The Evolution of Worker’s Rights in the Age of AI
Anne Fehres and Luke Conroy & AI4Media / Better Images of AI / Humans Do The Heavy Data Lifting / CC-BY 4.0
The Evolution of Worker’s Rights in the Age of AI
by Calum McDonald, Engagement and Participation Lead, Scottish AI Alliance
on 30 April 2024
Technology has a tradition of buzzwords. In the world of AI, “Fourth Industrial Revolution” is the phrase of the moment.
Framing the rise of AI technologies like this summons shimmering images of the billowing economic efficiency of the steam engine and the Spinning Jenny, the boosted productivity of the assembly line and the telephone, and the boom in connectivity of the internet and personal computers. When the mirage passes, the image becomes clearer – these advancements arrived in tandem with societal impacts from which equally important innovations developed: labour condition improvements, worker’s rights and the trade union movement. These other innovations ensured that the technological advancements would benefit those delivering them more fairly and more safely.
If AI’s Fourth Industrial Revolution seeks to combine and multiply the efficiencies, productivity and connectivity promises of those that came before — so too the importance of delivering them fairly and safely is multiplied. Just as previous technological advancements spurred crucial innovations in workers’ rights, today’s AI advancements need to be paired with similar protections.
In this tradition of innovation, on 18 April 2024, the Trade Union Congress published their draft of a proposed Bill to regulate the use of AI systems in the workplace, focussing on the interaction between decision-making AI systems and workers, employees and jobseekers.
AI is increasingly being used in recruitment and performance monitoring in the workplace, with 42% of companies globally using AI to “improve recruiting and human resources”, according to IBM’s Global AI Adoption Index, and 40% considering adopting the practice. While the potential benefits of AI are increasingly clear, there is still work to be done to make the protections and frameworks around its use clearer.
With the impact of the Fourth Industrial Revolution being rapidly seen and felt by people in the workplace, with this Bill the TUC proposes a series of parallel rights and obligations that would ensure that the technological advancements of AI do not degrade fairness or safety, ensuring that AI adoption is a benefit to those impacted by it. This approach is reflective of Scotland’s AI Strategy: grounded in trust, ethics and inclusion.
The Bill is progressive, trade union-led, and a step towards a healthy relationship with AI technologies in the workplace. Complementing the Bill, the TUC also worked with YouGov to perform a survey around these issues, investigating what people in the UK think about the use of decision-making AI systems in the workplace. These valuable insights create a picture of a populace who are largely supportive of the Bill’s approach.
Here’s what you need to know about the Bill and the survey:
The proposed rights and obligations in the Bill apply based on how AI is used.
While we are currently experiencing a rise in the use of Generative AI technologies, significantly affecting the world of work and learning, this Bill does not reference them specifically. Rather, it focusses on how AI technologies are used to make or influence decisions being made.
Clarote & AI4Media / Better Images of AI / User/Chimera / CC-BY 4.0
If an employer uses an AI system to make a “high risk” decision whether to “act or not act” concerning workers, employees or jobseekers, the rights and obligations of the Bill would apply.
Actions that the Bill presumes to be high-risk unless proved otherwise include the use of AI in the process of recruitment, assessing performance at work and in relation to terminating employment.
This extends to profiling as well as using data to evaluate, compare, analyse or make predictions.
If AI technologies are being applied in this way, an employer will be obligated to take several steps to ensure that their uses will be transparent, observable, explainable and challengeable.
The Bill protects against discriminatory algorithms.
The Bill considers the full scope of AI development and deployment, which is referred to as the “AI value chain” and includes data collection, data manipulation, model selection, and all other steps in the development of an AI system.
This will allow for the wider aspects of the use of AI to be considered and will help with strengthening fairness around issues of discrimination and bias within the whole lifecycle of AI technologies.
This is further strengthened by the Bill’s proposal to reverse the burden of proof around discriminatory algorithms, with employers needing to prove that their AI use is not discriminatory under the Equality Act 2010.
This is a huge step forward in ensuring that AI technologies protect, rather than degrade, our human rights as well as our employment rights.
The Bill improves transparency around AI and champions the voices of those impacted by AI.
If deploying a high-risk decision-making AI system in the workplace employers would be obligated to regularly perform a specific risk assessment, which is termed a Workplace AI Risk Assessment (WAIRA).
This will increase the transparency and accountability around employers’ uses of AI to make workplace decisions.
Rolled into this system is a process of employee and trade union engagement. The TUC liken this to the current statutory rights relating to engagement within redundancy processes.
This provides a positive interface for the voices of workers to be included in decisions that impact them around AI.
This is coupled with an obligation to register the use of the AI system, an approach adopted in Scotland with the Scottish AI Register, recently mandated for public sector uses of AI. New rights for employees to receive a statement on the use of AI, and the right to request a human review of the decision made, are established as well.
With the combination of these rights above, the world of work can become one which is robust enough to deliver the benefits of AI in a way that is trustworthy, ethical and inclusive.
The Bill prohibits the use of emotion recognition technology to profile workers, employees and jobseekers.
The bill proposes that any AI system that can identify or predict people’s attention, emotions or intentions in the workplace cannot be used to make or influence decisions.
Emotion recognition technology has crept into the workplace alongside the rise in working from home – with employers lauding the benefits of being able to closer track a newly decentralised workplace, and detractors understanding it as a fresh new step on the road of surveillance capitalism.
The use of AI for emotion recognition is barred within the EU AI Act across policing, border management, workplaces and schools – with other organisations also calling for a wider blanket ban.
For an application of AI that can be considered pseudoscience, its prohibition in UK workplaces will allow a more trustworthy approach to the use of AI while bringing it in line with the approaches of global partners.
Most people in Scotland think that employees should be consulted before employers introduce new technologies like AI in the workplace.
In the YouGov survey, 68% of those surveyed in Scotland believe that employers should consult employees before introducing new technologies such as AI, compared to 24% who think that they should not, and 9% who don’t know.
Involving everyone in the decisions made around AI is an important tenet of trustworthy, ethical and inclusive AI – and these figures show that this extends to the workplace.
People in Scotland are notably less supportive of using AI for workplace decisions compared to other UK nations.
The YouGov Survey canvassed people on their support and opposition to employers using AI to make decisions in the workplace in three areas: hiring, performance management and bonus decisions, and firing.
In Scotland, 6% of those surveyed supported employers using AI in making hiring decisions. In England, this was at 15%. In Wales, this was at 10%. In Northern Ireland, this was at 18%.
People were generally more supportive of employers using AI in performance management and bonus decisions, but with the trend of people in Scotland being less supportive than across the UK – with 14% in support in Scotland, 20% in support in England, 19% in support in Wales, and 20% in support in Northern Ireland.
The least support for employers using AI to make decisions in the workplace was around firing, across the whole of the UK.
In Scotland, only 3% of those surveyed supported this use of AI. In England, support was at 8%. In Wales, 5%. In Northern Ireland, 9%.
These figures show that while there is a variance across the different uses of AI to make decisions in the workplace, support is generally very low, and it is consistently the lowest in Scotland.
Most people believe that humans should have the final say when AI is used in employer decision-making.
Across the same three realms of hiring, performance and bonus decisions, and firing, most people surveyed hold that a human should have the final say.
In the UK, for hiring, the figure was 88%. For performance and bonus decisions, 83%. And for firing, 90%.
From this YouGov Survey, it appears that while people in the UK are majorly against employers using AI to make these kinds of decisions, they are also solid supporters of the concept that humans must be in the loop if it is.
To deliver on Scotland’s AI Strategy, our relationship with AI technologies must put trust, ethics and inclusion at the centre of conversations around AI. The protections, rights and obligations proposed within TUC’s Bill allows trust, ethics and inclusion to be at the heart of these conversations in the workplace. And from YouGov’s survey, the Bill seems well-pitched to the mood of the people it will protect.
If we want technological revolution, we need societal evolution — this proposed Bill is a solid match to the shimmering images of a Fourth Industrial Revolution.