AIFS0015

Written evidence submitted by Unite the union

Unite the union response

This submission is made by Unite, the UK’s largest trade union with over one million members across all sectors of the economy, including manufacturing, financial services, transport, food and agriculture, construction, energy and utilities, information technology, service industries, health, local government, and the not-for-profit sector. Unite also organises in the community, enabling those who are not in employment to be part of our union.

Of relevance to this submission, Unite represents workers in the financial services industry, including banks, insurance companies and outsourcers.

The use of innovative technologies is a rapidly evolving in banking and other financial services.

The 2024 Trade Union Congress (TUC) conference agreed a Unite the union motion which argued that while AI does change the nature of many workplaces, it is vital that “proactive steps are taken by government, trade unions and employers so workers have the skills and support to thrive in an AI-driven economy.”

The global trade union federation UNI Global concluded that “the coronavirus pandemic having led to an unprecedented rise in remote working across the globe, the desire of managers to find digital tools that can support them in leading their teams will only increase.”

It is important to recognise that whilst supporters of AI tools in workplace make vast claims about productivity gains, these tools are untested and there are fears about the huge costs and bias risks.

Unite views the key areas of consideration in relation to AI in the financial services sector are: recruitment, workplace decision making and performance management.

Recruitment and AI

One of the largest uses of AI is in recruitment practices, ranging from creating job adverts to screening applications and automated interview processes.

Unite has concerns that the use of automation in this area without appropriate supervision lacks transparency and can breed bias. There have been reports that some algorithms show bias against women, ethnic minorities, or other groups. This is not to suggest the technology is sexist or racist but that the data is tainted with human bias. This is the basis of many of the issues of management algorithms, they are designed by and evaluated on one homogeneous population.

Unite rejects the suggestion that AI software based on voice analysis or facial screening can assess workers trustworthiness.

It is vital that human beings participate in all aspects of recruitment and job selection. AI tools cannot alone fulfil these functions.

Workplace decisions

Across the financial sector Unite has seen AI used in workplace functions such as shift allocations, leave requests, and allocating tasks to employees.

As well as the advantages of AI management of day-to-day functions there are clear risks with algorithms attempting to remove all downtime and enforce great intensity which is simply not viable for a human being in the workplace.

The extreme pressure when micro rests are removed between calls in contact centres can leave people in a high state of constant stress and anxiety. Hyper efficiency driven by AI dehumanises the workplace leading to huge mental health challenges.

As UNI Global sets out: “For employers, this pursuit of efficiency at all costs can have unintended downsides too: increased staff turnover and burnout, high sickness and absence rates, and reduced employee engagement, ultimately leading to a lack of innovation and a higher risk of costly mistakes being made when workers don’t feel they have permission to think for themselves and are trained to blindly follow instructions from a computer.”

The over dependence on AI for workplace decisions can also remove all compassion and discretion from a workplace. An algorithm cannot recognise the importance of emergency family leave or a mental health crisis for a dependant.

It is vital that human managers continue to oversee workplace decisions and remain accountable to those who collaborate with them. It is true that whilst routine tasks like shift allocations can be initially performed by AI, the need for oversight and change remains with management in our workplaces.

Safeguards around workplace decisions must be established. Staff and trade unions must be told what data is being used to make workplace decisions and challenge the algorithm when needed.

Unite is concerned that the increased reliance on AI across customer contact functions in the financial services sector poses a risk for consumers and most notably the vulnerable as organisations attempt to reduce staff costs.  The expertise, skills and emotional intelligence lost when staff are not engaging with customers is a large risk and not one that can be measured using cost spreadsheets.

Performance management

The use of AI for the surveillance and evaluation of the workforce is increasingly common. Across the banking and insurance sector there are performance management algorithms for:

-          Tracking physical or digital worker activity i.e. logging in or out

-          Reading content of worker emails

-          Measuring outputs

-          Measuring performance

Unite the union has concerns about this level of surveillance using AI and the collection of this data. For instance, managing toilet breaks using software is deeply problematic and intrusive on both ethical and medical grounds.

In contact centres software is used to monitor staff increasingly working remotely. The software can track mouse clicks, keystrokes, and other computer activity. Some organisations even use webcam to ensure workers are at their desks. It is vital to note that excessive monitoring and over focus on performance metrics can lead to workers becoming disengaged.

It is without question vital that all employees are aware of what data is being collated about them.

On a positive note, AI can help a manager identify where a colleague is having difficulties or identify additional training needs in a particular area.

To improve AI productivity in financial services there are examples of organisations who are working with Unite the union to retain and upskill the employees with knowledge of their business. One example is the Unite and LBG skills agreement which actively encourages the retraining of customer facing staff who risk redundancy if all or some of their existing role is moving to technology. Instead of imposing redundancies the employer is working to upskill its workforce and ensure the talent is not lost for the business.

Conclusion

Employers must never use AI to avoid responsibilities for their decisions. Transparent criteria and explainable reason must always be available.

The Labour 2024 policy based on a Unite motion at conference states: “Labour must prioritise legislation to prevent the misuse of AI and new technology in the workplace. The interests of workers and businesses must be balanced; major corporations and bad bosses must not be allowed to use artificial intelligence in a way that exploits working people or diminishes their rights and pay. Workers need protection from AI-powered decision-making, particularly ‘high-risk’ decisions like recruitment, performance assessments and discipline.”

No financial services organisation should use customer ratings to feed directly into employee’s performance review outcomes. This is too easily skewed by bias and discrimination.

As UNI Global expresses perfectly: “Like most tools, they (AI) are neither inherently good or bad; from a hammer to a steam engine to a computer, it is the use to which we put these tools that determines their ethical impact, as well as the rules and frameworks put in place to govern them. Unions have an obvious interest in ensuring that these new analytical tools are both of good quality and used well. The first step is to properly understand what these algorithms are, how they are being used, and the key risks and opportunities that unions need to bear in mind.”

March 2025