While it is likely most roles will sooner or later be touched by (Gen.) AI, there is growing scientific evidence that not all workers will be equally impacted by the technology. This new body of research shows that Generative Artificial Intelligence can substantially raise average worker productivity by “raising the bottom”. Two empirical studies published in the past six months show just that :
“Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence” by S. Noy and W. Zhang: 444 college-educated professionals were assigned writing tasks, with half of them using ChatGPT. The results show that ChatGPT significantly improves productivity, reducing time taken and increasing output quality. It also reduces inequality between workers, benefiting low-ability workers more, and shifting tasks towards idea-generation and editing. Participants reported increased job satisfaction, self-efficacy, and mixed emotions about automation technologies.
“Generative AI at Work” by E. Brynjolfsson, D. Li and L. Raymond: The study examined the implementation of a generative AI conversational assistant in customer support. The tool boosted productivity by 14% on average, benefiting less experienced workers the most, while having minimal impact on experienced workers. The AI model helped transfer knowledge from skilled workers to newer ones, enhanced customer satisfaction, reduced the need for managerial intervention, and was shown to improve employee retention.
In short, Gen. AI increases productivity (we knew that), with specific benefits for low-ability workers and less experienced employees (we did not know that). Additionally, both studies highlight positive effects on job satisfaction, and improved customer outcomes as a result of AI implementation.
This has stunning implications. Technology has long been a force for increasing inequality, but AI may be different. By giving people with solid foundational skills the tools to learn faster and be more productive, they might become as good as the average quicker than they otherwise would have, and be rewarded more for it. Meanwhile, talented individuals will not be pulled down towards the average; they may even become marginally better at their role. Meritocracy survives… it just becomes a little fairer.
This is not to recommend that paralegals become lawyers or nurses become doctor. This is to say that education is expensive, especially for the jobs mentioned. And that we have a dire need for more well-trained workers, whose human traits cannot (yet) be imitated. And that giving a young nurse the ability to become proficient quickly, freeing time for experienced colleagues to give better care may just about save a crumbling system.
With this is mind, we have a chance to create a better corporate world for all. But it will not happen by itself. We have to make it happen. Today, two paths have emerged ahead of us.
Path 1: without government intervention and internal calls for a Fairer Deal within companies, executives will use this technology to devalue the roles of the middle 60% of the organisation, replacing them with younger, cheaper alternatives.
We would move towards a world with a few super-expert… and everything else done via automation and minimum-wagers doing for cheap what was once expensive. A world where no-one’s labour is scarce because we’re all too equal. And all those savings will go towards the bottom line and, ultimately, shareholders. This is still the most likely future, and may lead to an uprising. But that’s a story for another day.
Path 2: as fertility falls, populations age and immigration is further criminalized, we see the worker-to-retiree ratio skyrocket. We realize that we have more work than people and begin relying on automation to allow one worker to do the work of two. We start paying something resembling a Universal Basic Income thanks to the wealth created by AI. People start working less and create great art.
Sure, it’s an unfair system: some workers will produce more, and will not necessarily be paid more. But society may yet thrive, and in the long run, we could see a drastic reduction in economic inequalities. Unfair… but differently unfair than what we’re living through today.
Both paths are not mutually exclusive, of course. What’s most likely is that we end up somewhere in the middle. But the more we fight for the rights of the few, the more we fight for our children’s future, the more we get closer to eliminating both billionaires and homelessness. Would that be so bad?
AI Could Reduce Economic Inequalities. We just have to be willing to make it happen.
Good luck out there.
