A new study from a team of Harvard Business School researchers provides some good news for below average workers. This group will see the most upside from using generative AI.
The study involved 758 workers at Boston Consulting Group who completed a range of 18 realistic consulting tasks that were deemed to be within the capabilities of current AI systems. The study showed that while all participants saw improved ability to complete more tasks (12% increase) and to do so faster (25% increase), the biggest gains were for below average workers. Those workers saw a 43% improvement while their higher performing counterparts only experienced a 17% gain. Essentially, using generative AI helped to level up average or below average performers to be close to on par with high performers.
“Consultants across the skills distribution benefited significantly from having AI augmentation, with those below the average performance threshold increasing by 43% and those above increasing by 17% compared to their own scores.” (Dell’Acqua et al, 2023)
Things that make you go hmmm…
There’s been a lot of discussion about augmented intelligence which speaks to the idea that using AI for tasks that it's capable of performing will make us all better. The assumption has been that this benefit will be equally distributed and that everyone will see the gains of using AI. If you are already a very good performer, the assumption is that with AI you will become a stellar performer, while an average performer might become a good performer. In other words, you will still retain a competitive advantage through your own talent.
This study suggests that if you are already good at your job - that is to say you are above average or even outstanding - you will see the least benefits from using AI.
That’s not to say you won’t see any benefits at all. All of the participants were able to do more and faster. But that won’t give you a competitive advantage because everybody improved on those fronts. What are the implications for this finding?
I saw the sign (and it opened up my eyes)
At an individual level, you might wonder if it's worth the extra work and sacrifice to become a top performer when AI is going to level up your less talented colleagues to essentially the same place as you. Since your gains through AI are nowhere near as dramatic and since the overall output level achieved by both you and the underperformer using AI are very similar, you might decide it's not worth the sacrifice.
Why work on becoming better when the payoff isn’t there? Stepped out to a societal level, if a lot of people make this choice, we may see the rise of mediocrity as an unintended outcome of self-interested rationality.
From a business perspective, this finding might illustrate whose jobs are most at risk, namely highly paid talented workers. If a business can get similar outputs from AI plus average skilled workers, then why pay for top talent? Why not invest in those with adequate or even below average skills and then boost their output with an AI system? The business can save the costs of paying for top performers and get essentially the same outputs. You might consider this the zone of adequacy and seek to retain talent that is in the zone. The zone is where the mix of performance outputs and costs to retain that performance (e.g. employee pay) are fully optimized for your business. Over time, you may also find the higher performers reduce their compensation expectations in order to stay in the zone.
If we consider what this means from the perspective of educators, we might think about the value of education in the context of training workers for future jobs. In this world of augmented intelligence, less could be more. Doing the minimum might be completely fine, even a benefit.
It’s a tough pitch to tell students to work hard to become top performers when being a C or C+ student may actually serve you just as well in the AI-enhanced job market.
A similar study of performance impacts involving law students suggested that AI might even be detrimental for top-performers, noting that "the effect of AI assistance varied significantly depending on baseline student performance; low-performing students received a substantial boost, while top-performing students may have been harmed by access to AI." (Choi and Schwarcz, 2023, p 31)
Mixed with particular political ideologies, the finding that human skill development is not necessarily the key indicator of success in an AI-enhanced workplace might be further justification to defund or deprioritize education. It might also cause individuals to rethink the idea of pursuing traditional post-secondary degrees. Both of those decisions have enormous impacts that extend well beyond workplace success.
This research serves as an early warning sign to think about the bigger consequences of AI augmented work. Of course, this is only one study and I don’t believe it’s even been peer reviewed yet. While other studies suggest some consistency in this finding, such as this one by Brynjolfsson et al, I’d like to see further research on this topic. There are also a range of other interesting findings including a discussion about how AI should be used. Should we be more like Centaurs, where we automate specific things, or more like Cyborgs where we fully integrate AI into the overall process?
We should also pay attention to how research studies are framed. I have, admittedly, focused on only one of the many findings in this research study in order to highlight some ethical concerns that I hope readers will find rather obvious. Others, such as MIT's Sloan School of Management have chosen to position this same research by highlighting an "up to 40% increase in performance for highly skilled workers" which just goes to show how easy it is to position the details in service of a narrative.
By Katrina Ingram, CEO, Ethically Aligned AI
Sign up for our newsletter to have new blog posts and other updates delivered to you each month! Ethically Aligned AI is a social enterprise aimed at helping organizations make better choices about designing and deploying technology. Find out more at ethicallyalignedai.com
© 2023 Ethically Aligned AI Inc. All right reserved.