Update, September '19
I'm not sure this is the right problem for me anymore. According to a supergroup of MIT academics organized by MIT's president that has been studying the problem for the last two years, AI is more likely to lower the average quality of jobs than replace them altogether. My experience labeling data (and hiring people to label data) corroborates this.

--- past thoughts below ---

Unemployment, yikes
I'm worried that we're headed for an unemployment crisis and that we're underprepared for it. The only solutions I'm aware of are education (retrain displaced people into productive work) and socialism (basic income). I'm interested in any related sociology, economics, public policy, organizational research, and examples of companies trying to solve the problem from various angles.

At my last job I tried higher education reform at the the Minerva Project, where I helped build futuristic classroom software based around active learning.

I don't think that's quite the right problem anymore, though. The problem isn't knowing how to learn (and it's certainly not a lack of educational content): it's that most people have no idea what to do. They have incomplete information and so the job market is incredibly inefficient and they make terrible employment choices. That's why - and I promise this isn't a plug, you'll just have to trust me - I work at Sourceress on an AI job matchmaker. We're trying to develop the ability to quantify a) a job's requirements b) a person's skills and c) the gap between the two. Imagine being able to see at a glance the gap between yourself and a job you want, and being able to see other people that have surmounted a similar jobs gap in the past and how they did it. Imagine a recruiting process that wasn't biased against minorities, or that tried to estimate your potential instead of just your experience, or that was interpretible instead of opaque. We've made solid progress on all of these fronts, and we've got solid product-market fit, but we've got a long way to go.

How might we help the structurally unemployed? Job retraining programs, like bootcamps or trade schools, can be time- and cost-effective, especially when they make money by taking a cut of the first year's salary of their students' jobs. Part of the reason I'm such a believer is because they worked for me: Dev Bootcamp changed my life's trajectory in 2013 (nowadays I recommend Lambda School). There are also programs that provide people with a door into home healthcare jobs; home healthcare is the only occupational category projected to grow faster than software engineering according to the Bureau of Labor Statistics thanks to our aging population. Increased ability to make lateral career moves and a better understanding of one's options would go a long way towards making the job market more efficient.

What other kinds of work are vulnerable to automation? A 2013 Oxford study stated that 47% of US jobs are vulnerable to automation over "the next one or two decades". High risk occupations: "most workers in transportation and logistics occupations, together with the bulk of office and administrative support workers, labour in production occupations, a substantial share of employment in services, sales and construction occupations, and sales occupations like cashiers, counter and rental clerks, and telemarketers". Jobs with a low risk of automation include "most management, business, and finance occupations... most occupations in education, healthcare, as well as arts and media jobs" due to their requiring "generalist tasks requiring social intelligence".

What technology related to finding and evaluating people seems inevitable? Great question!

What technology related to learning employable skills seems inevitable? Also a great question!

How and why does quantifying people go wrong? I'm in the market for case studies.

Will the bootcamp model survive and/or spread to other, non-tech industries? There's been a consolidation in recent years due to very low barriers to entry and a lack of regulation creating some outright fraud, but the model is fundamentally sound.

When and in what form will UBI take and save our collective bacon? It's easy to lose hope these days, but the ACA was a great step, and many experiments are under way.
Updated: September 28th, 2019

Sources, notes:

Key: ✍️ is a blog post, 📘 is a flashcard, no emoji means notes & highlights.
His 2020 Campaign Message: The Robots Are Coming September 20th, 2018
Why the Republican Party wins when robots take your job September 15th, 2018
Notes on "Self-Driving Trucks Are Going to Hit Us Like a Human-Driven Truck" July 5th, 2018
A list of recent predictions about driverless cars July 5th, 2018
Council for Opportunity in Education » Press Releases May 7th, 2015
How Bobby Jindal Could Decimate Louisiana's State Universities May 7th, 2015
Opinion: Why competitive model fails schools. No one should lose in education. | Get Schooled April 26th, 2015
Remarkable turnaround for Regis - The Boston Globe April 26th, 2015
Push, Don’t Crush, the Students April 25th, 2015