The U.S. Senate AI Insight Forum recently held a session on AI and the Workforce, where
Information Technology and Innovation Foundation
president Rob Atkinson presented counterpoints to concerns over AI creating drastic structural changes in the U.S. economy and labor markets. That statement and his "Top 10 Things I Disagreed With" article provide a good summarization of key issues raised in the discussions. (See Forum #3 here).
I agree with most all of his written statement and encourage follow through on recommendations.
However, in reviewing the list of disagreements I found room for some further considerations, which I will share here not at all as critique, but in expanding the dialogue.
As with all postings, these positions are solely my own and do not reflect those of my employer or any group with which I am affiliated. Their intent is to stimulate critical thinking in these important discussions.
AI & Workforce Findings / Debated
Please visit the original article for the full text paragraphs of these points.
1. Automation lowers wages.
- Atkinson: Neither logic nor good academic literature support this assertion. It’s based on two myths.
- There is a conversation in Kurt Vonnegut Jr’s “Player Piano” (1952) that frames part of the concern over lowered wages. The structure behind it can be debated, but there is also the psychological and sociological component that drives the political and economic response.
'What have you got against machines?‘
'They're slaves.‘
'Well, what the heck, I mean, they aren't people. They don't suffer. They don't mind working.‘
'No. But they compete with people.‘
'That's a pretty good thing, isn't it –considering what a sloppy job most people do of anything?‘
'Anybody that competes with a slave becomes a slave.’
- We can see some of this sentiment playing out in the politics behind middle America working-class disdain for the tech industry of coastal elites – a sense of being minimized by change that has left their way of life behind.See also: ?Player Piano, the One-Dimensional Society, and the Emergency Brake of History
2. Technology has been the cause of inequality.
- Atkinson: This has been a common refrain over the last two decades, but it isn’t true. The fact that the top 25 U.S. hedge fund managers earned an obscene $26 billion in 2021 had nothing to do with tech-driven automation and everything to do with the institutional structure of the U.S. economy.
- Agreed, though I think there is a very valid question raised in what may be the result of that institutional structure (which is not being otherwise curtailed) further enabled to consolidate gains through capital investment in automation technology rather than sharing profit with the workforce. The profit-focused incentives in Friedman’s Shareholder Theory that drive contemporary financialization of industry explicitly discount accountability of the firm to societal obligations – and as often implemented in bias toward short-term returns, even reduce long-term competitive strength.
- Driven by shareholder interests, executive leadership can employ AI and sophisticated automation toward skill commoditization, labor surplus, and shift costs from wages to capital investment.? The long-term impacts would continue to reinforce inequality by reducing wages and shifting long-term gains (and their risks, despite risk of longer-term sustainability) toward immediate payment to investors and executives.
- See also: Questions on the Future of Work (2): Institutions & Economics
3. The pace of AI innovation is like a tsunami.
- Atkinson: To be sure, AI and large language models are ushering in a new era of computing that could define the next decade or so. But that’s a pattern we’ve seen many times before, roughly every decade.
- I agree, though think this requires more time to observe.
4. We run the risk of Great Depression-like unemployment rates.
- Atkinson: nearly every study on technology-based productivity growth has found that it produces no increase in unemployment, for the simple reason that productivity lowers costs, which boosts wages and lowers prices, which in turn leads to more consumption, and hence more jobs.
- This is a question of timing. Yes, the prior studies counter this position, but the employment disruptions due to the steam engine, electric motors, telecommunications, etc. did not have the generalizable impact that AI/automation may have upon the breadth of cognitive and skilled labor.
- One structural change we must confront is the expectation regarding the legacy pattern of education, followed by long-arc (“career”) employment, followed by retirement. At minimum, we must consider continual education throughout shifting job responsibilities and expected skillsets. This impacts not only the readiness of educational institutions (with community colleges dead-center for later-career students) but also how to support workers during these periods of transition.
- See also: Continuous Learning and the Corporation by
Chris Shipley
and Jobs Are Over by
Heather E. McGowan
5. There has been rapid adoption of automation technologies.
- Atkinson: Actually, productivity growth rates over the last 15 years have been abysmal. This is in part because capital expenditures on automation have been relatively anemic.
- Also a factor that requires more time to let play out and then assess.
6. The tax code favors capital, not workers.
- Atkinson: The notion that the tax code is stacked in favor of excessive automation is promulgated to get policies to slow automation.
- Perhaps the more valid issue to raise for tax and payroll policy relates to corporate burdens of employer-based healthcare in the US. Capital maintenance is not even comparable to healthcare costs, as they can be amortized across assets and costs may be more reliably predicted than the organic counterpart (whose costs may scale linearly+).
- Employers should be focused on their business; dealing with health benefits is arguably a distraction and hit on productivity. Similarly for the workforce, their employer’s (potentially arbitrary) selection of plans can stifle personal productivity – even more so by adding complexity in job transitions, where discontinuities require workers to adjust to different provider rules and networks, re-justifying treatments, etc.
- See also: Healthcare After Quarantine
- Add to this calculation the exposure to labor “availability reduction” for sick time, personal time, vacations, weekends, and even sleep, eating, and bathroom breaks.
7. Employers have strong incentives to use AI to harm workers.
- Atkinson: The OECD recently found that most workers reported an improvement in job quality from the introduction of AI into the workplace.
- This OECD report is excellent! And I'm in absolute agreement relative to the opportunity to reduce or even eliminate Dirty, Dull, Dangerous, and Dear aspects of jobs that are inherently unpleasant and undesirable.
- But let’s turn this around and consider this potential for AI to improve job (and life) quality of workers and the corresponding employer incentives or disincentives. In Shareholder Theory (as mentioned in response to #2), maximization of profit / returns to shareholders guides company strategy.
- Let’s assume we employ 10 bottom-tier workers to perform tasks A (40% of time) and B (60%). Automation technology becomes available to reduce labor on task B to 10% of the worker’s time (so A + B = 50% in total compared to the original). Based on incentives to maximize profit, one can argue the simplest default action for management would be preserving the output while reducing costs to that of only 5 workers.
- Yes, there are opportunities to retrain, expand job scope, etc. – but these will not be the default. And the long-sought after “work-life balance” that could be attained in the alternative scenario for retaining 10 workers for a 20-hour work week while holding pay constant is unthinkable. (Despite the idea being introduced by John Maynard Keynes in 1930).
- So are there any viable incentives to rethink the 40+ hour work week and days off? And more generally, how may incentives create opportunities to advance national and societal interests that may have been neglected in legacy employment structures? (“Unprofitable” activities such as care for children, elderly, sick, and disabled, education and social connectedness, pursuit of general scientific discovery - including medical care, improving the environment, and creative passions like art, music, and storytelling.)
- See also: Questions on the Future of Work (5): Values and Questions
8. Luddites were not antitechnology.
- Atkinson: As long as you don’t define destroying automatic looms with sledgehammers as antitechnology, we are in complete agreement.
- Same.
9. Artificial general intelligence could be here in five years.
- Atkinson: Let me just quote MIT professor and Rethink Robotics CEO Rodney Brooks, who observed with respect to AI: “Misled by suitcase words, people are making category errors in fungibility of capabilities—category errors comparable to seeing the rise of more efficient internal combustion engines and jumping to the conclusion that warp drives are just around the corner.”
- Agreed, and Rodney should be a primary source for a lot of sense-making on these topics. I'll add that while I believe the technology is indeed accelerating, I see Kurzweil’s singularity to be more of an “asymptote” – the convergence will slow due to multiple counter-forces, including technological complexity, regulation, social/psychological acceptance, energy costs, etc.
10. AI should be used to complement workers, not eliminate jobs.
- Atkinson: if this had been the historical American ethos, then we’d still have elevator operators, telephone operators, newspaper typesetters, and lots more gas station attendants, travel agents, and bank tellers.
- I think the examples selected here are perhaps too reductive. I believe the intent of this posture is well-framed in the title analogy and article “Automation Should be Like Iron Man, Not Ultron” by Tom Limoncelli.
- That is, place the human at the center (Iron Man), preserving human strengths in versatility, improvisation, and introspection while leveraging machine strengths such as precision, tireless focus, and consistency.? For example, there is a strong temptation to employ AI in place of medical scribes – but there are secondary benefits to pre-med students employed in that capacity, both in the reflection of the doctor’s awareness of the informed audience and in the hands-on learning for the scribe/student.
- The alternative (Ultron) approach instead aggressively automating tasks opportunistically with a prioritized focus on not only quality but also productivity (to reduce costs, risks) through capital investments in technology – starting with simple tasks, and ever-expanding to all tasks. Limoncelli cautions that the “leftover” tasks for human labor in this approach result in jobs increasingly difficult to fill and retain as the work ever-increases in difficulty, intensity, and skill.
- Certainly underemployment and pointless “Bullshit Jobs” are not the desired goal, and a symptom of overly resisting adoption of these technologies. But there are timely issues to consider in terms of the purpose of firms in the context of the influence of financialization.
Importance of Work & Purpose of Institutions
In?The Wealth of Humans Ryan Avent asks,
“Can we build [new] institutions for those who do not work [only] because their work is [no longer] necessary to generate economic growth?”
As automation technology advances to unleash unprecedented productivity, we enter the realm of possibility for a world as depicted in Player Piano – where labor surplus may undermine essential social contributions our culture demands from individuals: earn your keep (dignity), satisfaction from doing (pride), and existential identity (purpose).
AI should also be used to complement the purpose companies serve, not eliminate their part in the human economy – reduced to solely financial performance metrics. Financial institutions once primarily served businesses in facilitating growth through capital investments in human activities (“the real economy”) as opposed to the present-day layers intermediating equity, debt, and credit. (Aspects of which, once upon a time, were considered a heresy.)
Pivoting from serving businesses to a position as their master, financial firms may only narrowly regard the importance of the products and services provided by businesses – while seeming notoriously indifferent to resulting poverty, inequality, or personal needs for dignity, pride, and purpose across its workforce. The financialization of Shareholder Theory effectively misdirects corporate dignity and pride, while eroding the purpose of the company as a side-effect. This has, in turn, eroded American confidence in capitalism itself.
In 1937, Ronald Coase’s The Nature of the Firm offered these attributes for the purpose of the firm:
- continuity (of access to needed talent, tools, materials),
- communication/coordination between people enacting processes,
- development of intellectual resources (practices, culture, training, knowledge),
- creation of products and knowledge whose complexity exceeds informal efforts of individuals, and
- sustaining long-term bets on innovation beyond demand-response.
All compatible and coherent with the dignity, pride, and purpose of both the firm itself and its employees - and sustaining long-term competitive strength to deliver products and services in "the real economy."
AI & Workforce Findings / Agreements
Concluding - I wholly agree with the two points on which Atkinson cites strong agreement:
- We need better policies and programs to help workers manage technological dislocation.?
- Employer engagement with workers when AI is deployed can improve AI effectiveness.?
