The ‘Digital Drag’ between Gen AI promise and adoption in the workplace
Whilst Gen AI is being touted as a solution to low workplace productivity, for it to work we need to understand the behavioural challenges behind adoption
Many countries are considered to have a workplace productivity problem with some estimates suggest that workplace productivity growth has been relatively stagnant for about 40 years. Into the fray steps Generative AI (Gen AI), frequently considered a panacea for faltering productivity, or at the very least a helpful signal to investors that these companies at the forefront of technological adoption, deliver enhanced productivity. Indeed, startups that mention ‘AI’ attract up to 50 percent more investment than those that don’t.
But with all the promise that this suggests, how do we counter the finding that 70% of all organizational transformation initiatives fail to achieve the anticipated results. And specific to Gen AI, the finding that over two-thirds of business leaders in the US saying that AI integration either remains limited or is non-existent is clearly concerning for the productivity challenge.
Of course, it is clear that the successful adoption of any new tool within an organisation takes time: it is frequently suggested that effective deployment typically depends less on technological capability than on human flexibility and the ‘people factor’. This can mean concerns such as the fear of the unknown, with the uncertainty of change creating anxiety and loss of control where employees can feel a decline of autonomy over their working day.
While these are certainly key considerations for adoption, these sorts of concerns perhaps do not represent the full story: when an issue is characterised as being due to an apparent deficit on the part of the individuals involved, we often need to look harder.
With this in mind, we set out the way in which the modern organisation has a complex set of issues sitting behind the apparent gap between the promise of Gen AI and workplace adoption, what we are calling the ‘Digital Drag’. The research we cover certainly suggests that responsibility for effective adoption of Gen AI and translation into productivity gains sits much more broadly across the organisation and not solely at the feet of the individual worker.
Reasons adoption of Gen AI can be tricky
Complexity2
As we have set out previously, the hierarchical structures of the traditional workplace, with their rigid command-and-control dynamics, are often poorly-suited to the complexities of the modern world. While these can be highly effective in environments that are relatively stable due to their clear chain of command, consistency in processes, and efficiency in decision-making, they can also be liabilities in today’s more precarious business environment. These systems, built for predictability, can translate into rigidity and slow pace of change - precisely the opposite of what todays unpredictable landscape calls for. No surprise that occupational psychologists such as John Amaechi are increasingly advocating for a shift towards more adaptive and decentralized models, reflecting a significant shift in thinking about organisational leadership.
But while decentralised models offer a real advantage to organisations through agility and responsiveness, it means that it can be harder to work out what is actually happening in an organisation. This is illustrated by the huge amount of questions and discussion about workplace trends where commentators attempt to make sense of confusing shifts in the workplace such ‘quiet quitting’ popularised on TikTok, where a user described it as “not outright quitting your job, but quitting the idea of going above and beyond.” Another is quiet hiring, which is asking existing employees to take on new tasks as well as using contractors to fill in needs at companies that are struggling to find workers.
Note the use of the term ‘quiet’, perhaps suggesting how complex work structures means that much employee navigation of the organization is tacit rather than explicit. This means understanding unspoken rules, reading between the lines in communication, and knowing how to get things done within the unique context of the organization.
Add into this mix the way that GenAI arguably differs from traditional tools due to its open-ended nature, its ability to ‘learn’ and improve over time and the ongoing sense that new capabilities are likely to be available shortly. If we contrast this with other workplace tools, they (arguably) typically have more defined functions and capabilities as well as a longer lead time for changes or enhancements, then we can see that adoption of this complex tool into an equally complex work environment is no small feat.
No wonder then that in today’s more decentralised workplace it can take time for a less well defined tool to find its feet, being more dependant on the tacit, informal structures of the business to make change happen.
Also throw into this the way that people are acutely aware of the disruptive impact of AI not only on their nature of their work but also for their job security.
Will I keep my job?
One of the sometimes unspoken concerns that workers have in the adoption of new technology is that their livelihood can potentially be at risk. And the media coverage of AI certainly encourages this understanding with Sam Altman, the CEO of OpenAI saying:
“Jobs are definitely going to go away, full stop”
And he is not alone: the International Monetary Fund says that 40 percent of the workers in the world have jobs that “will be affected by artificial intelligence.” It is not simply the middle or lower spectrum workers that might be concerned - C-Suite execs tend to command high salaries, so there is financial incentive to consider how some of these functions could be undertaken by AI: indeed Polish drinks maker Dictador recently appointed a humanoid robot called Mika as its "experimental CEO."
Resistance in the workplace to technology adoption is nothing new; in the Nineteenth Century the Luddite movement famously involved textile workers challenging the manner in which new technology was leading to jobs and wages cuts. So perhaps no wonder that employeees may not be falling over themselves to look for ways in which their work can be done more cheaply and efficiently by a machine: the mantra may be that this frees up more time for more rewarding work, anda recent study found that 95% of workers see value in working with Gen AI, but approximately 60% are also concerned about job loss, stress and burnout. While almost all people appear to consider that Gen AI can offer value, has the argument really been won?
Value creation is by no means always certain
It is no secret within the workforce that there is a very live debate about the extent to which Gen AI will offer the degree of value versus the extent to which this is being hyped. One example (and by no means isolated) is a recent article in the New York Times suggests the latter:
“The question (today) isn’t really whether AI is too smart and will take over the world. It’s whether AI is too stupid and unreliable to be useful. … It feels like … AI is not even close to living up to its hype. In my eyes, it’s looking less like an all-powerful being and more like a bad intern whose work is so unreliable that it’s often easier to do the task yourself. That realization has real implications for the way we, our employers and our government should deal with Silicon Valley’s latest dazzling new, new thing.”
Added to this are the questions concerning the business model of Gen AI with Peter Cappelli, a management professor at the University of Pennsylvania Wharton School suggesting that overall, generative AI and LLMs may create more work for people than alleviate tasks. They can be complicated to implement, and “it turns out there are many things generative AI could do that we don’t really need doing.” In other words, if the infrastructural costs of offering Gen AI outweigh the business value that is accrued by organisations (and therefore the amount they are willing to pay for it) then there is a question mark about its long term viability.
On this basis workers may be hedging their bets – is this something that they should be investing their time and energy in when there are more certain near-term ways in which they can enhance their work and careers?
What steps can be taken?
Recognise this is a marathon not a sprint
In Aesop's fable, a speedy hare mocks a slow-moving tortoise, who then challenges the hare to a race. Overconfident in their speed, the hare takes a nap during the race. Meanwhile, the tortoise continues at a steady pace and ultimately wins, demonstrating that perseverance and consistency can triumph over arrogance and haste.
With this in mind, while slow adoption or resistance may appear simply to be a self-centred concern for employment conditions, or anxiety about one’s ability to integrate into their work. However, people such as Thomas P. Hughes have pointed out that slow speed of adoption can also be seen as a crucial counterbalance to technological momentum, ensuring that technological advancements are thoughtfully integrated with consideration for effective use in the workplace. By slowing down the adoption process, it therefore perhaps allows for a more effective approach, ultimately leading to technologies that have more value and longevity.
In line with this, philosopher of technology Andrew Feenberg advocates for a participatory approach, emphasising the importance of carefully including a wide range of voices in technological decision-making. This allows broader interests of the workplace to be met if it is developed and implemented in ways that enhance human capabilities, given that technology is a product of wider workforce choices and values.
And we can look at this in behavioural terms: Construal Level Theory suggests that at an early stage the understanding of Gen AI in its application to workplace challenges maybe generalized and simplified. People might underestimate the complexities involved because they are not yet dealing with the specific details and practicalities.
As the project progresses and people move closer to implementation, they increasingly need to address specific details such as integration with other workstreams, training for their teams, data security and privacy issues, security concerns, and so on. These concrete aspects often reveal unforeseen difficulties and obstacles. In other words, people start to see the ‘how’ – the specific steps, resources required, and potential roadblocks. This shift from abstract to concrete thinking can lead to ‘Digital Dragging,’ where the initial momentum slows down due to the recognition of these challenges.
Some care needs therefore to be taken when interpreting speed of adoption – not least as the complexity of workplaces and of the technology itself means that a more considered route may well be needed.
Encourage and empower the early adopters
So just how can organisations best approach the adoption of new technology and, in this case, of Gen AI? One useful route is via Lead User theory, developed by Eric von Hippel, which focuses on the role of advanced users in the development and diffusion of new technologies. These are people in the workplace that often face needs ahead of their peers and are positioned to benefit significantly from innovations that address those needs. This means they can be relied on to be champions of new technologies, and help ensure that the solutions developed are highly relevant and practically applicable.
Left to their own devices, Lead Users will often seek to develop their own solutions and can operate in a skunk works way, circumventing the usual governance processes of an organisation if they are found to be unhelpful. This means it is critical to accurately identify who the lead users are, recognising they are often not in formal leadership positions but possess deep practical knowledge and experience. Organisations need to create incentives for these lead users to actively participate in innovation processes, perhaps with recognition, additional resources, or career advancement opportunities.
But also, perhaps equally as importantly, while Lead Users are good at developing highly customized solutions, there is often a challenge of how to scale these innovations to the broader organization so they can be generalized and implemented across different departments.
Consider the wider organisational culture
A new book, Cultures of Growth, by social psychologist Mary Murphy offers some guidance for ways to encourage Gen AI adoption in organisations (there is a great review of here book that we are drawing on here). The backdrop to the book is the work of her advisor, Carol Dweck whose book Mindset, sets out two ways people can conceive of themselves and their talents. Those with a fixed mindset consider that positive traits are inherent and immovable, in contrast to those with a growth mindset who tend to feel that good qualities can be developed over time through hard work, learning, and collaborating with others. The evidence seems to suggest that organisations that promote growth mindsets over fixed ones tend to do better in the form of happier, more empowered members who are more likely to come up with new ideas and collaborate well on projects.
In Cultures of Growth, Murphy suggests that any individual's self-image is shaped by their everyday environment so even if we try to maintain a growth mindset for ourselves, it will not be maintained if workplace culture is more oriented towards fixed mindset behaviours. With this, Murphy outlines two possible environments: in ‘Cultures of Genius,’ fixed mindsets dominate, and the focus is mainly on the contributions of star performers, who are considered inherently more skilled than the wider workforce. By contrast, in ‘Cultures of Growth,’ positive mindsets are supported more widely, with the assumption that anyone in the organisation, given the right resources and structures in place, has potential to contribute to success.
In an analysis of Fortune 500 businesses, Murphy found ‘Culture of Genius’ businesses are less adaptive and resilient, less inclined to take risks (as the price of failure is so high), and more likely to experience significant staff turnover. And worse, these organisations are also consistently associated with greater lapses in integrity and ethics. By contrast, ‘Culture of Growth’ businesses, "which embrace learn-it-alls over know-it-alls," have more satisfied employees who both collaborate and innovate better due to the growth mindset cultivated around them.
‘Growth’ businesses also tend to be more diverse than their ‘Genius’ counterparts, for example they tend to have more women on their board. This is in part due to ‘Cultures of Growth’ believe that good ideas come from everywhere and they value differences (cultural, economic, social)." ‘Cultures of Genius’ on the other hand, tend to have fixed ideas about who the best performers are – and focus on fixed (and often sexist and racist) notions of inherent worth, On this basis, ‘Genius’ companies fail to properly access the huge amount of talent, viewpoints, and ideas that a more diverse workforce can bring to the table.
From this we can clearly see the way that it is not enough to place the success or failure of adoption of Gen AI at the feet of individuals in the workforce – the organisational culture also needs to be addressed adoption is to be scaled from Lead Users to the wider workforce.
In conclusion
Understanding how to encourage the adoption of Gen AI in the workforce leads to a much broader discussion about the modern workplace. As set out in the beginning, the pressures to enhance productivity through rapid adoption of new technology such as AI can lead to a narrow conceptualisation of the challenge, leading too much of the responsibility being placed at the feet of the employee.
But the workplace is one which is more complex, the tools more nuanced and multifaceted. How Gen AI is to be applied is not necessarily simple and requires careful consideration. And in the midst of this is the question of where to spend one’s time in challenging market conditions and uncertainty of the value Gen AI will deliver.
There is a long history of ‘Digital Dragging’ actually benefitting companies such as the introduction of Personal Computers during the 1980s when many employees continued to use typewriters and filing cabinets, on the basis that they found these were more reliable and easier to use. Whilst this minimal engagement with new systems slowed down the overall adoption process it was found, in hindsight, to reinforce the importance of better training and user-friendly interfaces to support the integration of personal computers in the workplace.
But more broadly, there is a growing body of evidence that we cannot rely on individuals alone, regardless of how brilliant they are, to be facilitating adoption. Instead, we need to be looking more carefully at the culture of the business and the extent to which this is working effectively, as it seems that to generate the much- sought-after increases in productivity, then we need a diverse set of employees who can both collaborate and innovate with a growth mindset cultivated around them.