"We need to make sure to protect the right to organize, form and join trade unions against all union busting practices" @EstherLynchs speech at @ETUI_org event on "Shaping a new social contract for AI"
— EUROPEAN TRADE UNIONS (@etuc_ces) May 31, 2024
👉Every new #AI technique that comes out is used against workers and their… pic.twitter.com/wrnpwaCuqR
Shaping a new Social Contract for AI
Esther Lynch, General Secretary - European Trade Union Confederation
Launch of the ETUI publication "Artificial intelligence, labour and society" - 31 May 2024
Dear colleagues,
I am delighted to have been afforded the opportunity to give the keynote address today.
I begin, by thanking Aida Ponce Del Castillo for her leadership and human intelligence in creating and bringing to life this important ETUI publication Artificial Intelligence, labour and society.
I am particularly grateful Aida, for your chapter that underscores the need for the human in control principle and for the application of the precautionary principle as essential elements for the governance of AI, and importantly for protecting the health and safety of workers for the use of AI.
Friends, artificial intelligence is transforming the way we work.
Important decisions about people’s working lives, such as whether they get a job and how they are managed, if and how much they will be paid, or promoted, are increasingly made by technology rather than a human manager.
These new technologies present us all with opportunities for a better world of work, safer, with less monotonous work and with more reward for greater productivity. But there are also risks of unfairness, discrimination, work intensification, stress and injury, with work becoming an increasingly alienating, isolating and lonely experience.
The ETUC in our Manifesto for the 2024 Elections has called for the effective regulation of AI.
Our demands include four Directives that are part of our consideration of how to shape the new social contract for AI.
Four Directives: (i) a just transition Directive to deal with anticipation and management of change at work, including the right to training, (ii) a Directive on Psychosocial Risks at work, (iii) a Directive on telework and the right to disconnect, and (iv) a Directive dealing specifically with AI at work.
These four Directives are necessary, as the EU AI act does not deal with the world of work, and because ethical guidelines and voluntary agreements - while they can be well meaning and often helpfully spotlight the challenges that need to be addressed in different sectors - have been proved time and time again to be hopelessly inadequate to meet the challenge of the mismatch in power present in the employment relationship, especially in employments that are not covered by a collective agreement.
Our assessment is that new rights are necessary, and I would like to set out some headlines of an AI at Work Directive. I stress it is not the full or final list.
The first big headline is that collectivism matters more than ever in the face of technological control. This calls for even greater support for trade unions and collective bargaining at all levels.
Collective bargaining provides the best system of co-governance of new technologies, be it at the workplace, within occupations or at sector level. The AI Directive must guarantee genuine involvement, the right to collective bargaining on AI as a prerogative for trade unions. It should also ensure that when implementing AI, redundancy is not the go-to solution, but rather there is a guarantee of employment, fair working conditions, up-skilling and re-training.
Collectivism means that the benefits of AI can be fairly shared. Collective bargaining is the method through which the productivity gains of AI can be fairly shared with the workforce, underscoring the importance of increasing the number of workers covered by a collective agreement to 80% as required by the Directive on Adequate Minimum Wages in the EU. Empowering trade unions to bargaining on the gains brought about by AI is essential for distributional fairness and to prevent excessive concentration of wealth created by AI in the hands of a few at the top.
Reinforcing collective bargaining is also key to ensure that human labor does not lose its value.
Collective bargaining also provides protection from the use of 'individual consent' by employers to justify use of workers data, GPS, health and fitness trackers, CCTV or other AI monitoring that gathers data pertaining to the worker, a collective agreement can afford much greater protection, reinforcing the individual rights already provided under GDPR rules.
In addition, the right to your own image, your ideas, your privacy these are not just something that celebrities should be able to protect but workers must be given a realistic way to protect their rights too.
And colleagues, it is ESSENTIAL to protect the right to organise, to form and join trade unions from AI enabled union busting tools. In the hands of unscrupulous employers, AI is already being used to profile for likely union members organisers and activists, spy on their communications or meeting activities and then to retaliate against them.
This brings me to the second big headline issue, the need to provide legal protection for human expertise, human relationships and to guarantee the dignity of the human person.
I want to mention here a new right, the right to your human personality at work. There is a very real danger that AI emotion monitoring tools will be used to enforce the 'universal company employee', eternally smiling, always compliant and with all human emotion and personality supressed.
Right now, workers are required to hide emotions under the surface, masking actual emotions such as using a fake smile to hide one's true feelings, whereas when AI is deployed, workers will be required to deep fake it, modifying one's real feelings as any infringements will be spotted and punished. We should prevent this - and we can prevent this - by prohibiting the use by employers of emotion monitoring tools on their employees.
Colleagues, I am convinced that protecting human dignity and the value of human labour, respect for the worker and work that we do will become a defining battle ground in our struggle for a fair future of work and regulating AI will be the front line.
This makes our actions to shape the new Social Contract for AI an urgent task. We need legal protection to guarantee from the ground up that technology works for humans, rather than the other way round.
This means establishing the human in control principle so that humans, and in the workplace, I mean workers, will retain control over AI, not just technological control but also control over if, when and how AI is used and with what safeguards.
There must be no question that the precautionary principle applies to AI and that the general principles underpinning the Health and Safety at Work Directives extend to the use of AI in the workplace. Therefore, the same considerations that would usually apply when assessing health and safety risks apply to the use of AI too. The safe use of robots will need particular attention. The impact of AI on the safe organisation of way, for example how AI will change the content of a day at work. It can be expected that humans will be required to undertake more complex tasks during the working day as the predictable parts of the job will be done by AI, increasing the stress and pressure experienced during the working day for the worker. Targets, and task completion times that are increasingly tracked to the millisecond are already resulting in an decrease of wellbeing this is why we are also calling for a Psychosocial Risks at Work Directive to deal with all health and safety aspects of AI deployment at work.
The third big issue is that it’s vital that technology doesn’t entrench existing inequalities.
Strong collective bargaining is an important antidote against this.
Making everyone in the AI value chain — including tech companies — liable for discriminatory algorithms will also provide an important protection against this, as will mandatory equality impact assessments and making it clear in legislation that discriminatory data processing is always unlawful.
Workers must also have the right to a human review of important decisions that affect them. This means the right to in-person engagement — people should have a right to meet a person, in person, when significant decisions are being made about their working lives.
Colleagues, we need to exercise our moral imagination. Recall the hard learnt lessons of what can go wrong in the world of work, how the imbalance in power creates particular vulnerabilities.
As trade unions we must act to prevent the detrimental use of AI with all tools at our disposal. We can push for patent and product law for example, to support the rule that no AI may be put on the market that does respect the human in control principle or that presents a risk to human dignity.
The discussion today is only the start, the issues are wide ranging from safety, to appropriate transparency and explainability, accountability and governance, the recognition of new rights and the protection of existing rights in the new context of AI, collectivism, contestability and redress and ensuring a fair share of the productivity gains with the workforce, how AI will affect jobs over time and how it will impact on income and wealth inequality.
There is a role for the EU and national governments to try to steer AI innovation in ways that are more productivity-increasing and job-creating, rather than job-destroying and inequality increasing. The need to deal with the impact of AI on jobs and incomes reinforces the our call for a European Industrial Policy backed up with investments for quality jobs in every sector and region. Inclusive and progressive reforms can strengthen the labour force against AI-fuelled inequality, Governments can encourage this through social conditionalities to be attached to investment and public procurement rules.
In conclusion, AI is a valuable tool, the way it is deployed is a choice, collective bargaining and social dialogue are the tools to secure the fair implementation and use of AI at work.
The new Social Contract for AI must deliver the human in control principle, and be based on the precautionary principle.
It must be underpinned by real support for collective bargaining, it must be guaranteed with legally enforceable rights, it must promote economic equality and job creation, guard against all forms of union busting and discrimination and guarantee the dignity of the human person and reinforce the value of human labour.
Working together we can win this social contract, our future depends on it.