We Need Caution When Predicting The Future Of Work

As highlighted in a recent article, the release of ChatGPT in its various guises, along with numerous other generative AI-based technologies, has heralded a flurry of articles, studies, and headlines lauding the often catastrophic impact such technologies will have on jobs and society more broadly.

It’s the kind of simplistic and often doom-laden narrative that so often thrives on social media. As Greg Berman and Aubrey Fox remind us in their recent book Gradual, however, change seldom happens rapidly and almost never happens in such a linear fashion.

Messy change

They highlight how, in politics and policymaking, we are often tempted into thinking that change occurs after we elect the right leaders, who then craft the right policies, which then change practice on the ground. It’s an enticing vision of change, but it’s seldom one that is reflected in reality.

Predictions of technological change are often very similar, which perhaps explains why the Information Technology and Innovation Foundation (ITIF) famously found that just 10% of technology predictions made by experts between 1990 and 2007 were accurate.

The ITIF study found that many predictions were overly optimistic and that experts tended to overestimate the impact of new technologies and underestimate the challenges associated with implementing them.

Living up to the hype

Indeed, even research from Gartner (of Hype Cycle fame) found that about 80% of emerging technologies fail to meet their hype cycle expectations.

Gartner’s “hype cycle” is a graphical representation of the life cycle of a new technology, from its initial introduction to its eventual adoption or rejection. The study found that many technologies fail to live up to their initial expectations due to a variety of factors, including technological limitations, market conditions, and user adoption.

This finding was replicated in another study, this time by Accenture, which found that companies often overestimate the impact of new technologies on their business.

The study surveyed executives from 200 large companies and found that while most recognized the importance of new technologies, many were unrealistic about their ability to transform their businesses. The survey revealed that companies that took a more measured and realistic approach to technology adoption tended to be more successful.

Overall, these studies suggest that technological predictions are often overly optimistic and that many new technologies fail to meet their initial expectations.

Slow progress

So while many technologies are portrayed as being rapidly adopted, the reality is usually very different. The challenges are perhaps best summed up by Daniel Patrick Moynihan, who famously remarked that when considering change, “we constantly underestimate difficulties, overpromise results, and avoid any evidence of incompatibility and conflict, thus repeatedly creating the conditions of failure out of our desperate need for success.”

In my recent article, I explained that melodramatic visions of the future often prevent people from making the changes they might need in order to adapt to new technology, and bold visions also often fail to take sufficient account of how stakeholders might benefit.

These bold visions are often driven both by the marketers’ need to over-hype new technologies and the impact they can have, alongside the social media that rewards dramatic content that generates clicks, shares, and comments. Berman and Fox explain, however, that honesty is essential to actually deliver meaningful change

Bringing people along

They also highlight the importance of humility and respect to ensure that stakeholders are brought along too. This is highlighted in research from MIT and BCG, which shows that implementations of new technologies tend to be more successful when the organizations ensured that their employees also benefit from the technology.

This boost is greatest when employees feel like their job performance, relationships at work, and autonomy are improved because of AI. Managers, meanwhile, can improve the use of AI by focusing on the understanding of how the technology is being used and by developing trust in it.

Indeed, when respondents felt that they personally gained from using AI, they were over 3 times more likely to be happy in their jobs than those who struggled to gain any value from the tech. It’s perhaps also worth noting that the vast majority of people thought that AI would work alongside them rather than replace them.

The report suggests that if organizations, and indeed individuals, want to get the best out of AI, they should first focus solely on customer value and experience, then look at the clear benefits both the organization and its employees will derive from the technology. It’s also important not to make employees serve the AI, especially if it mainly benefits the organization and not the employees themselves.

Step by step

The dramatic visions of our supposedly AI-driven future are also at odds with a startup culture that is built on incrementalism. While founders and marketers often trumpet their ideas in revolutionary terms, they are encouraged to constantly experiment and pivot based on incremental insights.

It’s a concept that Karl Popper famously referred to as “piecemeal social engineering,” where problems are solved via small-scale reforms and a fundamental sense of trial-and-error. This leads to gradual progress over time rather than rapid changes delivered at once.

It requires a sense of intellectual humility that isn’t always evident in some of society’s more well-known entrepreneurs (even though research shows humility to be crucial to entrepreneurial success). Without this, there is a real sense that societal trust will be eroded in the same way that trust in politicians has been eroded due to the continuing use of expansive rhetoric that is never delivered on.

Facebooktwitterredditpinterestlinkedinmail