Navigating AI optimism and workforce realities

It’s hard to overestimate the level of hype surrounding the future of work at the moment. Nowhere is this more encapsulated by the furor surrounding the release of DeepSeek towards the end of January.

Just as the hype around the potential of generative AI had driven American tech companies to huge valuations, the emergence of the Chinese competitor wiped around $1 trillion from American stocks in what Marc Andreeson referred to as AI’s “Sputnik moment.”

Given such wild swings in expectations, it’s difficult to understand what the future might realistically have in store, as the huge valuations of American tech companies seem based on ephemeral and optimistic visions of the future rather than the realities of the present.

A flurry of reports have attempted to make sense of things. For instance, The Rise of the Superworker, by Josh Bersin, takes an optimistic perspective. It portrays a future where employees are superpowered by AI to become ever more productive.

Alternatively, Navigating Tomorrow, from the Global Labor Market Conference, looks at the global forces affecting the world of work, including demographic changes and climate change as well as technology. You also have the Pissarides Review, from Warwick Business School, which reminds us that any changes we make should have humans at the core, and ensure that things like inequality and the various systemic barriers that affect society are addressed (as indeed did the Digital Dialogues report from the Digital Futures At Work research center.

A transformative force

All of these reports have at their heart the fact that AI will be transformative, but do they underplay some of the challenges involved in adopting new technologies? After all, past industrial revolutions have generally unfolded over decades rather than months or years.

Indeed, recent research found that the initial adoption of generative AI focused more on the uninitiated, who tend to be bewitched by the technology's seemingly magical capabilities, whereas those with more knowledge of AI are better able to see it as it is.

Bersin's paper exacerbates this oversimplification by failing to address any of the other challenges involved in utilizing AI, including overreliance on the technology, biases that remain an issue, and, more broadly, any displacement that may occur.

The other two papers tread somewhat familiar ground in exploring the skills situation. I've written previously about how certain skills, especially in AI, help with the effective adoption of the technology, but there are also obvious challenges when it comes to those displaced. Both reports put employers front and center of any reskilling work, which is already problematic given both relatively low investment in training by employers and also the obvious fact that displaced people, who are most in need of support, won't have an employer to help them.

This does threaten to deepen inequalities that are already at rarely seen levels. The authors suggest that middle-skilled jobs are particularly vulnerable and we could see a hollowing out of society, with jobs in areas like management as plentiful as those in nursing and construction, whereas those in areas like coding and graphic design are gutted.

Much of the political polarization in recent years has been driven by a sense of powerlessness, that great changes are happening to people rather than with people. It’s not immediately obvious that the AI transformation will be noticeably different.

Involved in the change

Fitting his optimistic stance in general, Bersin believes that AI will liberate humans and allow us to focus on more strategic tasks, which makes such concerns somewhat redundant as AI will fulfill a liberating role. It's not a scenario that seems to be borne out by current evidence, and indeed, one of the case studies he cites of such a trend actually involved the redundancies of huge numbers of people, few of whom would have been consulted about the introduction of AI and automation into the workplace.

“Much of the political polarization in recent years has been driven by a sense of powerlessness, that great changes are happening to people rather than with people. It’s not immediately obvious that the AI transformation will be noticeably different.”

One can certainly see such a scenario playing out in many organizations, especially where efficiency and profitability trump everything else, but the Pissarides Review projects a more humane future in which humans are at the core of discussions around how technology will be deployed.

What all three reports have in common is the sense that work will need to be redesigned, both so the potential of the technology can be realized and also so that humans aren't swept up in the tide of change. The best organizations are co-creating this change with workers rather than managers forcing change upon people.

Rates of adoption

Both the Digital Dialogues and Pissarides Review also highlight the potential for regional differences in how technology is adopted, which could also exacerbate pre-existing inequalities. There have been longstanding challenges in encouraging smaller businesses to adopt the latest technologies, and there are likely to be regional divides too, with organizations in larger cities quicker to adopt (and adapt to) AI than smaller towns and rural regions.

This is a divide that has been all too evident in recent years and has underpinned the political polarization that saw Donald Trump rise to power (twice) and Britain vote to leave the European Union. Research has shown a link between the risk of automation and support for populist politicians, and this trend seems likely to deepen given the ties between Elon Musk and Donald Trump (not to mention populist parties around the world).

While policymakers have long been aware of this, they have been far less successful in responding to this issue. Indeed, the UK government's recent AI paper makes a number of farfetched claims about how AI can actually reduce regional inequality, such as by installing the vast number of data centers required by OpenAI et al (and potentially rendered redundant by DeepSeek's more efficient approach while the ink is barely dry on the report).

Leading the change

Most technological changes require process change for the new capabilities of the technology to be capitalized on. This is underlined by research showing that managers are actually growing in number in firms that are among the early adopters of generative AI.

What is perhaps key, however, is that those managers oversee the change in ways that serve the humans in these organizations rather than the organizations transitioning to better serve the technology. Thinkers like Gary Hamel have long called for work to be made more humane, and this has never been more important than it is today, both from an individual organizational perspective and from a societal perspective.

This is especially so as there seems to be precious little sign that the technology leaders themselves are approaching things from this perspective. With political leaders typically lining up behind them, there has to be a real concern that this transformation is being driven by those with their own interests front and center.

Sadly, as I’ve written before, deployments of AI to date have often resulted in workers serving the algorithms rather than the other way around.

So, while the hyperbole around AI is quite probably overblown, the shockwaves caused by DeepSeek may provide us with a very welcome chance to pause and take stock of just what kind of future we want.

It’s clear that AI will be a key part of that, but we need to take a more collaborative approach to what changes AI will introduce, with workers, technologists, leaders, and policymakers working together to create a future whereby AI empowers us all, not just a select few.