Since the outbreak of the pandemic, there has been a dramatic increase in the use of surveillance tools invading workers’ homes as they have to work remotely. This has gone hand in hand with the acceleration of the trend towards the use of artificial intelligence at work and raises ethical, transparency, accountability and privacy concerns.
There is also a renewed focus on ethical AI at EU level. UNI Europa’s priority is to ensure that this results in aligning the applications of artificial intelligence with workers’ fundamental rights. Technology should deliver improved working conditions instead of increasing mental and physical health problems, disempowering workers and impeding people’s privacy rights. We know that these issues can only be addressed through a focus on workers’ collective power to keep the work-related design and application of AI ethical, transparent, accountable and privacy-respecting.
Speaking on behalf of UNI Europa at valuing labour in the age of AI, a webinar by the Socialist and Democrats group in the European Parliament and FEPS, Birte Dedden outlined some key issues as well as avenues to address them from the workers’ perspective.
Invasive surveillance
A recent study suggests that 75% of corporations use data to understand workforce performance and productivity. The tools and methods used to extract this data are in many cases being applied without workers having a say over which data is being gathered, how it is being gathered or for what purposes it is being gathered.
Surveillance tools are already being rolled out in corporations that are known to suppress workers’ collective rights. At Teleperformance, face biometrics are used to ID workers on their computers. Amazon warehouse workers are required to wear a device that tracks their every move. There have also been leaks revealing Amazon’s secret programme to spy on workers outside of work, gathering private data to determine whether they are involved in efforts to unionise.
Many workers don’t know what data employers collect on them at work. According to Prospect in the UK, 1 in 3 workers has no confidence that employer would use collected data appropriately. That same study finds that 80% of workers would be uncomfortable with camera monitoring.
Algorithmic management
The data that is being gathered is increasingly being analysed and acted upon, not by people, but by automated processes.
Hiring has been one of the areas in which this has been highly prevalent. Talent screening software is being used to scan prospective employees’ social media feeds and automatically flag up potential problematic behaviour. AI-based tools that analyses facial expressions, tone of voice and accent during an interview are also in use. Using historic data to determine decisions about the future carries risks of discrimination, as illustrated by Amazon’s AI hiring tool that discriminated against women candidates.
Similar approaches are used to determine people’s everyday work. In contact centres, workers are frequently monitored and AI software is used to keep them in check if they are speaking too fast, sound too tired, or insufficiently empathetic.
Solutions
There is a clear need for social partners to be both involved to deliver an ethical approach to the design and implementation of AI-driven surveillance tools. Some key principles to guide this work include:
Justification: Data collection or surveillance of the workforce should be for a clearly justifiable purpose. Personal or other sensitive data should not be collected without explicit consent.
Transparency: Workers should have a right to be made aware of any algorithmic management tools being used that affect them and to challenge the use of any tools they consider harmful.
Rights-based approach: Clear red lines for unethical AI systems impacting fundamental rights, workers’ rights and privacy. The burden of proof should be on the organisation developing the AI system.
Human-in-command: Algorithms should advise, humans should decide. Workers should always have the right to appeal to a human authorised to override the algorithm.
Check out UNI Europa’s position paper on artificial intelligence for more.
