THEMATIC BRIEF

The impact of Artificial

Intelligence on the
labour market and the
workplace: What role for
social dialogue?
 |1

Global Deal Thematic Brief

The impact of Artificial Intelligence on the labour market

and the workplace: What role for social dialogue?

DECEMBER 2021

Disclaimer Note: The opinions expressed and arguments employed herein do not necessarily reflect the official
views of the OECD member countries, ILO member states or Global Deal partners.
2|

Acknowledgments
This Global Deal thematic brief was authored by Sandrine Cazes of the OECD Directorate for Employment,
Labour and Social Affairs. The report has also benefitted from valuable advice and suggestions made by
the Global Deal Senior Advisors and the OECD Employment, Labour and Social Affairs Committee
(ELSAC). The Global Deal Bureau, comprising the Government of Sweden, the ILO and the OECD, have
also provided practical suggestions and guidance. The opinions expressed and arguments employed
herein do not necessarily reflect the official views of the OECD member countries, the ILO or Global Deal
partners.
 |3

Introduction – The impact of Artificial Intelligence: main issues and new

challenges
1. Artificial Intelligence (AI) technologies - machine-based systems that can, for a given set of human-
defined objectives, make predictions, recommendations or decisions influencing real or virtual
environments 1 - will have an important impact on labour markets, workers and the workplace. As a result,
workers and firms will need to grapple with significant transitions and adjustments. Recent years have
seen rapid advances in the development and adoption of AI technologies, particularly in the areas of image,
text and speech recognition, computer programming and predictive analytics. These developments have,
for instance, fuelled new fears about large-scale job losses stemming from the ability of AI to increasingly
automate not only repetitive but also non-repetitive tasks 2, and its potential to affect every sector of the
economy. At the same time, AI will also create entirely new tasks and occupations, and drastically change
the nature of others: as a result, some tasks may become safer and less monotonous, but in others there
may be a risk of de-humanisation and even de-skilling. Moreover, the abilities among different workers and
different firms to take advantage of the benefits that AI brings could also trigger inequalities in the labour
market: while not new, the risk that workers who adapt more slowly to technology adoption may be
excluded is likely to be magnified by AI diffusion, as “letting algorithms drive economic activity will further
privilege the privileged” (Schor, 2018[1]).
2. Furthermore, there are concerns about autonomous decision-making in the workplace, particularly
in HR and management processes, which are linked to excessive surveillance, intrusive practices and
ensuring fundamental workers’ rights: decisions taken by AI, when tracking drivers or nurses connected
with applications, or when recruiting, firing 3 or monitoring workflows and performance, could change the
nature of the relationship between firms and workers, but also raise more fundamental ethical questions,
linked to the risk of potentially biased decisions, discrimination, data protection and human rights. AI
technologies that are able to process biometric data, but also have facial recognition and even detect
emotions and behaviours may, for instance, carry a high risk of privacy breaches and a violation of human
dignity.
3. However, AI, also brings real opportunities to create new business models and new types of jobs,
as well as alter the nature and task composition of existing ones. Since AI has the potential to complement
and augment human capabilities, it can lead to higher productivity, greater demand for human labour and
improved job quality (Lane and Saint-Martin, 2021[2]).
4. While AI therefore, as with any technological change, will bring both risks and benefits, many
experts suggest that its effects on labour markets are likely to be magnified by the speed and large potential
for application across multiple sectors and occupations (Brynjolfsson, D. Rock and C. Syverson, 2017[3])
and the greater power imbalance it will trigger (De Stefano, 2019[4]). The impact of AI, whether positive or
negative, will very much depend on how it will be implemented at the workplace level, what will be the role
of regulation (e.g. international legislation, such as the EU GDPR 4, international labour standards, national

1While there is no widely accepted definition of AI, this brief uses the definition of an AI system established by the
OECD’s AI Experts Group (AIGO): “An AI system is a machine-based system that can, for a given set of human-
defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. It uses
machine and/or human-based inputs to perceive real and/or virtual environments; abstract such perceptions into
models (in an automated manner e.g. with machine learning (ML) or manually); and use model inference to formulate
options for information or action. AI systems are designed to operate with varying levels of autonomy”.
2 According to an ETUC-commissioned survey that was carried out on trade union representatives across Europe
(Voss and Riede, 2018[47]), job destruction and job creation due to automatisation ranked as the most important
concern and opportunity respectively for trade unions.
3 (Lecher, 2019
[56]) reporting that Amazon’s system tracks workers’ productivity rate and automatically generates
warnings and terminations in light of quality or productivity levels without input from a human. Amazon, however,
replies that supervisors still have the possibility to override the automated process.
4 On 25 May 2018, the European Union replaced the Data Protection Directive (European Union, 1995), by the EU
4|
legislation and collective bargaining) in governing AI diffusion and the extent to which all stakeholders will
be involved. Previous OECD work has highlighted the instrumental role that social dialogue and collective
bargaining can play in the changing world of work by easing transitions and spreading best practices in
terms of the introduction of new business practices, training and safeguarding quality (OECD, 2019[5]). The
key role of social partners is further underlined by the ILO’s recommendations on employment policy (see
for example ILO R169) as well as their recent survey on employment (ILO, 2020[6]).
5. More specifically, social partners have a key role to play in determining what technology and
training are adopted; helping companies define tailor-made and fair solutions to organisational and
technological changes at workplace level and enhancing the quality of the working environment (OECD,
2019[5]). Depending on the various national regulatory settings as well as practices and traditions, social
partners can voice concerns, inform and advise their members through codes of conduct and guidelines
about privacy and ethical rights, as well as participate in decision making at the workplace. Finally,
collective bargaining, provided it has high coverage while leaving some margins of flexibility, can foster
inclusive and dynamic labour markets when systems are co-ordinated 5 (OECD, 2018[6]; OECD, 2019[5]).
6. This brief aims to assist all concerned stakeholders in understanding how social dialogue and
collective bargaining can facilitate AI transition at labour market and workplace levels. It provides key
insights on the instrumental role social dialogue and collective bargaining can play in complementing public
policies in this adjustment process, while demonstrating concrete examples of social partners’ initiatives in
raising voice, advising or signing innovative collective agreements. Finally, it discusses the impact that AI
tools and systems may have on social dialogue itself.

The role of social dialogue in facilitating the AI transition at labour market and
workplace levels

Social dialogue and collective bargaining can complement government efforts in

enhancing labour market security and adaptability

7. Social dialogue and collective bargaining have a key role to play in smoothening the impact of AI
in the labour market, facilitating the introduction of new technologies as well as complementing public
policies in the deployment of re-training and upskilling programmes. OECD work on displaced workers
(OECD, 2018[6]) has highlighted the significant role that collective bargaining, in particular at sectoral level,
can play in enhancing labour market security and strengthening workers’ labour market
adaptability. As rapidly evolving demands for products and services and AI diffusion will affect skills
needs, social partners can help ensuring that if AI technologies are implemented in the workplace, they
complement, rather than fully replace, existing occupations. Social partners can also provide active support
to those workers displaced from their existing jobs to help them back into new ones. The Swedish Job
Security Councils (JSCs) and the Austrian Outplacement labour foundations, two institutions owned jointly

GDPR framework (European Union, 2016). The EU GDPR introduced new rules governing the collection, process,
and free flow of personal data regarding data subjects in the European Union. When data originating in EU member
states are transferred abroad, the EU GDPR ensures that personal data protections travel with them. The EU GDPR
ensures high level of protection with consistency, while eliminating barriers to the free flow of data within the Union
(European Union, 2016). Some controversies exist however on whether the existing EU GDPR framework adequately
and sufficiently protect workers from the downsides of AI technologies, including for ensuring work-related stress and
excessive pressure from intensive work schedules defined by AI.

5
OECD work (2018[6]; 2019[5]) explores the link between collective bargaining systems and a series of labour market
outcomes. The main results show that wage inequalities are highest in systems with no collective bargaining or in
systems with firm-level bargaining only. Moreover, coordinated systems – in which social partners negotiate for
different groups of workers follow collectively defined targets when negotiating wage increases – are linked with higher
employment and lower unemployment, including for women and young people, than fully decentralised systems, in
which negotiations are firm-specific and not coordinated at all. Finally, coordination also increases labour market
resilience, in helping social partners to account for the business cycle situation and macroeconomic effects of wage
agreements on competitiveness when negotiating.
 |5
by employers’ organisations and unions, are among the most notable examples of this (OECD, 2015[7];
OECD, 2019[5]).
8. Beyond supporting displaced workers, social partners can play a key role in anticipating skills
needs: though their representation in skills councils and training provisions in collective agreements, as
well as their involvement in the process of developing, funding and managing adult educational and training
programmes, the role of social partners has been found to be beneficial both in terms of the quality of
training and accessibility for all workers (Boheim and Booth, 2004[8]; Dustmann and Schönberg, 2009[9];
Verma, 2005[10]; OECD, 2019[5]). This latter point is particularly crucial in light of the growing risk of
exclusion of workers lagging in AI technological adaptation. In the majority of OECD countries 6, social
partners are involved in skills assessment and anticipation exercises OECD (2015[7]). More generally, in
several OECD countries 7 social partners are represented on sectoral skills councils which produce
industry-specific long-term projections to ensure that current qualifications meet future demand for skills
(OECD, 2019[11]).
9. Crucially, social partners can help ensure that workers also get enough lifelong training to adjust
to ongoing changes. As highlighted earlier equipping workers with the right skills, in a context of
technological and occupation changes, is a key challenge for shaping an AI transition that is more inclusive
and rewarding. Access to long-life training for workers can be negotiated and secured in collective
agreements, and is an increasingly important issue of collective bargaining. 8
10. In a time of accelerating changes stemming from AI transition, the role of social partners in
managing transitions, anticipating and filling skills needs may be increasingly important. Investing in skills
may not only be important to strengthen labour market adaptability and help workers in case of
displacement, but could also be a winning and revitalisation strategy for social partners to reach out to new
members (Klindt, 2017[12]).
11. Along these lines, both unions and employers’ organisations have engaged in outreach and
awareness campaigns highlighting the need for new competences that will be required to work with digital
tools, robotics and data (ILO ACT EMP and IOE, 2019[13]; BusinessEurope, 2019[14]; ETUC, 2020[15]; UNI
EUROPA ICTS, 2019[16]). BusinessEurope (2019[14]) outlines the key role of social partners in facilitating
the establishment of a “data culture” and “awareness of AI”. The organisation also points that a “highly
educated and entrepreneurial workforce is required to promote the growth and use of AI in Europe”. The
European Trade Union Confederation (ETUC) calls for “AI and digital literacy schemes”, as education and
transparency of AI systems will be important for workers to understand, and be part of, a fair
implementation of new technologies (ETUC, 2020[15]). It proposes building “data literacy” through on-the-
job training schemes to make AI more accessible to workers and workers’ representatives. In the same
vein, UNI Europa ICTS (2019[16]) emphasizes the importance for employers to invest in human capital,
while highlighting the role of soft skills among the new competencies that AI transition will require 9.

6
For instance, in Sweden, the JSCs, are partly based on a skills barometer which is run twice a year and which allows
the JSCs to anticipate new skills needs. In Germany, a 2016 agreement in the metal engineering and technology
sector named “Training and qualification for Industry 4.0 – managing change successfully’’ committed to analysing all
vocational and lifelong training programmes offered by the industry to assess their adequacy to the growing use of
data exchange and automation in manufacturing.
7
In France, the existence of national committees for employment and professional training (commissions paritaires
nationales emploi et formation professionnelle) is a source of collective agreements, in particular in the banking sector,
where the agreement of the 5th February 2020 on professional training maintains employees’ skills in the face of
economic, technological and organisational changes through anticipation, support and adaptation to changes, in
particular artificial intelligence.
8
In Denmark, for instance, a national tripartite agreement was signed in 2017 that specifically focused on adult and
continuing training. It included a series of initiatives over four years to increase and improve the access to, and the
quality of, adult learning.
9
According to the organisation, employers focus too much on the promotion of STEM skills, without acknowledging
the importance of soft skills, including creativity, empathy and complex reasoning. This would be partly due to the fact
6|
12. Some recent actions have also developed at European and national levels: UNI European Finance
(2021[17]) proposed to introduce responsible AI in the insurance sector through the use of a “people plan”
to identify skills needs, career paths and training possibilities. In June 2020, ETUC, BusinessEurope,
CEEP and UEAPME signed the European Social Partners Agreement on Digitalisation to identify the
necessary digital skills and adjust training accordingly, at national sectoral and firm levels. In Spain and
Sweden, unions have engaged in educating and training their own members: the Spanish union UGT
agreed with Google to train two hundred unions representatives on digital skills so they themselves go on
to train union members, while the Swedish Trade Confederation, TCO, provided in 2019 a free online
course to workers and unions representatives on AI (TCO, 2019[18]).

Social dialogue and collective bargaining can help define fair, ethical and pragmatic
responses to AI introduction at the workplace

13. The introduction of AI technologies at the workplace may entail different types of risks on the
quality of the working environment, but also raise fundamental ethical issues, such as excessive
surveillance, breach of privacy, potential discrimination among workers and risk of de-humanisation 10 due
to the emergence of AI-based management in the workplace providing potential unprecedented form of
control at the workplace (Adams-Prassl, 2019[19]).
14. On the one hand, the use of these new management models can help employers increase control
over their workers and the workplace, advance their performance evaluation techniques including novel
rating systems, improve the performance and productivity of workers, rationalise the organisation of work,
reduce the cost of monitoring and surveillance, facilitate the profiling of workers, affect workplace
behaviours, ensure discipline and improve HR processes: when responsibly developed, AI has the
potential to reduce human bias in decision-making relating to gender and various forms of discrimination
at work.
15. The European Agency of Safety and Health at Work suggests for instance that AI technologies,
by reducing human bias in decision-making, can improve OSH surveillance, reduce exposure to various
risk factors, including harassment and violence, and provide early warnings of stress, health problems and
fatigue. AI-based monitoring could also support evidence-based prevention, advanced workplace risk
assessment and more efficient, risk-based, targeted OSH inspections. Information could be used by
organisations to identify OSH issues, including psychosocial risks, and where OSH interventions are
required at organisational level (EU-OSHA, 2021[20]).
16. On the other hand, new management models based on AI may go beyond what is strictly
necessary, and lead to unfair and intrusive practices. (Bodie et al., 2016[21]) The OECD (Forthcoming[22])
identifies several issues of concern related to the use of AI and data for surveillance and monitoring 11.
First, workers are not always aware that they are being monitored or managed via AI. Second, even when
employees are aware that they are being monitored by algorithms, they do not always have access to nor
any say over who uses their data and how, which can lead to decisions with significant consequences for
them. Third, continuous monitoring of workers through AI-processed data may lead to privacy breaches
and violation of human integrity or dignity. Moreover, AI development typically relies on data which may be
biased in ways that are socially significant (Cowls et al., 2019[23]; Schor, 2018[1]). That may lead to an

that many AI-driven analyses aim at identifying skills gaps using open data sets such as LinkedIn profiles, which do
not necessarily refer to soft skills.
10The TUC (2020[45]) conducted a survey data that suggest high rates of distrust when it comes to discrimination and
unfairness by workers in the UK: more than 60% of respondents indicated that if not properly regulated, AI and
monitoring could increase both unfairness and discrimination.
11 Data privacy and protection issues range through all AI related technologies, including cobots, individual virtual

coaches, smart applications and platforms that support decision-making in sensitive areas of HR management, such
as hiring and performance management processes (OECD, 2021).
 |7
amplification of already existing forms of discrimination 12 and exclusion or the creation of new ones at the
workplace.
17. Beyond ethical issues, the use of AI technologies, can have both positive and negative impacts on
job quality: through a better work organisation and task optimisation, AI can reduce stress, and fatigue
levels, and enhance the quality of the working environment. For example, AI can support or automate
repetitive or physically and mentally strenuous tasks, thereby allowing workers to focus on more interesting
and safe tasks. However, AI can also generate psychosocial risks due to excessive monitoring and ethical
issue outlined before; moreover, when embedded in work equipment, AI can lead to an intensification of
the workload, higher dependency of workers in their interaction with machines, as well as increased safety
risks, all key drivers of the quality of the working environment (OECD, 2015[7]). Finally, concerns about
transparency 13, explainability and accountability in the case of injury or damage (Moore, 2019[24]) 14
may exacerbate these risks in the workplace (OECD, Forthcoming[22]).
18. Social dialogue and collective bargaining have a fundamental role to play in all the aforementioned
issues. First, because evidence shows that social dialogue and collective bargaining can help companies
define tailor-made and fair solutions to organisational and technological changes (OECD, 2019[5]). ETUI
(2021[25]) highlights the importance of the flexibility that collective agreements offer to cope with the sectoral
and company-specific application of AI technologies, by offering tailored-made solutions and accounting
the interests of workers and employers, and applying the general principles laid down in legislation in
specific contexts.
19. Secondly, there is evidence that social dialogue and collective bargaining can enhance the quality
of the working environment 15 (OECD, 2019[5]). Along the same lines, a recent study also found that the
existence of employee representation tends to lead to job designs which, under automation, provide better
working conditions and reduce worker scepticism towards automation 16 (Belloc, Burdin and Landini,
2020[28]). Third, because social dialogue and collective bargaining may be even more necessary to
rebalance excessive imbalances stemming from AI diffusion 17 and ensure that the fundamental labour and
human rights are respected, as part of the international initiative to ensure a human-centric governance.
20. Many authors suggest that social dialogue is essential in further examining the effect of AI in
minority groups and re-examining health and safety to draft better regulation (Lillywhite and Wolbring,
2020[29]). The OECD (2019[28]), recognizing the importance of this topic, calls “for all actors involved at all
stages of AI applications to address the risks related to safety bias and discrimination”. Along these lines,
the OECD AI principles (OECD, 2018[6]) outlines the instrumental role of social dialogue in ensuring a fair
AI transition (Box 1).

12 AccessNow 2018 provides evidence that workplace discrimination can be facilitated by AI.
13
Workers need to be informed on their privacy rights, for instance how data are used, stored or shared outside the
employment relationship.
14
For instance, if AI technologies are used to implement practices that increase pressure on workers (e.g.
micromanagement), they may cause stress and anxiety (Moore, 2019[24]).
15
The OECD (2019[5]) found that the quality of the working environment was on average highest in countries with well-
coordinated social partners and a large coverage of collective agreements. It is also significantly higher in firms where
workers can voice their concerns through representative institutions and through channels of direct dialogue with
management.
16
The survey was conducted over a sample of more than 20 000 establishments from 28 countries. Additionally the
study notes that employee representation and automatisation through AI are positively associated.
17The use of AI may generate substantial power unbalances, for instance surveillance on platform workers can
generate information asymmetries in favour of employers who have unlimited access to data on employees. Such
asymmetries fundamentally change the power balance between workers and employers (Rani and Singh, 2019[55]).
8|

Box 1. OECD AI principles

OECD AI principles promote an AI that is innovative and trustworthy and that respects human
rights and democratic values. The OECD AI principles call on governments to build human
capacity and prepare the labour markets for labour transformation, by:
1. Empowering people to effectively use and interact with AI systems, including equipping them
with the necessary skills;
2. Ensuring a fair transition for workers as AI is deployed, including via social dialogue, training
programmes
3. Promoting the responsible use of AI at work, to enhance the safety of workers and the quality
of jobs, to foster entrepreneurship and productivity, an aim to ensure that the benefits are
broadly and fairly shared

21. Several papers also acknowledge the importance of collective bargaining to prevent intrusive
business practices and risk of de-humanisation: De Stefano (2020a[31]; 2020b[32]) argues that collective
agreements could lay down the specific limits of AI-enabled surveillance of workers. The author argues
that trades unions and workers’ representatives should organise and oppose undue surveillance, for
example by banning the most intrusive applications of technology including neuro-surveillance, and that
collective bargaining is arguably still the most effective tool to achieve those goals in a rapid and
customized fashion, considering how fast new technologies and are developed and introduced in the world
of work today. Hendrickx (2019[33]) argues in favour of collective action in response to the rise of AI and
surveillance. Hendrickx calls for fully coordinating human rights instruments, such as the European
Convention of Human Rights, data protection instruments such as the EU GDPR and labour regulation
instruments, including collective bargaining, to ensure that the use of new technologies at work is made
compatible with human rights.

Social partners are engaging in outreach, awareness raising and advocacy

22. International and national unions have expressed their concerns and raised voice about ethical
issues, through position papers and guidelines about the application of AI in the workplace. The ETUC has
called for the reinforcement of workers’ protections from undue surveillance, as well as from biased
discrimination at the workplace in its resolution on the European Strategies on AI and data (ETUC,
2020[15]). In its foresight brief, ETUI emphasizes the need for a preventive engagement of workers and
trade unions in the way algorithms are designed and deployed, and call for collective bargaining to ensure
the interest of workers and protect fundamental rights (ETUI, 2021[25]). They can also offer the required
flexibility to cope with the sector- and company-specific application of technologies. Consequently, it is vital
that trade unions are aware of the risks of algorithmic management and that they plan adequate responses
to these risks.
23. UNI Global Union produced principles 18 for ethical AI and workers’ data privacy and protection to
be implemented within collective agreements, at various levels or global framework agreements (UNI
Global Union, 2019a[34]; UNI Global Union, 2019b[35]). AFL-CIO (2019[36]) expressed their concerns as
algorithms and AI tools to make decisions about hiring and firing, promotions and work organisation without
the consent of workers. They propose EU GDPR style of legislative introductions and tailored collective
bargaining agreements “since optimal rules for data collection and use may vary considerably among
workplaces”. The European social partners in the insurance sector (UNI Europa Finance; Insurance

18
Principles related to workers’ surveillance privacy and human dignity include: the right to have access to, and
influence over, data collected on them; the exemption of biometric data and personally identifiable information; the
avoidance of location tracking equipment unless necessary, the application of data minimalisation principle and more
generally the respect of privacy laws and fundamental rights throughout the company, etc.
 |9
Europe; Amice; Bipar, 2021[17]) also outlined the importance of assessing the risk of bias stemming from
AI use and mitigate it.
24. Some unions (ETUI, 2021[25]; TUC, 2021[37]) 19 have also started reflecting and calling for the
introduction of new rights, such as the right not to be subjected to fully automated decisions (e.g. without
human intervention), the right to explanation for decisions made by algorithms or machine learning models,
as automated decisions may result in incorrect performance assessment, biased allocation of tasks, etc.
The British Trade Union Congress (TUC) produced an AI Manifesto proposing the introduction into
legislature of new rights, such as the right to data reciprocity giving workers the right to collect and combine
workplace data, the right to human review of high-risk decisions, the right of human contact when important
decisions are made about people at work (TUC, 2021[37]).
25. In the United States, the Teamsters Union strongly positioned in favour of bargaining over the
permission given to employers to monitor the workplace using cameras and define more generally the
parameters of workplace surveillance (Teamster, 2018[38]). In Germany, (DGB, 2020[39]) called for a more
active participation of trade unions in the social dialogue around the topics of workplace surveillance and
privacy in the era of AI.
26. For employers’ organisations, AI is generally perceived as an opportunity for business and
economic growth in ensuring competitive advantage (BusinessEurope, 2018[40]). Hence, employers'
organisations have engaged in strategies to address challenges such as barriers for AI diffusion, including
the needs of up-skilling and re-skilling, data sharing practices and cybersecurity, as well as funding issues.
In its AI strategy, BusinessEurope (2020a[41]) proposed for instance the creation of common European
data spaces for business-to-business data access and sharing, cloud interconnectedness of data spaces
build on trust, openness, security, interoperability and portability or no mandatory data access rights unless
other options are explored first. Still, in their common report, ILO ACT EMP and IOE also outlines the role
of employers and business representatives in changing business paradigms and emerging new models
(ILO ACT EMP and IOE, 2019[13]). Finally, in a survey on the introduction of digital technologies addressed
to European managers (CEC European Managers, 2018[42]), half the managers expressed some concern
about the implications for privacy of digital technologies, followed by transparency and labour rights. 20
Those were yet primarily expressed as possibly disruptive to the business processes.

Social partners are developing action plans and signing innovative agreements

27. Beyond informing, alerting and participating in decision making, social partners can also provide
guidance through framework agreements, 21 as well as negotiate collective agreements ensuring adequate
safeguards when AI-enabled tools and algorithmic-management practices are implemented. The
European Social Partners Framework Agreement on Digitalisation (2020[43]), signed for instance by ETUC,
BusinessEurope, CEEP, and SMEunited, provides guidance on most of the issues outlined above and
calls for the respect of human dignity, while enabling workers’ representatives to address issues related to
data, consent, privacy protection and surveillance, and the need to systematically link the collection and
storage of data to ensure transparency – using the EU GDPR as a reference. 22 The framework also calls
for a fair deployment of AI systems, i.e. ensuring that workers and groups are free from unfair bias and
discrimination.

19
Those initiatives build on the EU GDPR.
20 The Activity Report 2018 - 2021 covers CEC European Managers’ activities from May 2018 to May 2021
21 European framework agreements are a result of EU-level social dialogue. Framework agreements are just one of
several possible outcomes of EU social dialogue. The term ‘framework’ is intended to highlight the particular nature of
the agreement as providing an outline of general principles to be implemented in the Member States ‘either in
accordance with the procedures and practices specific to management and labour and the Member States or at the
joint request of the signatory parties, by a Council decision on a proposal from the Commission’ (Article 139(2) EC).
22Notably to the article 88 of the EU GDPR which refers to the possibilities to lay down by means of collective
agreements, more specific rules to ensure the protection of the rights and freedom with regards to the processing of
personal data of employees in the context of employment relationships.
10 |
28. More recently, social partners have also started engaging in “algorithm negotiations”, i.e. they are
including as a subject of bargaining the use of AI, big data and electronic performance monitoring (“people
analytics”) in the workplace, as well as their implications for occupational health and safety, privacy,
evaluation of work performance and hiring and firing decisions (De Stefano, 2018[44]).
29. To date, a few collective agreements have already been signed in OECD countries related to AI.
In Spain, UGT and the employer organization Ametic signed an agreement asking for compliance with the
EU GDPR with regards to data collection and processing. The agreement also stipulates that new
technologies should be used for the benefits of the workers as well. Another agreement was signed
between the Spanish Government, CCOO, UGT, CEOE and CEPYME that guarantees platform workers’
rights to algorithmic transparency. In Switzerland, the trade union Syndicom agreed with the company
Swisscom to a “Smart Data” policy that includes principles when processing employees’ data.

How is AI affecting social dialogue?

30. If social partners can and should contribute to a fair AI transition, they also have to face additional
pressures brought by AI, stemming from the complexity and lack of transparency of AI technologies. In
addition, they have to develop strategies enhancing their affiliates’ trust towards AI. More generally, to
address fundamental ethical questions linked to human dignity, a more general policy framework besides
social dialogue and collective bargaining may also be required. At the same time, AI technologies may
also bring opportunities to social partners, for instance in helping strengthening workers’ organisation or
voice.
31. Particular attributes of AI technologies, such as their complexity and opacity, make it difficult for
users and social partners to understand the social implications of their implementation or to make claims,
and the allocation of liability may be unfair or inefficient. For instance, having a fair algorithm and respecting
labour standards may require social partners to be involved in the design of the algorithm. However, having
the mathematical code is often not sufficient to understand the purpose behind the algorithm and identify
who are the targeted individuals, what are the trade-offs made in the input of values and variables, such
as race, gender or other characteristics. In order to be able to critically understand AI’s role and its impact
on their work, social partners need to become “AI literate” (ETUI, 2021[25]).
32. According to a survey conducted by the British Trade Union Congress (TUC), 36% of the survey
participants felt that neither they nor their representatives could effectively challenge what they considered
unfair decisions, due to lack of knowledge about the algorithmic operations and use, problems accessing
data and management claiming infallibility of algorithms (TUC, 2020[45]).
33. The lack of transparency and explainability, while undermining trust in AI technologies 23, may also
damage mutual trust between social partners, one of the key conditions for successful dialogue and
collective bargaining (OECD, 2018[6]). This concern has been recently expressed by social partners in
various papers − see for example, UNI Europa Finance; and BusinessEurope (2020b[46]), calling for an
approach to “excellence and trust”.
34. Furthermore, social partners will also need to continue their efforts in maintaining and/or improving
their representativeness, as ensuring a large coverage is another key factor for a well-functioning social
dialogue in a changing world of work (OECD, 2019[5]). According to a recent survey conducted by ETUC,
most collective agreements around digitalisation have been by far concluded in large companies. Far fewer
have been concluded on a sectoral level and most often in sectors strongly affected by new technologies.
(Voss and Riede, 2018[47]). While collective agreements on AI technologies specifically are still in an early
stage across OECD countries, the sectoral patterns are likely to be similar.
35. Along these lines, unions are diversifying their strategies to reach potential members and
represent non-standard forms of employment (OECD, 2019[5]). New minority independent unionism is on
the rise, especially in AI and digitalisation related sectors, to bring bargaining counterparts to the table,
even though they usually lack formal collective bargaining participation (Johnston and Land-Kazlauskas,

23
According to the same TUC survey, 56% of the respondents found that AI introduction damaged trust between
workers and employers (TUC, 2020).
 | 11
2018[48]). Employers also highlight the need for businesses to reach out to underrepresented or emerging
economic actors and in particular giving “a seat at the table” to small and medium enterprises, as well as
new business models based on AI technologies (ILO ACT EMP and IOE, 2019[13]).
36. Finally, since AI systems may take decisions that violate fundamental human rights – whether or
not this is known or intended by the developers and applying business, social partners’ actions may not be
sufficient and require adjustments in national and international legislation governing AI as well as effective
enforcement by public authorities. Such an issue should be further investigated in forthcoming research.
Codes of conduct, soft law and other self-regulatory instruments are not legally binding. Evidence shows
that their impact is rather limited and companies using them have a limited ability to evaluate their success
or implementation (Auplat, 2012[49]; Jenkins, 2001[50]). Furthermore, the lack of enforcement mechanisms,
as well as explicit sanctions for non-compliance is a real concern (ETUI, 2021[25]).
37. The use of new technologies, including AI-enabled technologies, also provide an opportunity for
workers’ representatives to increase workers’ voices and organise (Adler-Bell and Miller, 2018[51]).
Following the onset of the internet, which has had a tremendous impact in terms of facilitating
communication and presenting renewal opportunities for unions (Martinez and Walker, 2005[52]) AI can
further boost trade unions through increased outreach, especially to younger members, new forms of
management, facilitation of membership renewals, but also data analysis and machine learning techniques
aiming to inform the union strategy about concerns related to the world of work (Vandaele, 2018[53]).
Flanagan and Walker (2020[54]) provide an illustration of an application of AI used to build unions’ power in
Australia and the United States: using a reconfigured chatbot to reflect an “organising” rather than a
“servicing” ethos, they argue that chatbots can offer various possibilities to unions trying to enhance their
resources or capabilities. 24
38. Ranking applications, initially developed by platforms for their customers and employed to put
pressure on workers of those platforms through monitoring and discipline, are also re-appropriated by
workers to rank employers and their working conditions (AFL-CIO, 2019[36]; Johnston and Land-
Kazlauskas, 2018[48]). Unions representatives and workers, are developing new initiatives, as they become
aware of the potential of digital technologies to help them organise and reach out more workers from the
digital economy 25 (Voss and Riede, 2018[47]).

Conclusions

39. This brief argues that social dialogue and collective bargaining can play an important role in
addressing some of the key challenges driven by AI technologies. Previous evidence has shown that when
social partners work cooperatively, social dialogue and collective bargaining can support and usefully
complement public policies in easing transitions in the labour markets. Moreover, collective bargaining
systems, when coordinated, can also reduce inequalities and foster inclusive and dynamic labour markets.
40. However, the fundamental ethical issues that the use of AI poses in terms of workers’ privacy, data
protection, surveillance and discrimination, may call for international and national actions to ensure the
right ethical and legal framework is in place. Furthermore, the complexity and lack of transparency of AI
algorithms represents a serious challenge for social partners who may be unable to understand, explain
and adequately defend their members’ interests. Along these lines addressing transparency, explainability
and accountability will be key to build trust around AI adoption and diffusion.

24
The reconfigured chatbot was used to enable otherwise marginal workers to receive basic information in a manner
that reinforced union narratives of power and worker solidarity, and workplaces to be mapped more efficiently. The
chatbot did not act as a labour-saving tool, but stimulated wide-ranging learning by bringing implicit tensions between
'servicing' and 'organising' conceptions of knowledge, power and expertise to the surface.
25For instance, new applications, such as Alia for domestic workers, or Weclock, have been developed to empower
workers and unions. New online tools, such as the faircrowd.work website portal set up by IG Metall, the Austrian
Union Confederation and the Swedish Unionen, provide support and information to platform workers on their working
conditions. Along the same lines, the Lighthouse in the United Kingdom, provides an online tool for unions to educate
themselves and find out where they stand in matters of data governance on their projects and as a whole.
12 |

References

Adams-Prassl, J. (2019), What if Your Boss Was an Algorithm? The Rise of Artificial Intelligence [20]

at Work, (SSRN Scholarly Paper No. ID 3661151) Social Science Research Network,
Rochester, NY.

Adler-Bell, S. and M. Miller (2018), The Datafication of Employment, How Surveillance and [50]

Capitalism Are Shaping Workers’ Futures without Their Knowledge., The Century Foundation.

AFL-CIO (2019), Report of the AFL-CIO Commission on the Future of Work and Unions (Report to [35]
the AFL-CIO General Board).

Auplat, C. (2012), Nanotechnology and sustainable development, Routledge. [48]

Belloc, F., G. Burdin and F. Landini (2020), Robots and Worker Voice: An Empirical Exploration, [27]
IZA, Discussion Papers 38.

Bodie, M. et al. (2016), The Law and Policy of People Analytics, Fac. Scholarsh. [22]

Boheim, R. and A. Booth (2004), Trade Union Presence and Employer Provided Training in Great [9]
Britain, Ind. Relat. J. Econ.Soc.43, 520-545.

Brynjolfsson, E., D. Rock and C. Syverson (2017), Artificial intelligence and the Modern [3]
Productivity Paradox: A Clash of Expectations and Statistics, NBER Working Paper Series,
No.24001, NBER.

BusinessEurope (2019), Artificial Intelligence Reccomendations. [15]

BusinessEurope (2018), Robotics and Automation. [39]

BusinessEurope (2020a), A European Strategy for Data. [40]

BusinessEurope (2020b), AI: A European Approach to Excellence and Trust (Position Paper). [45]

BusinessEurope, SMEunited, CEEP and the ETUC (2020), European Social Partners Framework [42]
Agreement on Digitalisation, https://www.etuc.org/system/files/document/file2020-
06/Final%2022%2006%2020_Agreement%20on%20Digitalisation%202020.pdf.

CEC European Managers (2018), CEC European Managers Survey, https://www.cec- [41]
managers.org/wp-content/uploads/2021/06/CECActivity-Report-2018-2021.pdf.

Cowls, J. et al. (2019), Designing AI for Social Good: Seven Essential Factors, (SSRN Scholarly [24]
Paper No. ID 3388669). Social Science Research Network,
https://doi.org/10.2139/ssrn.3388669.

De Stefano, V. (2019), Introduction: Automation, Artificial Intelligence, and Labour Protection, [4]

Social Science Research Network, Rochester, NY.

De Stefano, V. (2018), ‘Negotiating the Algorithm’: Automation, Artificial Intelligence and Labour [43]

Protection, (SSRN Scholarly Paper No. ID 3178233). Social Science Research Network,
https://doi.org/10.21 forma39/ssrn.3178233.

De Stefano, V. (2020b), ‘Masters and Servers’: Collective Labour Rights and Private Government [31]

in the Contemporary World of Work, Int. J. Comp. Labour Law Ind. Relat. 36.

De Stefano, V. (2020a), Algorithmic Bosses and What to Do About Them: Automation, Artificial [30]
Intelligence and Labour Protection, Springer International Publishing,
https://doi.org/10.1007/978-3-030-45340-4_7.

DGB (2020), Artificial Intelligence (AI) for Good Work. [38]

Dustmann, C. and U. Schönberg (2009), Training and Union Wages. Rev., Econ. Stat. 91, 363– [10]
 | 13

376., https://doi.org/10.1162/rest.91.2.363.

ETUC (2020), Resolution on the European strategies on artificial intelligence and data, ETUC [16]
Executive Committee.

ETUI (2021), Algorithmic management and collective bargaining (Foresight Brief). [26]

EU-OSHA (2021), Impact of Artificial Intelligence on Occupational Safety and Health. [21]

Flanagan, F. and M. Walker (2020), How can unions use Artificial Intelligence to build power? The [53]

use of AI chatbots for labour organising in the US and Australia, New Technol. Work Employ,
https://doi.org/10.1111/ntwe.12178.

Hendrickx, F. (2019), Privacy 4.0 at Work: Regulating Employment, Technology and Automation, [32]

Comp Lab Pol J 41, 147.

ILO (2020), Promoting employment and decent work in a changing landscape: Report of the [6]
Committee of Experts, ILO,
https://labordoc.ilo.org/discovery/delivery/41ILO_INST:41ILO_V2/1268239920002676?lang=en
&viewerServiceCode=AlmaViewer.

ILO ACT EMP and IOE (2019), Changing Business and Opportunities for Employer and Business [14]
Organizations, International Labour Organization and International Organisation of Employers.

Jenkins, R. (2001), Corporate codes of conduct. Self-regulation in a global economy, Technology, [49]
Business and Society Programme Paper 2, United Nations Research Institute for Social
Development.

Johnston, H. and C. Land-Kazlauskas (2018), Organizing on-demand representation, voice, and [47]
collective bargaining in the gig economy, International Labour Office,
https://www.ilo.org/wcmsp5/groups/public/---ed_protect/---protrav/---
travail/documents/publication/wcms_624286.pdf.

Klindt, M. (2017), Trade union renewal through local partnerships for skill formation, Transfer: [13]
European Review of Labour and Research, Vol. 23/4, pp. 441-455.

Lane, M. and A. Saint-Martin (2021), The impact of Artificial Intelligence on the labour market: [2]
What do we know so far? (No. 256), OECD Social, Employment and Migration Working
Papers, OECD Publishing.

Lecher, C. (2019), “How Amazon automatically tracks and fires warehouse workers for [55]
‘productivity’”.

Lillywhite, A. and G. Wolbring (2020), Coverage of Artificial Intelligence and Machine Learning [28]
within Academic Literature, Canadian Newspapers, and Twitter Tweets: The Case of Disabled
People, Societies 10, 23., https://doi.org/10.3390/soc10010023.

Martinez, M. and S. Walker (2005), The Networked Union? The Internet as a Challenge to Trade [51]
Union Identity and Roles, Crit. Perspect. Int. Bus. 1, 137–154.,
https://doi.org/10.1108/17422040510595645.

Moore, P. (2019), The Mirror for (Artificial) Intelligence: In Whose Reflection?, SSRN Electron. J., [25]
https://doi.org/10.2139/ssrn.3423704.

OECD (2019), Negotiating Our Way Up: Collective Bargaining in a Changing World of Work, [5]

OECD Publishing, Paris, https://dx.doi.org/10.1787/1fd2da34-en.

OECD (2019), OECD Employment Outlook 2019: The Future of Work, OECD Publishing, [12]
https://doi.org/10.1787/9ee00155-en.

OECD (2019), Scoping the OECD AI principles: Deliberations of the Expert Group on Artificial [29]
14 |

Intelligence at the OECD (AIGO) OECD Publishing, Paris,, OECD Publishing,

https://doi.org/10.1787/d62f618a-en.

OECD (2018), OECD Employment Outlook 2018, OECD Publishing,, [7]

https://doi.org/10.1787/empl_outlook-2018-en.

OECD (2015), Back to work: Sweden: Improving the Re-employment Prospects of Displaced [8]
Workers, OECD Publishing, https://doi.org/10.1787/9789264246812-en.

OECD (Forthcoming), AI at the workplace: the fundamental ethical issues, OECD Publishing. [23]

Rani, U. and P. Singh (2019), Digital Platforms, data, and development: Implications for Workers [54]
in Developing Economies, Autom. Artif. Intell. Labor Law 41, 23.

Rosenberg, A. (1984), “Automation and the work preservation doctrine: Accommodating [56]
productivity and job security interests”, UCLA L. Rev., 32,, p. p.135.

Schor, J. (2018), The Platform Economy: Consequences for labour, inequality and the [1]
environment. An Excerpt from Work in the Digital Age: Challenges of the Fourth Industrial
Revolution., Rowman & Littlefield spencer.

Solomon, L. and N. Hyman (1989), “Improving the Competitive Position of US Manufacturing [57]
Firms: A Legal and Policy Analysis”, JL & Com, Vol. 9, p. p.21.

TCO (2019), TCO’s board accepts AI challenge, https://www.tco.se/nyheter-och- [19]

debatt/Pressmeddelanden/2019/tcos-styrelse-antar-ai-utmaning/?epslanguage=sv (accessed
on 22 April 2021).

Teamster, D. (2018), The Future of Work is Now bargaining strategies: From Machine Breaking to [37]
Artificial Intelligence 84.

TUC (2021), Dignity at Work and the AI Revolution. A TUC Manifesto, Trade Union Congress. [36]

TUC (2020), Technology managing people - The worker experience, Trade Union Congress, [44]
https://www.tuc.org.uk/research-analysis/reports/technology-managing-people-worker-
experience (accessed on 20 January 2021).

UNI Europa Finance; Insurance Europe; Amice; Bipar (2021), Joint declaration on Artificial [18]
Intelligence by the European social partners in the insurance sector, https://www.uni-
europa.org/wp-content/uploads/2021/03/20210316_ISSDC_Joint-declaration-on-Artificial-
Intelligence-1.pdf (accessed on 13 April 2021).

UNI EUROPA ICTS (2019), Position on Artificial Intelligence. [17]

UNI Global Union (2019b), Top 10 Principles for Ethical Artificial Intelligence, UNI Global Union. [34]

UNI Global Union (2019a), Top 10 Principles for Workers’ Data Privacy and Protection, The [33]
Future World of Work, http://www.thefutureworldofwork.org/docs/10-principles-for-workers-
data-rights-and-privacy/.

Vandaele, K. (2018), Will trade unions survive in the platform economy? Emerging patterns of [52]
platform workers’ collective voice and representation in Europe, ETUI Working Paper 35.

Verma (2005), What do unions do to the workplace? Union effects on management and HRM [11]
policies, J. Labor Res.

Voss, E. and H. Riede (2018), Digitalisation and workers participation: What trade unions, [46]
company level workers and online platform workers in Europe think,
https://www.etuc.org/sites/default/files/publication/file/2018-09/Voss%20Report%2.
THE GLOBAL DEAL FOR DECENT WORK AND INCLUSIVE GROWTH

The Global Deal is a multi-stakeholder partnership that aims to address

the challenges in the global labour market to enable all people to benefit
from globalisation. It highlights the potential of sound industrial relations and
enhanced social dialogue to foster decent work and quality jobs, to increase
productivity, and to promote equality and inclusive growth. The Global
Deal welcomes governments, businesses, employers’ organisations,
trade unions, as well as civil society and other organisations to join the
partnership.

[email protected] www.theglobaldeal.com @theglobaldeal