The canaries in the AI employment coalmine

Artificial intelligence is already in the spotlight for its potential environmental impact, primarily due to its high energy consumption. But AI technologies will also likely have major social impacts, and affect corporate governance at multiple levels. So AI should be a big concern for ESG investors and those interested in sustainability overall.

Here I’d like to focus on the social impact of AI, specifically on the labour market.

AI is such a big issue that no one can predict its eventual impact on jobs with any certainty. But we are getting some hard evidence of its initial effects, including from a fascinating study: “Canaries in the Coal Mine? Six facts about the Recent Employment Effects of Artificial Intelligence[i]." This study – by Stanford University academics Erik Brynjolfsson, Bharat Chandar and Ruyu Chen – is based on large-scale data from ADP, a payroll software provider, on the US economy.

Canaries were traditionally used in coal mines to warn of poisonous gas (the warnings were provided not by a trill, but by the unfortunate canaries’ deaths). The study suggests that the “canaries” in the US labour market – i.e. the quiet collapses – are to be found in the reduction of jobs for the young and entry-level employment.

In the study’s own words, (US) “employment for young workers has declined in AI-exposed occupations.” The chart below, taken from the report, provides a simple graphical representation of this. The blacker the line, the more exposed the employment is to AI.

Figure: US employment categorised by sector AI exposure, rebased so October 2022 = 100.

Source: Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, Canaries in the Coal Mine? Six facts about the Recent Employment Effects of Artificial Intelligence

The report reinforces this point by noting that “though overall (US) employment continues to grow, employment growth for young workers in particular has been stagnant.” Importantly, the report says that “labour market adjustments are visible in employment more than compensation.”

The impact of AI on employment has already been substantial: the report reckons that employment levels for US workers aged 22-25 in AI-exposed industries have declined by 13% since late 2022.

What we don’t know is if this is temporary, or whether firms will eventually reconfigure these entry-level jobs to make them more appropriate and viable in an AI-dominated world. But the report should still encourage us to have a stronger debate on AI and employment, supported by data, rather than simply falling back on broad generalisations – for example, the assumption that manual jobs will be relatively unaffected by AI, a guess which may well be wrong.

The pace of future AI transformation is also uncertain: “AI as Normal Technology[ii]" by Arvand Narayanan and Sayash Kapoor of Columbia University uses lessons from past technological revolutions to explain why they think transformative societal and economic impacts will be slower than some expect. They say that the diffusion of AI in safety-critical areas has been slow, that personal adoption may not be as rapid as it seems, and that productivity gains (as with past technologies) may take some time to show. For the authors, one major issue is a future potential division of advanced labour between AI and humans (with humans focused on AI control).

The “AI as Normal Technology” essay also contains a useful historical analogy, which I think should inform future conversations on AI and employment. As the authors put it, at the start of the first industrial revolution (i.e. the 18th century), it would have been “useful to try to think about what an industrial world would look like and how to prepare for it, but it would have been futile to try to predict electricity or computers.”

In other words, while we cannot know what exactly lies ahead, we still need to start thinking hard about possible scenarios and how to manage them. Employment issues have to be part of this AI and sustainability discussion.

This article is also published on the author's blog. illuminem Voices is a democratic space presenting the thoughts and opinions of leading Sustainability & Energy writers, their opinions do not necessarily represent those of illuminem.

Track the real‑world impact behind the sustainability headlines. illuminem's Data Hub™ offers transparent performance data and climate targets of companies driving the transition.