SUMMARY
This resource has been developed to support trade union organisers dealing with the introduction of artificial intelligence (AI) in the workplace. It outlines existing laws and organising tools relevant to AI and how collective bargaining can be used to ensure workers benefit from and are not harmed by AI.
The page has five interconnected sections:
- Principles for a worker-centred approach to AI. This section outlines core union principles on AI.
- Engaging employers on AI. This section outlines recommended steps unions can take when engaging with employers on AI.
- Impact assessment tool. This section provides a quick- fire impact assessment tool that will inform you what impact an AI tool might have on workers, so you can tell how much work you might need to do to engage with an employer about it.
- Health and safety. This section outlines some of the existing health and safety practices that can be used to deal with the health and safety risks of AI.
- Recommended collective agreement clauses. This section provides a set of model clauses that may be useful in negotiating collective agreements.
THE ISSUE
The term artificial intelligence (AI) describes technology that performs tasks requiring human-level intelligence. Current AI technology has a wide range of uses and employers are using it more and more in workplaces.
Common forms of AI used in the workplace include things like ChatGPT to help with writing, AI-assisted surveillance systems, customer service chatbots, recruitment algorithms, data analysis tools, and automated management systems.
There is a lot of debate about the impacts AI may have on employment. Some people argue AI could mean more good jobs and higher wages. Others fear it will cause unemployment and increase employer surveillance (monitoring) of workers.
The evidence so far is mixed. AI does not appear to be causing many job losses yet, and many people who work with AI say it has improved their job satisfaction. However, some uses of AI, such as AI surveillance systems, productivity monitoring, and automated management are very harmful to workers.
In the end, the impact AI has on workers will depend on the collective decisions we make about its development, use, and regulation. Trade unions therefore have an important role to play in making sure workers benefit from the use of AI in the workplace and are not harmed by it.
The New Zealand Council of Trade Unions Te Kauae Kaimahi has produced this resource to help New Zealand trade unions in their mahi to put workers’ interests first in this rapidly changing world of work.
The main approach we recommend is twofold: (1) use existing laws and organising tools to protect workers from harmful uses of AI; and (2) use collective action to introduce new tools that ensure workers benefit from and are not harmed by AI.
PRINCIPLES FOR A WORKER-CENTRED APPROACH TO AI
The NZCTU has developed the following principles to guide its approach to AI in the workplace. The objective is to maximise the benefits of AI for workers and prevent negative impacts.
These principles can be used to guide unions in their mahi to protect workers’ rights, support worker voice, and ensure workers benefit from AI.
Te Tiriti o Waitangi must be upheld
AI technologies must support and not undermine tino rangatiratanga, mana motuhake, and equity outcomes for Māori. This means employers need to engage Māori workers on the use of AI in the workplace, and laws need to reflect the interests and concerns of Māori.
Worker voice is essential
Workers must be involved in decision-making on the introduction, use, and regulation of AI technology. High-risk AI (AI that might impact a worker’s or job seeker’s legal rights and responsibilities, or have another significant effect) should only be used if workers and their unions have agreed.
AI should be used to improve job quality
AI should be used to make work better for workers, by cutting down on boring tasks and lifting skills. Workers must have education and training so that they can productively and safely use AI. Employers, unions, and government have a joint responsibility in delivering this education and training.
Workers’ fundamental rights must be protected
Some uses of AI, like automated surveillance (monitoring), can be harmful to workers or create unacceptable risks. AI that has negative impacts on workers and communities must not be used.
A just transition is essential
Workers must be supported through any AI-related change, especially if employers cut the number of jobs. Where the introduction of AI will have a significant impact on the work itself, proper management of change processes must be followed.
Humans in command
Humans must make all decisions that affect work or workers. Employers must not give these decisions to AI. Workers must have the right to get a clear explanation of how decisions about their work have been made and be able to challenge those decisions.
Workers’ data sovereignty must be upheld
Workers are the owners of data (information) about themselves, and regulations and employment practices must support this. Workers must be able to access their data in an understandable format, correct this data, and use it to support their own interests.
The right to disconnect must be embedded
Workers must keep the right to have clear work/home boundaries, which means having the ability to log off and be unavailable.
ENGAGING EMPLOYERS ON AI
If an employer has introduced or is planning to introduce AI in a workplace your union covers, we recommend considering the following steps.
Consider how workers can be engaged
Worker engagement, participation, and representation is key to ensuring safe and productive uses of AI. We recommend that unions make sure employers engage workers at all levels of decision-making on AI, particularly if the tool might be high-risk.
If possible, we recommend establishing a forum, or using existing channels (such as a health and safety committee), where the employer and union can regularly engage on the issue of AI. This forum could be used to develop a strategic approach to the introduction and use of AI. The approach should consider issues like the prevention of AI-related job loss, training and education for workers, health and safety implications, and data and privacy processes.
Gain an understanding of the technology
There are many ways AI can be applied in the workplace. If an employer is considering introducing AI in a workplace your union covers, it is important to understand how that technology works and what its strengths and weaknesses are.
Ask the employer to provide information to workers and the union about what the technology is, how it works, and what it is for. We also recommend conducting an assessment to determine what the risks of introducing the technology may be (see page 9). This will help your union work out if it is necessary to engage on this issue. Unions may need to invest in appropriate training so that union staff understand how these technologies work.
Work out if there is a good reason to introduce the technology
Employers may have good reasons for using AI technology in the workplace, such as to improve the speed or accuracy of certain tasks. However, other reasons may be harmful and should be challenged, such as using AI to increase unfair surveillance of workers. Unions also need to watch out for “scope creep”, where AI introduced for a good reason, such as improving health and safety practices, is then used for a harmful reason, such as increasing unfair surveillance of workers.
If an employer is considering introducing AI in a workplace your union covers, consider if the proposed technology has a good reason. Pay particular attention if the technology has a surveillance component.
Identify relevant clauses in existing collective agreements
Your union may already have in place collective agreement clauses that establish processes for the introduction of new technology or changes to ways of working. The issue may also be covered in existing industry frameworks or workplace policies. Look at these provisions to work out how these interact with the new technology and if there are any gaps that need to be addressed.
Understand the legal rights and responsibilities of employers and workers
New Zealand law already covers multiple issues raised by AI in the workplace. In addition to the Employment Relations Act 2000 and the Health and Safety at Work Act 2015, you may need to consult the Privacy Act 2020, the Human Rights Act 1993, the Bill of Rights Act 1990, and the Copyright Act 1994, as well as any other acts relevant to the workplace your union covers.
Develop an impact assessment
If AI could have significant impacts on workers and the workplace, we recommend that you develop an impact assessment. Ideally, you should do this with the employer before the AI is introduced. The assessment should consider the reason for introducing the technology, the benefits
and risks of the technology, whether and how risks can be safely managed, and possible cultural impacts, among other things. If there is an impact on worker health and safety,
the employer has a legal duty to engage with workers on its introduction.
Consider opportunities for collective action
Because it can have a big impact on workers’ jobs, AI can be a widely and deeply felt issue. This makes it a great organising opportunity. We recommend engaging with workers to help raise awareness of the opportunities and risks of introducing AI, how they can be prepared, and the possible role of collective bargaining.
IMPACT ASSESSMENT TOOL
If an employer is thinking about introducing AI in a workplace your union covers, we recommend you look at and score the following risk factors. A higher score on any of these factors will usually mean you should engage the employer more actively.
What is the risk that the AI being introduced will:
- Cause workers to lose their jobs?
- Substantially change the nature of workers’ jobs?
- Make the quality of the work worse?
- Have an impact on management decisions like hiring or performance management?
- Introduce risks to health and safety, including psychosocial health and safety?
- Increase monitoring and surveillance of workers?
- Damage workers’ ability to organise or be part of their union?
- Put workers’ privacy at risk?
- Increase bias and discrimination in the workplace?
- Reduce human contact in the job?
- Blur boundaries between work and home life?
- Reduce workers’ ability to develop and maintain skills?
- Any other potential risk factors?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☐ | ☐ |
EXAMPLE 1: AI CHATBOT
The employer at a telecommunications company intends to introduce a customer service chatbot to handle routine customer questions.
What is the risk that the technology being introduced will:
Cause workers to lose their jobs?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☐ | ☒ |
The chatbot could reduce the number of customer service workers required by the business.
Substantially change the nature of workers’ jobs?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☐ | ☒ |
The chatbot will likely change the types of tasks performed by customer service workers.
Make the quality of work worse?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
It is possible the chatbot will improve job satisfaction among customer service workers by eliminating boring tasks and enabling them to engage in more interesting ones. But it is also possible the chatbot will provide lower-quality service to customers.
Have an impact on management decisions like hiring or performance management?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Introduce risks to health and safety, including psychosocial health and safety?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Increase monitoring and surveillance of workers?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Damage workers’ ability to organise or be part of their union?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Put workers’ privacy at risk?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Increase bias and discrimination in the workplace?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Reduce human contact in the job?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
It is possible the chatbot will reduce contact between customer service workers and customers.
Blur boundaries between work and home life?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Reduce workers’ ability to develop and maintain skills?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
There is a moderate risk the chatbot will reduce workers’ ability to develop and maintain skills if it replaces customer service workers and these workers are not shifted onto other jobs.
Summary
The assessment indicates the chatbot would introduce several major risks for workers. Most significantly, it has the potential to cause job loss and to substantially change the nature of customer service workers’ jobs. The potential introduction of the chatbot therefore requires a high level of engagement between the union and employer.
EXAMPLE 2: AI-ASSISTED VIDEO MONITORING
The employer at a logistics company intends to introduce AI-assisted video monitoring in dispatch warehouses to identify health and safety hazards.
What is the risk that the technology being introduced will:
Cause workers to lose their jobs?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
There is no direct risk to employment posed by this technology, but there is a risk it could be misused by the employer for disciplinary purposes.
Substantially change the nature of workers’ jobs?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
There is some risk that workers would change how they perform tasks because of the presence of video monitoring.
Make the quality of work worse?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☒ | ☐ | ☐ |
There is some risk that surveillance will increase work stress and anxiety, which would negatively affect job quality.
Have an impact on management decisions like hiring or performance management?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☒ | ☐ |
There is a high risk the video monitoring could be misused by the employer for disciplinary purposes.
Introduce risks to health and safety, including psychosocial health and safety?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☒ | ☐ |
There is a critical risk the video monitoring could lead to unfair surveillance of workers, which is known to be a psychosocial risk.
Increase monitoring and surveillance of workers?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☒ | ☐ |
There is a critical risk that the employer will use the video monitoring to increase unfair surveillance of workers – for example, monitoring worker productivity.
Damage workers’ ability to organise or be part of their union?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☒ | ☐ |
If the video monitoring is used to increase unfair surveillance this could have a chilling effect on worker organisation.
Put workers’ privacy at risk?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☐ | ☒ |
Any surveillance of workers creates privacy risks.
Increase bias and discrimination in the workplace?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☐ | ☐ | ☐ | ☒ | ☐ |
The video monitoring could be misused by the employer to make biased or discriminatory decisions. However, the tool itself does not present an inherent risk.
Reduce human contact in the job?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Blur boundaries between work and home life?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Reduce workers’ ability to develop and maintain skills?
| No risk | Minor risk | Moderate risk | Serious risk | Critical risk |
| ☒ | ☐ | ☐ | ☐ | ☐ |
Summary
The assessment indicates the AI-assisted video monitoring would introduce a range of significant risks for workers. Unlike the chatbot example, video monitoring does not create direct risks of job loss or substantial change to workers’ jobs. However, there is a significant risk the video monitoring could be misused by the employer, could cause psychosocial harm to workers, and impact privacy, among other issues. The potential introduction of the AI-assisted video monitoring therefore requires a high level of engagement between the union and employer.
HEALTH AND SAFETY
AI tools can create significant health and safety risks such as intensifying work and increasing surveillance This can have a serious impact on the mental and physical wellbeing of workers.
This section outlines some of the existing health and safety practices that can be used to deal with the risks of AI in the workplace.
Worker engagement, participation, and representation
- Employers have legal requirements to engage with workers when they propose changes that may affect the health and safety of workers.
- Engagement requires the employer to share all relevant information about the matter with workers in a timely manner, and to give all workers a reasonable opportunity to express their views and contribute to the decision- making process.
- We recommend negotiating with employers to establish ways to engage with workers so that they have a say in the design and operation of safety protocols around the use of AI before they introduce any new risks to the workplace.
Health and Safety Representatives (HSRs)
- HSRs are worker representatives for worker health and safety. They have powers under the Health and Safety at Work Act 2015 to be engaged, be given information, and monitor measures taken by the employer.
- HSRs can support workers to be engaged on how AI is used in the workplace, and to raise concerns if AI is creating health and safety risks for workers.
- Ideally, HSRs will be union members and supported in the same way the union supports its delegates.
Health and Safety Committees (HSCs)
- An HSC is a forum for workers, management, and ideally unions to come together to develop and review health and safety policies and procedures.
- HSCs provide a structured and consistent way of ensuring ongoing engagement in the workplace. Where there are HSCs in the workplace, the employer must consult them on matters of health and safety.
- We recommend utilising HSCs to monitor AI being used in the workplace and monitor the control measures that are in place to keep workers safe.
RECOMMENDED COLLECTIVE AGREEMENT CLAUSES
This section provides model clauses that can be used when negotiating collective agreements on AI. In most cases, you will need to adapt these clauses to the workplace your union covers.
Getting agreement on these things will help make sure workers benefit from, and are not harmed by, AI and are engaged on decisions about its use.
Strategic approach to AI. The employer and union will agree to a strategic framework for the use of AI in the workplace. This framework should set out principles that will govern the introduction and use of AI. The framework should cover the introduction and use of AI, including:
- whether AI will be used and the metrics for deciding if, when, and how AI may be used;
- where AI is to be used or trialled, its parameters, and how its use will be reviewed to ensure that quality of work is not negatively affected;
- management of change processes, with the intention of preventing AI-related job loss;
- training and education requirements of workers;
- health and safety;
- privacy;
- sharing of benefits.
The employer and union will meet regularly to discuss, and where necessary update, this strategic framework.
Introducing AI in the workplace. Employees and their union shall be given advance notice of AI being considered for use, how the tool works, and how it could impact upon employees or the nature of their work. Introduction of AI should be by mutual agreement between the employer and union.
Prior to introducing new AI technology, an impact assessment shall be conducted, in consultation with employees and the union. This should consider, among other things:
- the purpose of introducing the technology;
- the benefits that may be created by introducing the technology;
- the risks that may be created by introducing the technology;
- whether identified risks are proportionate to achieving the intended aim;
- whether identified risks are proportionate to the expected benefits to the individuals who carry the risk;
- the potential cultural impacts and effects of the technology on Māori;
- how identified risks can be safely managed.
The employer will provide workers and the union with all relevant information including information about how the AI operates.
AI technology shall only be introduced if agreement between the employer and union has been reached following the impact assessment.
Employees and their union shall be engaged on the implementation of new AI technology, the management of risks, and the establishment and maintenance of review processes.
Job loss. AI technology will not cause job loss or significant changes to the nature of work without prior consultation and agreement with the union.
Limitation of purpose. AI technology shall only be used for the purposes it was introduced for. These purposes can be changed upon securing agreement with the union. AI technology shall not be introduced for the purposes of supporting disciplinary action.
Prevention of unfair surveillance. Monitoring and surveillance technology shall only be installed for legitimate and delimited health and safety purposes. The employer will not install monitoring and surveillance equipment
or software for this purpose unless they have secured agreement with the union. AI technology must not be used to replace other essential health and safety practices.
Privacy. No collection or analysis of workers’ sensitive personal data shall occur without employee consent. The employer will always comply with the Privacy Act 2020 and relevant codes of practice.
Data protection. The employer shall develop and agree with the union a policy on how data is to be collected, stored, and used. Data shall only be collected for valid work purposes or when directed to by law. The scope of “valid work purposes” should be agreed between the employer and union.
Access to data. Employees shall have the right to access, review, and, if inaccurate, correct any data about them that is held by the employer or a third party. Employees shall have the right to insist on the deletion of any data that is inaccurate or was obtained outside of the scope of agreement.
Employees have the right to withdraw consent for their information to be collected and held at any time.
Human in command. The principle of human in command shall be adhered to. This means that responsibility for consequential decisions shall not be given to non-human agents, such as automated management systems,
and employees shall have the right to obtain a readily understandable explanation of how decisions relating to their work have been made and be able to challenge those decisions.
Right to disconnect. Employees shall maintain the right to disconnect from work outside of their regular working hours, and shall not unreasonably be required to monitor, read, or respond to work-related communications outside of their regular working hours.
Algorithmic bias. Algorithms will be regularly reviewed to ensure they are not creating biased or discriminatory
outcomes. The results of such reviews will be made available to anyone affected by decisions that are made by or informed by algorithms.
Training and education. Training and education shall be provided to employees to ensure they are able to use AI productively and understand the risks associated with the technology and how these can be safely managed.
Job quality. The employer and union agree AI should not be used if it diminishes job quality. The employer and union will conduct regular assessments of the impact of AI on job quality. Where AI is found to be having a negative impact on job quality, the employer and union will make necessary adjustments to deal with these negative impacts.
Sharing of benefits. The employer and union shall implement mechanisms to ensure that workers share fairly in any benefits that the use of AI creates, such as increased productivity.
Regular review. The employer and union shall implement mechanisms to ensure AI clauses are regularly reviewed.