AI governance must involve labour unions

As companies adopt AI, labour and civil society groups need to have a say to boost overall economic productivity and wages

Ever since ChatGPT arrived in our midst, it has generated fear and euphoria in equal measure. Some believe that it is a harbinger of a glorious future where Artificial Intelligence (AI) and AI-driven machines will relieve the drudgery of our lives, boosting productivity and incomes. Others fear that AI-driven machines will replace human workers en masse, creating wealth for a small set of Big Tech firms.

Many firms may be tempted to deploy AI-based tools to track and control employees rather than to empower them. (Thinkstock)

To ensure that the benefits of the new automation technologies are not captured by a few, it is important for the rest of us to have a say on how such technologies are deployed and regulated, two influential American economists, Daron Acemoglu and Simon Johnson write in their new book Power and Progress: Our 1000-Year Struggle Over Technology and Prosperity.

Most economists see technological disruption as a force for good. Even if the disruption leads to initial job losses, it eventually ends up creating many more new jobs, they believe. New technologies can boost the productivity levels of an average worker. This allows a firm to generate more output using the same resources as earlier, and with the same pool of workers. Higher revenues enable the firm to generate higher profits even while paying higher wages to employees. As workers spend more, the economy grows further, creating a new cycle of job creation.

The two professors at the Massachusetts Institute of Technology (MIT) challenge this techno-optimist narrative. They use examples from distant and recent history to argue that the gains of technological change are often captured by elites, unless countervailing mechanisms and institutions force them to mend their ways. For instance, during the first phase of the Industrial Revolution in 18th century England, new factories disrupted the livelihoods of craftsmen and weavers. These factories employed women and children because they could be made to work for long hours at exploitative wages. Economy-wide productivity went up even as wages remained low.

The Dickensian world ended only in the second half of the 19th century as new labour laws allowed workers to form trade unions while placing restrictions on the use of child labour. The extension of voting rights to a wider section of the British population led to other welfare measures that improved access to education and public health facilities. The Labour Representation Committee, formed to represent trade unions, eventually gave rise to the Labour Party, which championed the cause of workers. In the second half of the 19th century, real wages in England nearly doubled, almost keeping pace with the rate of increase in worker productivity.

Similar pressures from worker groups and political parties ensured that the first few decades after World War II also saw wages rise in tandem with productivity growth. New labour safety standards, consumer activism, anti-monopoly regulations, and conservative banking regulations checked the power of big businesses and banks while empowering workers. Since the 1980s, the pendulum has swung the other way, with the balance of power shifting away from workers and consumers once again. Increasing automation and digitisation over the past few decades has eased some aspects of our lives, and enriched Big Tech firms. But it hasn’t boosted overall economic productivity or wages.

“Today we are moving closer to H. G. Wells’s Time Machine future dystopia”, Acemoglu and Johnson write. “Our society has already become two-tiered. On top there are the big tycoons, who firmly believe they have earned their wealth because of their amazing genius. At the bottom we have regular people whom tech leaders view as error-prone and ripe for replacement. As AI penetrates more and more aspects of modern economies, it looks increasingly likely that the two tiers will grow further apart.”

A course-correction is desperately needed if the next wave of automation is to benefit society at large, Acemoglu and Johnson argue. Many firms may be tempted to deploy AI-based tools to track and control employees rather than to empower them. Others may use AI-based tools to replace workers – ostensibly to raise efficiency levels – but without actually realising such gains. Since such decisions are likely to cut costs in the short run and win favour with share-holders, many managers may jump onto the AI bandwagon even when they aren’t sure of the long-term results. In such a scenario, we wouldn’t see any appreciable rise in economy-wide productivity after the AI bubble bursts, even as the human workforce shrinks dramatically.

To tilt the balance of power towards employees, civil society and worker groups need to have a say in how new automation tools will be deployed, Acemoglu and Johnson suggest. Instead of providing tax breaks to companies for buying machinery and robots, governments should provide tax breaks for training or retraining workers, they argue. Big Tech monopolies must be broken up. Governments must not depend on tech leaders to build the very regulations that will govern their conduct, they warn. If anyone else had made such arguments, they would have been dismissed as technophobic Luddites. But it is hard to pin such labels on MIT professors, especially when they also happen to be among the best-known academics of their generation. Acemoglu and Robinson should be read by those who fear and cheer AI alike.

Pramit Bhattacharya is a Chennai-based journalist. The views expressed are personal.