Knowledge Vault 2/41 - ICLR 2014-2023
Erik Brynjolfsson ICLR 2018 - Invited Talk - What Can Machine Learning Do? Workforce Implications
<Resume Image >

Concept Graph & Resume using Claude 3 Opus | Chat GPT4 | Gemini Adv | Llama 3:

graph LR classDef economics fill:#f9d4d4, font-weight:bold, font-size:14px; classDef technology fill:#d4f9d4, font-weight:bold, font-size:14px; classDef inequality fill:#d4d4f9, font-weight:bold, font-size:14px; classDef policy fill:#f9f9d4, font-weight:bold, font-size:14px; classDef future fill:#f9d4f9, font-weight:bold, font-size:14px; A[Erik Brynjolfsson
ICLR 2018] --> B[Brynjolfsson: MIT professor,
technology's economic impact. 1] A --> C[Past tech disrupted jobs,
boosted growth. 2] A --> D[AI: pervasive general purpose technology. 3] D --> E[Tech power requires responsible
innovation. 4] D --> F[Techlash: job loss, fake news,
bias, monopolies. 5] A --> G[Tech outweighs trade in
recent inequality. 6] G --> H[Horses replaced by tech,
human labor next? 7] G --> I[Job loss tied to deaths
of despair. 8] G --> J[Tech grows pie, doesn't
guarantee even split. 9] G --> K[Productivity up, median wages stagnant,
gains to top. 10] G --> L[Skill-biased change: educated gain,
less educated lose. 11] G --> M[Tech superstars reap outsized
rewards, concentrating wealth. 12] G --> N[Labor share of GDP down,
capital share up. 13] D --> O[Machines learn what we can't explain. 14] O --> P[Machine learning may impact
higher-wage jobs. 15] O --> Q[No jobs fully automatable,
but most partially. 16] Q --> R[Effects vary: clerical work
vulnerable, creativity resilient. 17] D --> S[AI can augment not
just automate jobs. 18] A --> T[Policies can spread gains:
education, taxes, safety net. 19] A --> U[Businesses can share tech
prosperity more widely. 20] A --> V[Inequality trends global, vary
by policy. 21] T --> W[Education for human skills
may help workers adapt. 22] T --> X[Tax capital gains, not
just labor income? 23] A --> Y[Cheap software capital may
lower interest rates. 25] Y --> Z[Income shifts to innovative
intangible capital creators. 26] A --> AA[Technologists, policymakers, entrepreneurs
must steer innovation. 27] AA --> AB[Next decade could be
best or worst. 28] AA --> AC[Growing wealth and disruption
require shared prosperity. 29] AA --> AD[Society's choices, not just
tech, pick winners and losers. 30] class A,B,C economics; class D,E,F,O,P,Q,R,S,AA,AB,AC,AD technology; class G,H,I,J,K,L,M,N,Y,Z inequality; class T,U,V,W,X policy;

Resume:

1.-Erik Brynjolfsson is a professor at MIT studying the economic effects of information technology, including AI's impact on jobs.

2.-Major technological changes like the steam engine have triggered steep economic growth but also disruption for some workers in the past.

3.-AI is likely a general purpose technology (GPT) on the scale of steam engines or electricity, with pervasive effects across the economy.

4.-With great technological power comes great responsibility - technologists should consider the societal consequences and steer innovation in a positive direction.

5.-There is a growing backlash and "techlash" against big tech companies over issues like job destruction, false news, algorithmic bias, and monopolies.

6.-Economists agree technology has contributed more to recent wage inequality than globalization, though both play a role.

7.-Horses were once ubiquitous in the economy but were largely replaced by technology - could something similar happen to human labor?

8.-Job losses are correlated with troubling social trends like rising "deaths of despair" from drugs, alcohol and suicide among less-educated Americans.

9.-Technology makes the economic pie bigger overall, but doesn't guarantee the gains are evenly distributed - some may lose even as others benefit greatly.

10.-Recent decades have seen a "great decoupling" where productivity continues to rise but median wages have stagnated, with gains flowing to the top.

11.-Skill-biased technical change has increased the demand and wages for highly educated workers while reducing them for less educated workers.

12.-Superstars in many fields can now leverage technology to reach huge markets and reap outsized rewards, leading to more concentration of wealth.

13.-The share of GDP going to labor has declined while the share going to capital owners and investors has increased.

14.-Machines can now learn to do things like image recognition that we know how to do but can't explain - overcoming Polanyi's Paradox.

15.-Machine learning may impact higher-wage jobs like radiology in addition to lower-wage routine work, unlike previous waves of automation.

16.-No occupation is completely automatable yet based on current technology, but almost all have some tasks that could be automated to some degree.

17.-The effects vary by job - those heavy in clerical work are more affected than those requiring manual dexterity, interpersonal skills, or creativity.

18.-Technologists can choose to develop AI that augments and enhances human capabilities rather than only automating and replacing human jobs.

19.-Policies like education, progressive taxation, wage subsidies, and a stronger safety net can help share the gains from technology more broadly.

20.-Private sector entrepreneurs can develop business models that create more widespread benefits and prosperity from technological progress.

21.-Trends of rising inequality are happening across developed countries, though specifics vary based on different policies and starting points.

22.-Reforming education to focus more on uniquely human interpersonal and creative skills may help prepare workers for the changing economy.

23.-Tax policies currently put a thumb on the scale rewarding capital over labor - arguably this should change to better promote employment.

24.-The reasons for long-term declines in interest rates remain an open question and mystery that economists are still debating.

25.-One theory is that cheap, easily replicable software acts as a new form of capital, driving down returns to both labor and ordinary capital.

26.-In this model, income is increasingly shifted to those who can create innovative new "intangible capital" rather than just providing labor or capital.

27.-There is an important role for technologists, policymakers and entrepreneurs in steering innovation to benefit society as a whole.

28.-The next ten years could be either the best or worst decade ever, depending on the choices made - the future isn't predetermined.

29.-Growing overall wealth alongside increased disruption means gains must be actively and consciously shaped to be shared prosperously and equitably.

30.-Technology itself doesn't decide winners and losers - it's up to us as a society to make those choices through action and policy.

Knowledge Vault built byDavid Vivancos 2024