How will AI change work?

A new report pays particular attention to the way AI will change hiring, monitoring, and the managing of staff. By Annie Gray.

A University of Otago report looking at the impact of artificial intelligence on jobs and work in New Zealand includes a major chapter on how AI will change work.

The report, which is funded by the New Zealand Law Foundation, investigates what it will be like to work alongside AI and assesses regulatory changes designed to maximise the benefits and minimise the harms of AI in the workplace.

The report, by Professor James Maclaurin, Professor Colin Gavaghan and Associate Professor Alistair Knott, notes in the executive summary that it’s difficult to assess claims that AI will generally enhance jobs.

“As with previous types of automation, AI will sometimes relieve us of onerous and unpleasant tasks and sometimes leave occupations deskilled, reducing the mana and bargaining power of workers.”

The summary says it pays particular attention to the way AI will change hiring, monitoring, and managing staff.

“AI promises to make hiring faster and less expensive, to better match applicants to jobs, and to help increase diversity in the workplace. However, there is also a risk that AI will introduce unfair bias in job advertising and in the vetting of job applicants.

“When algorithms are trained on historical data that reflects historic discrimination or inequality, this use of ‘dirty data’ is likely to skew outcomes for already disadvantaged individuals and groups.”

The Impact of Artificial Intelligence on Jobs and Work in New Zealand’s executive summary says that AI is already used widely in the recruitment of workers, in targeting job ads to potential employees, shortlisting applicants, and evaluating the performance of candidates in interviews. “There are dangers of unfair discrimination at every stage.”

It says that AI can also be used in management of workers, performing a variety of tasks that were previously the preserve of human managers.

“New AI-based management methods promise to improve accuracy and efficiency in decision-making, and to reduce opportunities for human favouritism and unconscious biases. Algorithmic management also has the capacity to improve the lives of workers. It could be deployed in consultation with workers, so as to accommodate the needs of workers with families, enhance leisure time and educational opportunities.”

However, it says that at the other end of the scale, it could leave workers feeling isolated and de-humanised, or placed under greater levels of pressure or surveillance.

The report says it’s essential that AI does not entrench deep disparities between the power of workers, managers, and capital owners.

“An algorithmic management recommendations or system trained on profiles of previous workers could make predictions based on characteristics that are irrelevant or discriminatory. Such AI threatens to entrench historical discrimination.”

Managerial decisions in general are covered by both the Employment Relations Act 2000 (ERA) and the Human Rights Act 1993, “but issues may arise regarding implementation, compliance monitoring and enforcement of the legislation”.

It says that a range of auditing tools already exist to help employers avoid inadvertent discrimination, but as with recruitment, concerns exist about the criteria employed by different tools – what notions of bias or fairness they use, for instance, and what jurisdiction’s laws they are aligned with.

“The opacity of AI systems that could potentially inform discipline or dismissal makes it difficult for those affected to assess whether employers have complied with their legal obligations. Employers should make sure task allocation algorithms are ‘explainable’, in terms that are meaningful to their workers.”

The report also notes that another concern relates to growing use of AI-enabled workplace surveillance, “which can threaten the autonomy and dignity of workers”.

“In due course, these technologies may require specific legislative attention.

“In the meantime, we would welcome attention from the Privacy Commissioner to the possibility of a code of practice directed at workplace surveillance technologies, or perhaps workplace surveillance more generally.”

It says that WorkSafe could also have a role to play in regulating the potentially harmful effects of algorithmic management and surveillance.

Meanwhile the section of the report also points to the fact that increasingly AI will appear in the workplace in the form of collaborative robots (cobots). “These may greatly enhance productivity but will force a rethink in the way we address safety concerns.

“It will no longer be possible to ‘separate and contain’ such machinery, as it will be working amongst us. WorkSafe should consider issuing a code of practice dealing with workplace robots and particularly ‘cobots’, perhaps based on the ISO standard for collaborative robots.”

The report says that AI is a new phenomenon in most workplaces and it is poorly understood by many of those who now use it or are affected by it.

“Efficiency gains and cost decreases will drive its rapid adoption in many industries. New Zealand government and regulatory agencies should facilitate this transition to ensure that harm does not come to workers or other stakeholders.”

It highlights:

• Consideration should be given to requiring hiring tools to include functionality for bias auditing, so that client companies can readily perform audits of each recruitment decision process.

• Discussions should take place between developers, employers’ organisations, unions and other relevant stakeholders to consider the development of guidance and standards for auditing of algorithms used in employment situations.

• The New Zealand private sector should use algorithm impact assessments, assessing factors such as privacy and equality. Given the likely challenges for smaller employers developing these in-house, templates should be made available along the lines of those developed by the NZ Privacy Commissioner for Privacy Impact Assessments.

A copy of the full report can be found at https://www.otago.ac.nz/caipp/research 

Visited 10 times, 1 visit(s) today
Close Search Window