Why empowered workers are vital for reaping the full benefits of AI

The demand that workers share equally in the benefits of artificial intelligence (AI) is the latest chapter in a long story of unions organizing to make technology work for all.

Since the dawn of the Industrial Revolution, unions have fought for the implementation of new technologies that respect the humanity of workers as well as uphold the need for safe work and fair compensation.

These were the same goals my union had when I was a shop steward in a jet engine factory in the 1980s. My employer introduced “numerically controlled” technology which lifted the difficult task of turning the lathe’s wheel off of workers. I welcomed the new machines because the work became less physically burdensome, and no workers were displaced in the process.

But there were real concerns about the future uses of this technology, so our union negotiated protections. These included advance notice before new technology was introduced, which gave the union time to understand and negotiate over issues such as job and income security, health and safety, staff training and assignments. Many other unions in manufacturing secured the same commitments.

Similarly over the past few decades, unions have in some cases curbed the excesses of algorithmic management, which accelerates the pace of production by creating highly-pressurized data driven environments. To counter the unrealistic and often unsafe quotas, many unions have negotiated limits on surveillance and protections against discipline, among other points.

Fast forward to today, and workers, their unions and, in some cases, works councils are making many of these same demands with respect to AI.

The rapid introduction of generative AI (genAI) over the past year has spurred head spinning promises about making work better and more productive in many sectors. So, it is no surprise that workers and their unions are speaking out with legitimate concerns about the future of employment in the areas likely to be impacted by genAI.

Indeed, workers in media, care, telecommunications, customer service and beyond are now placing themselves at the centre of this debate.

Members of UNI Global Union affiliate the Writers Guild of America (WGA) were the first workers to strike with generative AI as a central issue. And after nearly 150 days on picket lines, they signed a groundbreaking contract that contains hard-fought language to protect the integrity of the writers’ profession and to guarantee their compensation during this period of experimentation. The two sides will meet twice a year to consult about these developments.

SAG-AFTRA has just concluded another long strike over the use of genAI in connection with images of the actors, negotiating guard rails through compensation and limits. Other media unions have echoed the demand for a transparent, human-centred approach and stressed that AI should be a “tool” which complements existing workers in the craft.

Deutsche Telecom and its works council have just issued a 'Manifesto' which describes a joint approach towards assessing AI and its implications, along with protections for key rights.

And most recently, Microsoft signed a collective agreement with the Communications Workers of America covering hundreds of workers at its ZeniMax video game studio that obligates the company to inform the union whenever the implementation of AI or automation “may impact work performed” by union members. If requested, the company will negotiate over the impact on employees.

One of the only large, real-world studies on using generative AI at work involves call centres, where genAI reduced call times by 14% and improved worker satisfaction, particularly for lower-skilled agents. But key questions remain: will workers be paid more since they are now able to answer more calls? Will the work become even more stressful?

The answers will be determined in large part by whether workers can negotiate and exercise power at the bargaining table.

Recent steps forward in regulation have recognized the role of workers and their representatives, but there is a long way to go.

In the newly-issued Executive Order on AI, US President Joe Biden acknowledged that “all workers need a seat at the table, including through collective bargaining, to ensure that they benefit from these opportunities”.

Unlike in many countries, the effects of technology are a “mandatory subject” for bargaining in the US, and there is a long history of bargaining on these topics. But the executive order itself does not address the significant obstacles towards creating a collective bargaining relationship in the US, or the fact that very few workers are able to exercise these bargaining rights.

And, indeed, the new EU Act on AI nods at this issue by requiring that companies are transparent with workers and their representatives about the introduction of artificial intelligence. The door remains open to develop further regulation on this topic and it is expected that European unions will push for a more robust commitment to enable bargaining on technology.

For regulators who are alarmed about the prospect of job displacement and the new world of work – the promotion of bargaining and social dialogue may be the best remedy. It may seem old fashioned, but unions know how to do this. We just need to open the door.

Judging the success of generative AI solely by metrics of productivity and profit misses the mark. Forecasts of adding $7 trillion annually to the global economy or boosting corporate profits by $4.4 trillion should not be hailed as triumphs if they merely amplify existing inequalities, destabilize democracies and degrade job quality.

We must make sure AI benefits all and ensuring workers’ bargaining power is the way to do it.