Artificial intelligence is cleverer than you, workers told

Wales TUC - supported by Dr Juan Grigera from Kings College London and Adam Cantwell-Corn from Connected by Data - are investigating how workers in a range of sectors are responding to digitalisation and AI at work. This write up is from the second workshop in the project. The first write up is available here.

“Artificial intelligence is cleverer than you”. That’s what some workers are being told when they challenge the use of the new technology to manage their work.

Convened by Wales TUC, a group of full-time trade union officials in Wales who represent workers from a range of sectors including retail, logistics, care and health met to discuss digitalisation and AI.

Some participants are concerned that workers are losing their jobs because of the introduction of AI and related technologies. Others fear the impact of constant surveillance by their managers through technology. In some industries, new technologies are being introduced without consultation with workers.

Throughout the discussion it was interesting to note that different types of technologies, from digitalisation, automation and artificial intelligence were conflated. This reflects how technology is marketed, used and then developed – with applications building on another.

It also indicates the need for the TUC and unions to do more help workers gain a more detailed understanding of the tools that are in use.

More generally, there is a determination to resist the negative effects of AI and other technologies, through a combination of union organising and demands for regulation. The processes and skills needed by unions to ensure a worker-centric technological transition also need to be developed.

“It's a minefield for the union and its members”

Electronic surveillance through a variety of technologies including AI is widespread and causing concern to workers.

T said that monitoring is prevalent in local government:

“People feel as though they're constantly being monitored,” whether in the office, at home or out and about he said. T added that refuse collection drivers have cameras in their vehicles to monitor for road accidents, which can also be used to pursue complaints against workers in disciplinary procedures.

Workers are unhappy at this intrusive monitoring and unsure how to counter it. There is a growing body of evidence that suggests that surveillance leads to worse outcomes for productivity and workplace culture and can backfire on employers.

In the health service, the digitalisation of patient records is merging with potentially concerning micro-management. “Members are already being pulled up on their record keeping so if AI was used here to identify errors or different ways that staff work that could cause problems,” said S.

He also raised a concern that the reduction of patient care to pure numbers might impact on the relational and personalised aspects of care. “Somebody might be taking longer with the patient than the manager likes, and the software would pick that up” he said.

On a related matter, S said that patients’ data could be held on databases and analysed by third parties in ways that patients and staff were not aware of or able to influence. He said that there was also the prospect that AI could be used to analyse conditions and decide how soon patients receive their treatment. “It's a minefield for the union and its members,” he said. 

A said “we're getting the hard sell now from IT companies. We've been told nurses won't have to deal with note taking. That would be great if it's true, but we always have to maintain professional standards in healthcare. We can't have that being diluted.”

The experience of the full-time officers resonates with wider research that shows that many workers feel distrustful of the manner and impact of the roll out of digitalisation and AI, in particular generative AI. Polling conducted by Opinium for trade union Prospect found that most workers are in favour of government action to protect jobs and job quality. Only 12% of workers believe that the benefits of AI are likely to outweigh the costs.

“When we challenge employers they say ‘AI is cleverer than you.”   

D said that social care workers were very concerned that a data-driven management approach that cuts out human-centred work would be extended to their sector, in both service delivery and working conditions. Care providers could set standardised targets for completing certain tasks for people in their homes, without taking any consideration of the context for the patient or worker.

Retail workers often had good evidence that these target times did not sufficiently allow for the activity but had limited ways to challenge management. “People are being disciplined for failing targets. But when we challenge employers about inaccurate task time targets, they say ‘AI is cleverer than you.’” Workers did not have access or influence over what data was used and what parameters established to set the targets.

In a sector already facing a deep workforce crisis and the resultant impact on citizen’s care, D said that union’s membership in the care sector were as very concerned about this exacerbating an already bad situation.

The blanket application of targets without allowing for context or adjustments is also prevalent in the logistics sector, which has implications for equalities and fairness. “Target times for completing a task are set regardless of age or disability,” he said, adding that “people are managed out if they can't meet their target” regardless of other factors.

T concurred that in his experience he had seen how the targets could have a sexist effect. For example, if someone is pregnant their ability to do work quickly may be lower, but the algorithm doesn't take this into account and the worker will be penalised as a result.

Likewise, T said “People should be accountable for the decisions machines have made”, adding that “poorly made decisions by AI software could end up in disaster.”

Having strong provisions for human review of automated decision making would mitigate this, by enabling a manager to incorporate context and insight that the machine cannot pick up.

However, the UK government’s Data Protection and Digital Information Bill is weakening the requirements for human’s to review automated decision making, as well as access to data to help workers understand on what basis they are being managed.

“It's important that people understand that AI isn't just robots, it's software”

As well as people being managed out of their jobs for failing to meet algorithmically determined targets, technology has contributed to general redundancies. Especially in supermarkets, where far fewer staff work on tills, due to the introduction of self-checkout as well as a boom in online shopping.

As a result, D said that there have been “massive” layoffs of staff, saying that “at one time 1,000 people were working at a large superstore. Now it's 250 and they are mostly part time workers.” A added that many retail staff were female while the new jobs as drivers and warehouse staff are mainly male dominated.

A 2019 study by the Royal Society For the Encouragement of Arts, Manufactures and Commerce found about 75,000 jobs as sales assistants or checkout operators previously taken by women have gone in the last seven years due to automation and e-commerce.

While men lost 33,000 of the jobs over the period 2011 to 2018, these were largely offset by increases in roles in warehouses and as delivery drivers. T said that many of these jobs are precarious and outsourced, meaning that reaching and organising these workers is particularly hard for union reps.

Digitalisation and AI will certainly displace jobs, but it will also create others across the economy. The question will be when those roles might be available, what qualities they will have and how people will be supported to access them.

S noted that in his view unions aren't putting out enough information for members on AI, meaning they would struggle to identify opportunities and concerns, on the shopfloor and sectoral level. “It's important that people understand that AI isn't just robots, it's software,” he said.

On a more hopeful note, T cited a trade union paper which looked at the influence of AI and asked whether it could be used to eradicate monotonous jobs leading to an opportunity for people to be upskilled into more fulfilling work.

“This is what we intend to do’ is the attitude of the retailers. They don't work with unions.”

The role of government and unions in meeting the challenge of AI and technology in the workplace was a matter of intense discussion in the group. It was felt that unions in retail are particularly under the cosh with AI and automation.

A said as the retail sector is fiercely competitive, these technologies and redundancies are being brought in to get an edge on rivals.: “As unions, we don't have access to the employers in terms of negotiating the introduction of technology. We are involved, but it's such a competitive marketplace that technology is rolled out regardless of unions’ views.  The companies buy the technology and just go ahead and introduce it.”

D said that sometimes there is a "shambolic" consultation, which does not offer meaningful engagement. One nationwide grocer introduced technologies for shoppers to scan their own groceries as they went through the store, without any consultation with unions.

“At the same time,” said D, “the retailers reduce the workforce, forcing customers to use the technology. We're being engineered to go this way.  ‘This is what we intend to do’ is the attitude of the retailers. They don't work with unions.”

This unilateral approach contrasts with other comparative nations. In Germany, ‘works councils’ are legally able to be established at firm level in cooperation with trade unions to represent worker interests, including engaging employer funded independent experts to provide technical advice on understanding and negotiating digital technologies.

“Unions must accept that not all AI is bad”

The UK’s regulation and culture of industrial relations is marked by hostility and some of the most repressive legislation in Europe.

However, A was more optimistic. She said that in her experience of representing call centre staff the “key union principles remain paramount” she said. A said that a good illustration is “reasonable practice”.

She gave the example of a call centre in which people's use of time is recorded, even the time they take on toilet breaks. She said that the question from a union perspective is simple: is this reasonable? 

“We should go back to basics as a trade union movement building on existing relationships where workforces are unionised” but that unions need to take a realistic approach to AI. “Fair Work standards should always be maintained when AI is introduced. The floodgates have opened so we can't resist outright. But, strong regulations and standards, including ethics are key requirements.”

The TUC has developed a range of resources to help workers, union organisers and policy makers work towards the goal of a worker-centric approach to AI.

“It's a scary time, but we have to be pragmatic,” she said. Touching on the wider debate of which way the AI transition will go, A said “Unions must accept that not all AI is bad - like the use of diagnostic tools in healthcare. If it improves patient outcomes it is a good thing,” she said.

S agreed and said “AI is the cart and we as unions are the horse - we are being dragged along. The question is how do we put ourselves back in front of the cart?”.

This write up is from the second workshop in a project investigating the impact and understanding of AI and digitalisation among Welsh workers, led by Wales TUC and supported by Dr Juan Grigera from King's College London and Adam Cantwell-Corn from Connected by Data. 

The first write up is available here.