Pearls and Irritations

In future AI is the new technology which is likely to have the greatest impact on our economy and our society. But how AI is used and developed is a choice, and so far AI has been predominantly focused on continuing the emphasis on automation. To realise the full potential of AI and minimise its harmful effects there needs to be more public consultation and regulation to determine the future direction of AI.

The previous article in this series of three discussed how the choice of technologies and the sharing of the benefits in the past depended on the extent to which there were institutions to balance the power of elites.

Looking ahead, artificial intelligence or AI is now sparking a revolution which is likely to be the most important technological change affecting the labour market over the next decade or more. As the OECD has said:

“Using a fast-evolving suite of algorithms and statistical models – in particular, machine learning – increasingly available big data and falling costs of computer capacity, AI has made rapid advances in its ability to supply answers to problems where formal rules are impossible to codify, and where humans have until recently had a comparative advantage in inferring decisions from their training or past experiences.”

The OECD further found that “AI appears different from previous digital technological changes in several ways: (i) it significantly expands the range of tasks that can be automated beyond just routine, non-cognitive tasks; (ii) AI is a general-purpose technology, meaning that nearly every sector and occupation will be affected; and (iii) the speed of development is unprecedented.” “In some areas, it has become difficult if not impossible to distinguish its output from that of humans.”

As a result, “high-skilled occupations have been most exposed to recent advances in AI, including business professionals; managers; science and engineering professionals; and legal, social and cultural professionals.”

The critical issues for the economy and more specifically the labour market are:

• The extent to which AI will complement or displace existing workers and/or lead to additional demand for workers with new skills.

• Which workers are most likely to be impacted and what does that imply for future equality.

Separate reviews of the existing evidence and literature by the IMF and OECD find that:

• The available data (to the OECD) suggest that the share of firms that have adopted AI remained in single digits last year, although approximately one in three large firms have done so.

• There is little evidence of significant negative employment effects due to AI so far, but this may be because AI adoption is still relatively low and/or because firms so far prefer to rely on voluntary workforce adjustments.

• In advanced economies, such as Australia, about 60 per cent of jobs may be impacted by AI.

• Roughly half the exposed jobs in the advanced economies may benefit from AI integration, enhancing their productivity and wages, while for the other half AI applications may execute key tasks currently performed by humans, which could lead to lower wages, and some of these jobs may disappear.

• University educated workers are better equipped to move from jobs at risk of displacement to high-complementary jobs.

• The gains in productivity, if sufficiently strong, could result in higher growth and higher incomes for most workers, [although the evidence so far is that AI has not lifted productivity growth].

• Model simulations suggest that AI will likely worsen labour income inequality and also the inequality between the developed and developing countries.

In sum, the assessments by the OECD and IMF regarding the impact of AI on the labour markets of advanced economies is fairly reassuring, that not a lot will change in terms of the availability of jobs and their relative wages. In addition, a survey of academics and technology leaders found that the overall majority thought that although there were downsides, AI would bring widespread economic and societal benefits.

However, these assessments of the impact on the labour market mainly reflect experience so far and hypotheses about what is possible, but it is questionable how far that information provides an adequate guide to the future.

Acemoglu and Johnson in their new book, Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity, agree that “Reassuringly, AI does not appear to be advancing so much that it will create mass joblessness.” But as they also note: “the evidence indicates that AI so far has been predominantly focused on automation, …. [and] most of the burden of AI automation to date has fallen on less-educated workers.”

Interestingly, Acemoglu and Johnson maintain that so far AI has not “turned out to be anything as productive or amazing as its boosters maintain, On the contrary, AI-based automation often fails to increase productivity that much.”

Nevertheless, that still leaves the question of what will happen as AI tries to realise its ambition to expand to non-routine tasks. As Acemoglu and Johnson comment, “If machines can be autonomous and intelligent, then it is natural for them to take over more tasks from workers”.

However, Acemoglu and Johnson also think there are limits to how usefully AI can be extended into areas presently relying on human judgements. They argue that “Human intelligence derives its strength from being situational and social”, and their concern is that AI by “taking away initiative and judgement from humans, sometimes makes things worse, not better.”

Acemoglu and Johnson fear that “These lessons about human intelligence and adaptability are often ignored in the AI community, which rushes to automate a range of tasks, regardless of human skill.” In their view “Statistical approaches used for pattern recognition and prediction are ill-suited to capturing the essence of many human skills.”

Acemoglu and Johnson acknowledge that “machines could be harnessed to the service of humans as a complement to our skills”, but this is not “the main area in which AI dollars are being invested. Industry’s focus continues to be on extensive data collection and the automation of narrow tasks based on machine learning techniques.”

They support this conclusion with examples of how AI has enabled flexible scheduling and monitoring of workers, which has then allowed employers to get more work out of workers and cut wages.

However, Acemoglu and Johnson insist that “None of this [focus on automation and cost cutting] was inevitable.” As with all previous technological breakthroughs, “The current approach that dominates the third wave of AI based on massive data harvesting and ceaseless automation is a choice. It is in fact a costly choice, not just because it is following the bias of elites toward automation and surveillance, and damaging the economic livelihood of workers. It is also diverting energy and research away from other, socially more beneficial directions for general purpose technologies.”

A better alternative, according to Acemoglu and Johnson would be “Instead of fixating on machine intelligence, we should ask how useful machines are to people, which is how we define machine usefulness (MU). Focusing on MU would guide us toward a more socially beneficial trajectory, especially for workers and citizens.”

Acemoglu and Johnson then discuss how “machines could be harnessed to the service of humans as a complement to our skills.” For example:

• “The same statistical techniques used for task automation can also be used for identifying in real time groups of students who have difficulties with similar problems, as well as students who can be exposed to more advanced material. The relevant content can then be adjusted for small groups of students.”

• “The situation in health care is similar: the right type of MU can significantly empower nurses and other health care professionals, and this would be most useful in primary health, prevention, and low-tech medical applications.”

Finally, based on their assessment of the impact of AI and its potential, Acemoglu and Johnson conclude that “Digital technologies did not have to be used for just automating work, and AI technologies did not have to be applied indiscriminately to amplify the same trend.” “There is nothing foreordained about this path of technology, nor is there anything inevitable about the two-tiered society that our leaders are creating. There are ways out of our current conundrum by reconfiguring the distribution of power in society and redirecting technological change.”

Interestingly, both the IMF and OECD support the need to ensure robust regulatory frameworks for AI, and the OECD is also on record that “collective bargaining and social dialogue have an important role to play in supporting workers and businesses in the AI transition.

But Acemoglu and Johnson seem to have in mind a more fundamental change in the regulation of AI. They want to break with the tradition whereby the direction of technological change has always been decided by elites to their advantage, and instead ensure that the choices made aim to optimise AI’s usefulness to society.

And of course, there are powerful reasons for regulation of AI to protect our social cohesion and democracy, but that is a separate important discussion in its own right.

 

Read part 1 of the series:

Sharing the benefits of technological progress