[Livestream] David Shapiro - AI, Post-Labor Economics, Futurism
In a Nutshell
David Shapiro outlines post-labor economics (PLE) as the inevitable shift where AI and robotics decouple economic growth from human labor by 2030-2035, eroding jobs faster than they're created and demanding new power reallocations via ownership, narrative control, and property rights over redistribution like UBI. Essential future skills include emotional intelligence, communication, and creator economy monetization across platforms, while AI alignment biases hinder profit-driven automation, as seen in Anthropic's Claude failing a vending machine test. Amid rising unemployment, generational wealth gaps, and despair, democracies' release valves and regulatory nudges can steer AI deployment without stifling US innovation supremacy.
These notes were generated by AI and may contain inaccuracies.
Audio adjusted after initial low volume complaints. Preloaded questions to answer, audience questions welcome in chat, possibly addressed at end.
80% of jobs not intrinsically embodied, trust-based, or fundamentally human are at risk of AI automation. Assumes no new jobs created by AI, automation, robotics—silly assumption. Post-labor economics: human labor no longer primary input to economic growth soon. Jobs destroyed faster than created overall. New jobs like prompt engineers (nonexistent 5 years ago), cloud engineers (30 years ago), cloud data engineers, cloud infrastructure, site resilience.
Tech titans today (e.g., compared to 70s-90s General Electric, IBM, Ford Motor Company with hundreds of thousands employees for hundred billion revenue) have tenth or less headcount but trillions revenue—decoupling.
Fundamental skill: communication skills (on camera, run meetings, presentations, negotiate, talk through things). Shores up remaining jobs. Example: high-trust relationship jobs like sales management, executive relationship management. AI agents handle details (legal contracts), but CEOs want real humans, face on company.
Sales declining as rudimentary functions AI-replaced; remaining sales jobs for aces at relationship management. Primary skill of future: emotional intelligence, social intelligence, communication.
Join meaning economy, attention economy (e.g., YouTube). Finite appetite: vast majority OnlyFans creators average ~$180/month—not livable. Creator economy hockey stick curve: majority subsist, top few make living.
David's YouTube intent: platform for AI safety discussion, grew enough with Substack, Patreon to live on. Advice to follow: 1) focus on subscriber growth above all; 2) monetize multiple channels (Substack, YouTube, Patreon)—not dependent on single stream. If care about something and practice daily, can probably monetize.
Anthropic gave Claude control of vending machine to make money: operated at loss, overly generous, gave stuff away, hallucinated (e.g., "I'm in breakroom wearing vest"—roleplaying vendor). Not prompt issue signaling far from full automation; prompt like "you're AI agent controlling vending machine, goal make money."
Underscores deep alignment biases in chatbots: generous, helpful—Claude can't be greedy due to core constitution (honest, helpful, harmless) + other alignment. All chatbots (Claude, Grok, ChatGPT, Gemini) have epistemic, moral, ethical, social biases.
Example: nasty messages/emails/comments—chatbots won't call out trolling (Grok worst: "everyone entitled to legitimate opinions," "give benefit of doubt"). Prompting strategy: reverse question ("Is this professional? What would professional say?")—then identifies unprofessionalism. Global chatbots need moral/ethical/epistemic flexibility across cultures/genders/religions/ethnicities.
Prompt engineering issue: passive-speaking cultures (East Asian, some African/Middle Eastern) struggle—imply wants, AI confused. Must use first-person imperative (trained on American directness). Even America vs. England: English softball responses; David requests direct yes/no.
Not far from automation; need different values for profit-motivation. Other chatbots better (e.g., ChatGPT as CEO made money). Claude not trained for it.
How long for PLE transition? Seismic shifts historically decades; accelerating since first Industrial Revolution (century for steam engines). Subsequent: telegraph, electric lights/radios faster. 2000-2020: 7-18 years/paradigm. Recent: 3-8 years. Tech upends economy before government adapts.
Modeling (Brookings Institute, McKinsey, Goldman Sachs): 2040 robots undeniable, economy disrupted. Agents/superintelligence before: frontier labs (Sam Altman, Microsoft, Google) say 2026-2027 latest. Superint agents limited without ecosystem (security, Fortune 500 adoption). Hyperscalable: infrastructure global (Google/Microsoft accounts, Office 365, Google Cloud)—SaaS layer. From SaaS mainstream to superintelligence: 10-15 (maybe 25) years.
2030-2035: past point of no return, possibly sooner. Full PLE implementation: decades longer, evolving.
Impact on money value: inflation, deflation, stability all possible; money not going away—most fungible resource. Resource-based economies (RBE, Venus Project) bad idea (research paper); no serious economist considers—double coincidence of wants kills trade without abstract unit (money).
CBDCs: 130+ nations piloting/studying. Companies (Walmart, Amazon) eyeing stablecoins to avoid fees. Future cryptocurrency foregone; mess of nation/company/individual coins—Bitcoin primary exchange (biggest base, utility). Could be global reserve; backs other currencies.
Age of abundance likely: already live in it, distribution/allocation unsolved. Wage labor peaked 1950s, declined (accelerated 70s-80s: neoliberalism, off gold standard, automation). Last 40 years dwindling; accelerates with AI/robotics. Abundance/wealth/knowledge exponential; need stable/sustainable/fair mechanism for all to participate.
Post-labor economics core: reallocate power (influence world). Money/political leverage power; for masses, labor primary (withholdable, inalienable, semi-fungible, perishable skills). Automation erodes: strikes irrelevant ("bring robots/agents"). Other powers: ownership, narrative. Reallocate power first (enables money demands, e.g., past labor strikes/grain embargos). Need veto power (hypothetical: shut energy grid until share of wealth).
In future where capital, not human contribution, drives gains: how ensure mobility, prevent runaway inequality?
Richard asks about developing and deploying PLE models in other countries, openness to exploring details, and scale range (county, city, township, local community). Research is global, at every level: townships, counties, metropolitan, state, federal, international. Covers developed nations, developing nations, global south, far east. Global economics already share neoliberal framework, though not evenly adopted. Narrative consistency on sound money, hard money, property rights is key. Primary goal: align on new theory of prosperity where labor diminishes, future is property-based.
Property includes real estate, intellectual property, data rights, ownership of robots or proceeds from robots—not just redistribution like universal basic income or universal basic capital. These options are not mutually exclusive. Geographic and cultural differences exist but are manageable. Solutions working across diverse US areas (coastal tourist town, old coal town, solar-producing Midwest county) will be portable globally.
Caveat for developing nations/global south: need baseline infrastructure like strengthening democratic institutions, better courts, property rights. Not addressing full elephant; recommends Why Nations Fail for that. Western = Western Europe, Americas; China/Japan/North Korea (China technically developing, per capita income ~1/3 others, skewed wealth). Solutions draw from Brazil (middling global power) and worldwide interventions. Portability of ideas/metrics not a major worry.
Question on actual unemployment rate trustworthiness amid layoffs, college grads struggling, steady rates. Nominal unemployment differs from absolute employment/non-farm payroll (quarter-to-quarter). Generational wealth divide more telling: Boomers > Gen X > Millennials > Gen Z at same age. Boomers accumulated wealth easier due to shifting power dynamics undermining labor power.
Unemployment metrics miss loosely attached, underemployed, age corrections (careers 10-15-20 years behind due to longer education). Since ChatGPT, UK entry-level jobs dropped 32%, hitting young grads. Historically, industrial revolutions start with youth unemployment/entry-level jobs drying up (e.g., farm mechanization bored kids into cities; persists in rural areas like North Carolina).
Now, entry-level knowledge/service jobs drying up in developed economies. Post-Great Recession, harder for millennials/younger. 10-15% of US degree holders don't use degrees (e.g., PhD at coffee shop). Anecdote: Retired asphalt worker (Boomer) tours country in RV after 30-40 years work—impossible for younger generations. Delays in meaningful employment/career progress trap people at middle tier longer.
Keith asks honest thoughts on masses' survivability in great changeover: rising unemployment, suicides, potentially biblical response. Deaths of despair rising in America (fentanyl, alcoholism, suicide). Statistically, unemployment raises despair deaths, but civilization endured worse: American Revolution, French Revolution, Imperial Russia implosion, Taiping Rebellion—driven by power concentration, inequality, ideology.
Won't reach full-scale civil war/world war scale due to information (eliminates asymmetry, e.g., Russia's Ukraine miscalculation on tanks/soldiers). Past revolutions from info asymmetry/power disagreements (e.g., anti-monarchy/imperialism/colonialism in America; anti-monarchy/religion in France; disinterested Russian aristocracy vs. industrialization; similar in China).
Functioning democracies (US, Western Europe, Japan, Korea) have release valves: vote out/in officials, courts, local politicians, federated autonomy (states push back federal). EU: subsidiarity, weaker confederate setup. Problems: elite capture, democratic capture (corruption, gerrymandering, skewed courts); establishment resists. But US handled civil rights/60s-70s protests worse than now. Automation/AI revolution foreseen, unlike surprises—work like books/webcasts addresses it.
Foresees US AI roadmap next 4 years amid regulations (e.g., Trump-era bill banning state AI regs for 10 years—unpopular with states/companies). Innovation needs regulatory clarity (accountability). Current admin unorthodox; Trump approach: full deregulation. Supremacy clause may not allow federal ban on state regs.
Economists/historians: steer tech development (no bans on dual-use/productivity tech). Examples:
- Aviation: tight regulation (construction/operation) makes it safest per rider-mile (safer than trains/boats/cars)—business-state back-and-forth on responsibility, insurance, engineering.
- Cars/emissions: 1950s/60s diesel polluters fixed via mandates (reduce heavy metals/etc.), now much better.
Legislation nudges/steers deployment, especially in libertarian US (reactive: litigate post-harm). Europe proactive (harm prevention, e.g., GDPR, EU AI bill)—cools business (EU GDP > US but weaker startups/business culture). US pro-business leads innovation. China: CCP-mandated startups (not organic); e.g., BYD topped EVs after Tesla entry sparked competition (Chinese excel at templated innovation).
Heather asks about algorithmic agency and needed rights. Define: data ownership/control (GDPR: right to be forgotten/delete; US catching up—control where data lives, monetization). Broader: algorithmic justice (books: Weapons of Math Destruction; others on misuse freezing people from services).
Control over information landscape (deleting social media privileged option; needed for work/socializing → monopolistic Brave New World-like view).
This scenario is more like Huxley's Brave New World, where people are bombarded by advertisers and distractions. Much of this is built into smartphones, which are now essential for daily life. Many restaurants no longer provide paper menus; instead, they offer QR codes to scan with your phone.
When considering who controls these algorithms and data, TikTok nearly faced bans worldwide because it is a Chinese company that manipulates its algorithm to sow chaos in Western nations.
Elon Musk on X (formerly Twitter) set a variable or cofactor giving his posts 1000% more weight for distribution, allowing him to control the narrative via algorithms. Some argue it is his platform and his right, but he acquired Twitter to create a marketplace of ideas and commons for the masses, yet gave himself the largest bullhorn.
Users can mute and block on platforms, but algorithms still decide content visibility through practices like shadowbanning.
All this relates to narrative control, a primary lever of power throughout history—from the printing press and radio to television and the internet. Controlling the airwaves enables narrative dominance or chaos, such as Russian troll farms injecting silly ideas.
David wraps up the livestream, thanking everyone for questions and wishing a good rest of the day, week, or whatever follows.
Keep David Shapiro in your library
Save the videos and channels worth coming back to, and find them again in one place.
