Let’s Chat about ChatGPT: Unravelling the Employment AI-ssues in South Africa

Introduction

A phenomenon that has taken the world by storm is the emergence of language models like ChatGPT. In the quick and ever-changing landscape of technology, artificial intelligence (“AI”) continues to reshape various aspects of our reality and will undoubtably have implications for the South African legal fraternity. The objective of this article is to delve into the legal considerations surrounding the use of ChatGPT in the employment landscape, exploring its impact on privacy and data protection, employment practices, and job security.

Unravelling the Legal Threads

What is ChatGPT and why should I care about it?

ChatGPT is a type of conversational AI model developed by OpenAI. It is based on the GPT (Generative Pre-trained Transformer) architecture, which uses a transformer network to process and understand text. It can perform a wide range of language tasks, including answering questions, providing explanations, generating conversational responses, and offering suggestions.

ChatGPT is trained to capture the semantics and syntax of natural language, allowing it to generate text that closely resembles human speech. It can understand some context, recognise patterns, and generate meaningful responses based on the input it receives. The model has been fine-tuned on specific datasets to ensure accuracy and quality in its responses.

ChatGPT has numerous applications, including assisting with information retrieval, providing customer support, offering writing suggestions, aiding in language translation, and facilitating interactive conversations. It aims to simulate human-like communication, making it a powerful tool for various tasks that involve processing and generating text.

The adoption of AI models such as ChatGPT by employers raises questions regarding the rights of employees and labour relations. ChatGPT has the potential to impact privacy and data protection and influence job security.

Privacy and Data Protection

 The Protection of Personal Information Act 4 of 2013 (“POPIA”) imposes obligations on employers to safeguard employee personal information. The impact of ChatGPT on POPIA would primarily revolve around the handling and processing of personal information during interactions with the AI system.

ChatGPT involves the collection and processing of personal information during conversations. Employers using ChatGPT should therefore ensure that they comply with POPIA’s requirements regarding informed consent and the lawful processing of personal information. Clear consent mechanisms and policies should be in place to obtain consent from employees and individuals whose data is being processed by ChatGPT.

ChatGPT relies on data inputs to generate responses. Employers must ensure that appropriate security measures are in place to protect the personal information shared during interactions with ChatGPT. This includes encryption, access controls, and safeguards against unauthorised access to or disclosure of personal data.

POPIA mandates that personal information be collected and used for specific, lawful purposes. Employers should ensure that ChatGPT is used within the scope of authorised purposes and that any personal data collected is not used for other unrelated activities. It is important to review and update privacy policies to include ChatGPT usage and data handling.

POPIA specifies that personal information should not be retained for longer than necessary. Employers should establish data retention policies that govern how long personal information collected during ChatGPT interactions is stored and ensure compliance with POPIA’s requirements.

Employees interacting with ChatGPT may exercise their rights under POPIA, such as accessing their personal information or requesting corrections. Employers must have processes in place to handle these requests and ensure that employees can effectively exercise their rights in relation to the data collected during ChatGPT interactions.

Employers should carefully evaluate the involvement of third-party AI providers in the operation of ChatGPT. It is important to assess the data protection practices and compliance of these providers with POPIA to ensure that personal information is handled in accordance with the law.

Discrimination and Fair Employment Practices

While ChatGPT can be a powerful tool for workplace interactions, there is also the potential for discriminatory practices to arise when using ChatGPT in the workplace.

ChatGPT learns from vast amounts of training data, which can include text from a wide range of sources. If the training data contains biased or discriminatory content, ChatGPT might inadvertently generate responses that reflect those biases.

If users, consciously or unconsciously, input biased or discriminatory prompts when engaging with ChatGPT, it can perpetuate discriminatory language or ideas. This can inadvertently reinforce and amplify existing biases and prejudices that may exist in the workplace.

ChatGPT or similar AI systems can be used in various workplace scenarios, such as recruitment, performance evaluations, or promotion processes. If these systems rely on biased or discriminatory criteria or data, it can result in unfair treatment and adverse impacts on certain individuals or groups.

ChatGPT may struggle to fully grasp the nuances of specific situations or understand the broader context in which certain conversations or workplace interactions occur. This limitation can potentially lead to misinterpretations, miscommunications, or biased responses that do not align with fair and inclusive practices.

Section 5 of the Employment Equity Act 55 of 1998 (“EEA”) holds that every employer must take steps to promote equal opportunity in the workplace by eliminating unfair discrimination in any employment policy or practice.

Section 6 of the EEA regulates that no person may unfairly discriminate, directly or indirectly, against an employee in any employment policy or practice on one or more grounds. These grounds include race, gender, sex, pregnancy, marital status, family responsibility, ethnic or social origin, colour, sexual orientation, age, disability, religion, HIV status, conscience, belief, political opinion, culture, language, birth, or any other arbitrary ground.

It is essential for employers to recognise the potential for discriminatory practices in AI systems, including ChatGPT. Employers must take proactive measures to ensure fairness, inclusivity, and adherence to labour legislation. These measures include:

• ensuring diverse and unbiased training data to mitigate the risk of perpetuating biases;
• implementing regular audits and evaluations of AI systems to identify and address any discriminatory patterns or outputs;
• providing clear guidelines and training to employees on the appropriate and responsible use of ChatGPT to prevent discriminatory inputs or prompts; and
• encouraging an inclusive and diverse work environment that promotes awareness, education, and open dialogue about biases and discrimination.

Employers must prioritise diversity, inclusive practices, and human oversight to mitigate the risks of discriminatory practices in the workplace and ensure that AI systems like ChatGPT contribute to a fair and equitable work environment.

Impact on job security

The integration of ChatGPT into business operations has the potential to render certain employment categories redundant. An emerging question is whether an employer can simply dismiss an employee whose employment has been rendered redundant due to ChatGPT.

In essence, while an employer cannot simply dismiss an employee whose role is deemed redundant, there is a possibility of them being retrenched, resulting in job loss. In terms of the Labour Relations Act, 66 of 1995 (“LRA”), employers can initiate retrenchment processes when faced with needs that necessitate terminating employment. The employer must adhere to the retrenchment procedures outlined in the LRA when carrying out such retrenchments.

Section 189 of the LRA permits employers to dismiss employees for operational requirements. These are defined as requirements based on the economic, technological, structural, or similar needs of the employer. An employer who employs 50 or fewer employees is subject to the process set out in section 189 of the LRA, which sets out the procedural and substantive obligations placed on the employer to maintain a fair retrenchment process.

The retrenchment procedure entails several key steps to ensure fairness and compliance with section 189 of the LRA. It begins with meaningful consultations between the employer and affected employees, along with any relevant trade union or workplace forum, to explore alternatives and discuss the reasons for retrenchment. This consultation is a process, not merely a once-off meeting. The consulting parties must attempt to reach consensus on, among other things, the possibility of avoiding dismissal and appropriate measures to mitigate the effects of retrenchment.

Fair and objective selection criteria must be established to determine which employees will be retrenched. The employer must issue notices to the employees who have been selected to be retrenched after the consultation process has been completed.

Employees are entitled to severance pay if they are retrenched for operational requirements. The requirements regarding severance pay are set out in section 41 of the Basic Conditions of Employment Act, 75 of 1997 (“BCEA”). Section 41 of the BCEA provides that an employer must pay an employee who has been dismissed for operational requirements severance pay equal to at least one week’s remuneration for each completed year of service with that employer.

The specifics of the retrenchment procedure may vary depending on the applicable agreements in place; consequently, seeking professional legal advice is recommended to ensure compliance with the law.

Will ChatGPT be able to replace attorneys

The short answer is no. Whilst you can ask ChatGPT to provide you with basic documents such as lease agreements, it would be impossible for it to assist you in recovering arrear rentals or any court process that involves an evaluation of the available evidence. Only an attorney would be able to advise whether your evidence is sufficient to win on the day or to cross-examine the lessee, as an example.

Conclusion

The introduction of ChatGPT in the workplace presents both opportunities and challenges from an employment law perspective. It is crucial for employers to consult with legal experts and comply with applicable labour laws and regulations when implementing AI technologies like ChatGPT in the workplace. Labour legislation in South Africa provides protections for employees’ rights, and employers should ensure that these rights are upheld in any AI-related initiatives. By proactively addressing these implications and implementing appropriate policies and safeguards, employers can embrace the benefits of ChatGPT while safeguarding employee rights, fostering fair employment practices, and promoting a harmonious work environment.