A flurry of articles in December and January assess the state of artificial intelligence

From Erik Brynjolfsson et al, optimism about productivity growth:

“Economic value lags technological advances.

“To be clear, we are optimistic about the ultimate productivity growth fueled by AI and complementary technologies. The real issue is that it takes time to implement changes in processes, skills and organizational structure to fully harness AI’s potential as a general-purpose technology (GPT). Previous GPTs include the steam engine, electricity, the internal combustion engine and computers.

“In other words, as important as specific applications of AI may be, the broader economic effects of AI, machine learning and associated new technologies stem from their characteristics as GPTs: They are pervasive, improved over time and able to spawn complementary innovations.”

Full post at The Hill here.

On her blog, Sarah Constantin writes about the pace of advances in five areas: games, image recognition, speech recognition, machine translation, and natural language processing. She concludes:

“General AI systems, as far as I know, do not exist today, and the million-dollar question is whether they can be built with algorithms similar to those used today, or if there are further fundamental algorithmic advances that have yet to be discovered. So far, I think there is no empirical evidence from the world of deep learning to indicate that today’s deep learning algorithms are headed for general AI in the near future. Discontinuous performance jumps in image recognition and speech recognition with the advent of deep learning are the most suggestive evidence, but it’s not clear whether those are above and beyond returns to processing power. And so far I couldn’t find any estimates of trends in cross-domain generalization ability.”

Link to the full post, where many commenters provide further reading.

  • Denny Britz has a more technical year-in-review at WildML, including a list of favorite courses/lectures on deep learning and reinforcement learning, and a list of favorite datasets for supervised learning. Link.
  • Wired emphasizes how AI-generated content is becoming indistinguishable from the real thing. “‘If you were to see a picture of me on the moon, you would think it’s probably some image editing software,’ Sotelo says. ‘But if you hear convincing audio of your best friend saying bad things about you, you might get worried. It’s a really new technology and a really challenging problem.’” Link.
  • Stanford’s 2017 AI Index. Link. ht Margarita


Historian of economics Beatrice Cherrier questions contemporary economists’ reluctance to consider ethical questions

“As [Al] Roth’s presentation moved from mundane to life-saving achievements, his slides became loaded with ethical statements. Moneyless kidney exchange is ‘fair,’ he argued. Global kidney markets shouldn’t be considered ‘exploitative.’ Yet he never saw fit to discuss the ethical underpinnings of the designs he advances. Is it because, as explained in his introduction, his address focuses on ‘practical’ marketplace design rather than on the identification of the theoretical properties of mechanisms? Or because the underlying ethics is obvious – isn’t raising the number of kidney transplants a universally accepted policy end? Or because he believes that the opposition to his international kidney market scheme betrays an unwarranted sensitivity on repugnance? Elsewhere, he has argued that economists should not take for granted citizens’ repugnance to engage in some kind of transactions (money landing [sic] or prostitution are other historical examples). That society has banned markets for such transactions has sometimes harmed their welfare, and Roth believes it is possible to carefully design such markets in a way that commodification and coercion won’t happen.

“My puzzlement is twofold. As a historian, I find economists’ contemporary reluctance to get their hands dirty with ethics (Roth is not alone in this) highly unusual. For, contra the popular received view, those economists usually considered the founders of contemporary economics like Paul Samuelson or Kenneth Arrow were constantly arguing about the ethical foundations of their discipline, in particular welfarism. And as an observer of a changing discipline, I fear that this ethical shyness may at best prevent them from communicating with their public, at worst backfire. Let me elaborate.”

Full post on Cherrier’s blog here.

  • In a follow-up tweet, Cherrier mentions a few economists who are seriously grappling with these issues, Matt Weinzierl and Glen Weyl. See the abstracts of their recent work here. Weinzierl on welfare: “I propose and formalize an argument for why economists working in the welfarist normative tradition should include nonwelfarist principles in how they judge economic policy.” Weinzierl’s paper is available free here. Weyl on voting: “We propose a simple design that offers a chance: individuals pay for as many votes as they wish using a number of ‘voice credits’ quadratic in the votes they buy.” A technical examination of this idea is here.


Free-riding, Right To Work, and the feedback effects of hampered union power

“Free-rider problems were nearly fatal to the Union under the Articles of Confederation, as Alexander Hamilton observed. The notion ‘that a sense of common interest would preside over the conduct of the respective members, and would beget a full compliance with all the constitutional requisitions of the Union,’ was disproven by ‘that best oracle of wisdom, experience,’ as contrary to ‘the true springs by which human conduct is actuated.’ […] Despite their common interests, each member ‘yielding to the persuasive voice of immediate interest or convenience has successively withdrawn its support, till the frail and tottering edifice seems ready to fall upon our heads and to crush us beneath its ruins.’

What was true of the Union is also true of unions.”

From an amicus brief filed to the Supreme Court in the case Janus v. American Federation of State, County, and Municipal Employees, Council 31. The case seeks to overturn the court’s 1977 decision in Abood v. Detroit Board of Education.

The brief skewers the “forced rider” argument—that workers are required to pay union dues against their will—which underpins both this lawsuit and Right To Work legislation more broadly. In its place, they offer an overview of free-rider problems—with citations ranging from Hume to Mill to Friedman to Mancur Olson—and evidence of the negative effect of RTW laws on public-sector union membership. Disaggregating the connection between sentiment regarding unions and participation in them, the brief holds that it is the presence of free-riders, rather than any underlying anti-union sentiment, that explains the reduction of union membership in RTW jurisdictions.

Read it in full here.

  • A new working paper from economists at NBER, SIPA, and Brookings examines the immediate and downstream political effects of Right To Work laws: decreased turnout, decreased Democratic vote share, fewer working-class candidates in down-ballot races, decreased liberalism in state policy. Link.
  • This research joins a wide field of studies on the effects of Right To Work laws, the majority of which focus on economic effects. A 2015 briefing paper from the Economic Policy Institute examines wage differences between comparable RTW and non-RTW jurisdictions. Link. A 2011 paper in the American Sociological Review finds “the decline of organized labor explains a fifth to a third of the growth in inequality—an effect comparable to the growing stratification of wages by education.” Link. A 2007 paper finds more business in RTW states but lower wages, employment and personal income in non-RTW states. Link. Another EPI paper found negative effects for business in Oklahoma following RTW legislation. Link.
  • The new NBER paper was written up in The Nation by Demos’s Sean McElwee, who reads its results as a wake-up call for Democratic leadership. Link.
  • Hyperlinked therein, related research on the use of policy feedback loops as a political weapon. Link.


  • On a distinction between interest- and idea-based outcomes. Link. ht Sidhya
  • AI, jobs, and the role of demand in employment amidst mass technological change. Link.
  • Related to last week’s spotlight on social wealth funds, a new post by People’s Policy Project with a radical proposal for collectively owned real estate. Link.
  • Using machine learning to resettle refugees: “the algorithmic assignment yielded higher employment rates in almost every location, including locations that had higher and lower baseline employment rates.” Link.
  • From the World Economic Forum: Eight Futures of Work (brief), Towards a Reskilling Revolution (more substantial).
  • “Record corporate profits and share prices, coupled with low corporate investment and wage growth.” On financialization in the non-financial corporate sector and declining worker prosperity. Link.
  • We previously sent this paper on monopsony from Azar/Marinescu/Steinbaum. Pro Market’s recent post situates the paper with other research. Link.
  • From FiveThirtyEight, an interactive set of gerrymandered maps. “We went back to the drawing board and drew a set of alternative congressional maps for the entire country. Each map has a different goal: One is designed to encourage competitive elections, for example, and another to maximize the number of majority-minority districts.” Link.
  • Power sharing and political violence. Link.
  • Davis is a Ph.D. candidate at MIT, and his group’s image processing algorithm can turn everyday objects into visual microphones—deciphering the tiny vibrations they undergo as captured on video.” Link. ht Will
  • Accidental geoengineering: “Studies have found that ships have a net cooling effect on the planet, despite belching out nearly a billion tons of carbon dioxide each year. That’s almost entirely because they also emit sulfur, which can scatter sunlight in the atmosphere and form or thicken clouds that reflect it away.” Link.
  • The New York Times Magazine’s cover story on the blockchain catches up, four years behind, on the social potential of blockchain technology: “For all their brilliance, the inventors of the open protocols that shaped the internet failed to include some key elements that would later prove critical to the future of online culture.” Link. From Steve Randy Waldman in 2014: “When you hear “cryptocurrency”, don’t think of Bitcoin or money at all. Think of Paul Krugman’s babysitting co-op. Cryptocurrency applications deal with the problem of organizing people and their resources into a collaborative enterprise.” Steve Randy Waldman continues the idea in 2016.

Each week we highlight research from a graduate student, postdoc, or early-career professor. Send us recommendations: editorial@jainfamilyinstitute.org

Subscribe to Phenomenal World Sources, a weekly digest of recommended readings across the social sciences. See the full Sources archive.