#141 – Erik Brynjolfsson: Economics of AI, Social Networks, and Technology

Episode Summary

Executive Summary: Eric Brinjolfsson argues that exponential technological progress is reshaping society faster than human institutions can adapt, creating both immense gains and serious risks. The conversation covers Moore’s law, AI, autonomous vehicles, GDP mismeasurement in the digital economy, social media’s distortion of truth, inequality, taxation, UBI, pandemic-driven remote work, and the need for policies that spread technological benefits more broadly.

Main Topics: Exponential growth and human intuition (Priority: 5/5): Brinjolfsson explains why exponential change is hard for people and institutions to understand, using COVID-19 and Moore’s Law as examples. He argues that digital technologies make the world increasingly exponential, while human instincts remain linear. AI, automation, and the future of work (Priority: 5/5): The discussion surveys how AI is likely to reshape tasks and occupations rather than eliminate all work. He emphasizes that machines are strong at structured tasks, while humans retain advantages in creativity, empathy, and complex social judgment. Productivity, GDP, and the value of free digital goods (Priority: 5/5): He argues that GDP and productivity undercount the value created by zero-priced digital products like Wikipedia, Facebook, Zoom, and Twitter. He proposes alternative measures such as GDP-B to capture consumer benefits more accurately. Platform design, misinformation, and truth (Priority: 5/5): Brinjolfsson warns that social networks often amplify outrage, lies, and 'nutpicking' because their incentives reward engagement over accuracy. He argues platforms should be designed to favor truth, while users and leaders share responsibility. General-purpose technologies, reinvention, and the J-curve (Priority: 4/5): Using electricity and self-driving cars as examples, he explains that major productivity gains often require years of reinvention after adoption. Early stages can even reduce measured productivity before the upside appears. Policy: taxes, inequality, UBI, and shared prosperity (Priority: 4/5): He critiques tax systems that favor capital and under-tax bad externalities, supports carbon and congestion taxes, and prefers earned income tax credits over unconditional UBI as a way to preserve work, dignity, and incentives to reskill. Pandemic effects and remote work (Priority: 4/5): COVID accelerated remote work dramatically, with large portions of the labor force shifting online. Brinjolfsson expects much of this change to persist, though it will deepen inequality between knowledge workers and others.

Key Arguments: Exponential change is counterintuitive; linear instincts fail when systems double repeatedly. Technological progress is not destiny; society can design institutions and products to amplify truth and shared prosperity. AI will mostly restructure work by automating tasks, not eliminate all human employment soon. Digital goods create real value even when their market price is zero, so GDP misses much of modern welfare. Social platforms are often optimized for engagement, which rewards outrage and falsehoods over accuracy. General-purpose technologies typically have a long lag before productivity gains appear because organizations must reinvent themselves around them. The current tax system over-taxes good activity like labor and under-taxes harms like pollution and congestion. UBI can provide safety, but work also supplies meaning, identity, and social structure; reskilling and wage supplements may be better tools. The pandemic showed that remote work can scale quickly and may permanently change how knowledge work is organized. A strong public investment role remains essential for basic research because private firms cannot capture most of its benefits.

Data Points: Early COVID case growth: A few dozen cases doubling every 2–3 days - Used to illustrate exponential growth and why Brinjolfsson became alarmed in early March Remote work share before pandemic: About 15% (one in six) of Americans - Estimated working remotely at the beginning of 2020 Remote work share during pandemic: Roughly 50% - Share of Americans working from home after COVID hit Occupations analyzed: 970 occupations - US O*NET dataset used to assess machine learning suitability Radiology tasks: 27 distinct tasks - Example showing that AI can automate some, but not all, parts of a profession Productivity trend comparison: Past 15 years lower than previous 15 years - Brinjolfsson cites weak productivity growth despite rapid technological progress Long-run horizon for AI: 50–100+ years - He speculates machines may do most economically valuable work over very long horizons Self-driving fatalities: About 30,000 annually in the US and about 1 million worldwide - Used to argue human driving is already far from perfect and machines must do better Electrical factory productivity lag: ~30 years - It took decades after electrification for factories to fully reorganize and see gains Battery/energy improvement concept: 'Coome’s Law' of halving energy consumption - He describes energy efficiency as a more important metric than raw chip speed for many uses

Pivotal Quotes: "The greatest shortcoming of the human race is our inability to understand the exponential function." — Eric Brinjolfsson (citing Albert Bartlett): Introduced while discussing COVID, Moore’s Law, and why people underestimate rapid change "Technology doesn’t shape our destiny. We shape our destiny." — Eric Brinjolfsson: Said while discussing social media, misinformation, and the need for conscious platform design "The big gains only came once smart entrepreneurs and managers basically reinvented their industries." — Eric Brinjolfsson: Explaining why electricity only transformed productivity after factories were redesigned

Implications: Listeners should expect faster AI-driven change but also long adjustment lags. Policy, platform design, and education will determine whether gains become broadly shared or deepen inequality and misinformation.