelerating disruption. (Start at 35:30)
Why is AI adoption so uneven across companies, leaving many blocked by IT and leadership risk aversion?
Time: 49:19 – 55:07
Category: AI in Workforce Disruption, AI Literacy & Public AwarenessAnswer: Wharton prof notes divides where regulated firms deploy AI freely while others face IT/legal blocks or case-by-case approvals, boiling down to executive risk tolerance; hosts cite ‘Law of Uneven AI Distribution’ holding since 2023, with finance/healthcare still blocking access. Leadership AI literacy is key, as bottom-up efforts fail without top urgency, creating massive value gaps. Consulting alliances aim to bridge this for enterprises. (Start at 49:19)
Has GPT-5.4’s superhuman benchmarks signaled the ‘Move 37’ moment for knowledge workers everywhere?
Time: 55:08 – 65:32
Category: AI in Workforce Disruption, AI-Driven Innovation Economy, Transhumanism & AI EnhancementAnswer: OpenAI’s GPT-5.4 exceeds humans on OS World (75% vs 72.4%), matches pros 83% on GDPVal across 44 occupations, solves Frontier Math Tier 4 problems with creative insight, per skeptic mathematician’s ‘personal singularity’. Hosts urge testing paid models, warning 3-year exponentials mean urgency beyond programmers, as even math PhDs embrace AI collaboration. This accelerates scientific discovery and job rethinking. (Start at 55:08)