ChatGPT: Exploring the risks of unregulated AI in South Africa

May 1st, 2023
x
Bookmark

ChatGPT is a technological breakthrough like no other. Even though the technology has been around for a long time the latest innovation of generative artificial intelligence (AI) exploded onto the tech scene and into our lives over the past few months. With its incredible ability to provide lengthy answers with humanlike sophistication, ChatGPT has been hailed a ‘game changer’ by many. It has not only changed the game – it changed everything – the players, the rules, the outcome right down to the most minute detail. It is revolutionary, with enormous potential to both benefit and harm society.

The fundamental problem is that the inexorable AI is on a collision course with democratic values like accountability, transparency, fairness, and equality. Opaque algorithms can now influence how we work, live and play by determining who qualifies for credit agreements, who gets considered for jobs, and who may be summarily arrested through predictive policing with facial recognition technology. This article explores the risks for society and humanity if AI proceeds without legal regulation.

What can ChatGPT do for lawyers?

For illustrative purposes, I prompted ChatGPT about the role of AI in legal practice, and this was the very considered response it generated.

Title: Embracing the future: Why lawyers must adopt AI in legal practice

Introduction: In an era where technology and artificial intelligence (AI) is rapidly transforming various industries, the legal profession cannot afford to be left behind. Despite the traditional resistance to change, lawyers must embrace technology and AI to thrive in the modern legal landscape. The benefits of AI adoption in legal practice are manifold, from increased efficiency and cost-effectiveness to improved accuracy and decision-making.

  • Increased efficiency: One of the most compelling reasons for lawyers to adopt AI is the increased efficiency it brings to legal practice. AI-powered tools can help automate routine tasks, such as document review, legal research, and contract analysis, allowing lawyers to focus on more strategic and higher-value work. By reducing time spent on mundane tasks, AI enables lawyers to serve more clients and tackle more complex issues, ultimately increasing productivity and revenue.
  • Cost-effectiveness: By streamlining workflows and automating repetitive tasks, AI can significantly reduce costs in legal practice. Clients are increasingly demanding more cost-effective legal services and embracing AI can help law firms meet this expectation. Furthermore, AI-powered tools can help lawyers identify potential cost-saving measures and reduce the likelihood of costly errors. As a result, law firms can offer competitive pricing and attract a broader clientele, driving revenue growth.
  • Improved accuracy and reduced errors: AI algorithms can analyse vast amounts of data and identify patterns with a level of accuracy that often surpasses human capability. This can be particularly helpful in tasks such as due diligence, where AI can quickly identify relevant information and flag potential issues. By minimising errors and ensuring a higher degree of accuracy, AI can help lawyers deliver better quality services and reduce the risk of malpractice claims.
  • Enhanced decision-making and predictive analysis: AI has the potential to revolutionise the way lawyers make decisions and advise clients. By analysing historical data, AI can identify trends and patterns, enabling lawyers to make more informed decisions and provide better guidance. Additionally, AI can be used to predict case outcomes based on previous rulings and case factors, allowing lawyers to develop better litigation strategies and advise clients on the likelihood of success.
  • Increased access to justice: By automating certain tasks and reducing costs, AI has the potential to make legal services more accessible to a wider range of clients, particularly those who may have been priced out of the market. This increased access to justice will benefit society as a whole, as more people can obtain legal representation and protect their rights.
  • Staying competitive in a changing legal landscape: As more law firms and legal professionals adopt AI, those who fail to do so risk falling behind their competitors. To remain competitive and attract clients, lawyers must demonstrate that they are at the forefront of technological advancements and able to offer the most efficient and effective legal services.

Conclusion: The adoption of AI in legal practice is not a question of if but when. By embracing technology and AI, lawyers can increase efficiency, reduce costs, improve accuracy, and enhance decision-making. Ultimately, AI adoption will not only benefit lawyers and their clients but will also contribute to a more just and equitable society. The time to embrace the future of legal practice is now.

Impressive and compelling indeed. Next, I prompted ChatGPT to tell me about the risks of adopting AI in legal practice:

Title: Caution ahead: The dangers of AI in legal practice

Introduction: While the integration of artificial intelligence (AI) into various industries has led to numerous benefits, it is crucial to consider the potential risks and downsides of incorporating AI in legal practice. By examining the ethical, privacy, and employment concerns, we can better understand the potential pitfalls of relying too heavily on AI in the legal profession.

  • Ethical concerns: The use of AI in legal practice raises several ethical concerns. AI algorithms can inadvertently perpetuate or even exacerbate existing biases in the legal system, as they are trained on historical data that may be tainted with human prejudice. Consequently, AI-driven decisions may inadvertently discriminate against certain groups, leading to unfair outcomes and undermining the core principles of justice and equality.
  • Privacy and security risks: Incorporating AI into legal practice often involves the processing of vast amounts of sensitive and confidential data. The collection, storage, and analysis of this data can pose significant privacy and security risks, especially if proper safeguards are not in place. A data breach or unauthorised access to sensitive client information could have severe legal and reputational consequences for law firms and practitioners.
  • Loss of human judgment and intuition: AI algorithms, despite their accuracy and efficiency, lack the human intuition and judgment that is often crucial in legal practice. Lawyers possess an innate understanding of human emotions, social contexts, and moral values that cannot be replicated by machines. Relying too heavily on AI can lead to a loss of this essential human touch, potentially resulting in suboptimal decisions and misguided legal advice.
  • Unemployment and job displacement: The automation of routine legal tasks by AI could lead to job displacement and unemployment among junior lawyers and support staff. These positions are often the entry points for young professionals seeking to gain experience and develop their legal careers. The loss of such opportunities may create barriers to entry into the legal profession and exacerbate existing inequalities.
  • Accountability and liability: As AI assumes a more significant role in legal practice, determining accountability and liability for errors and poor outcomes becomes increasingly complex. It may be difficult to ascertain whether a mistake is attributable to the AI system, the programmer, the data source, or the legal professional using the AI tool. This ambiguity can lead to challenges in assigning responsibility and may undermine clients’ trust in the legal system.
  • Over-reliance on AI: While AI can be a valuable tool in legal practice, there is a danger of becoming overly reliant on it. Lawyers must remain vigilant and avoid complacency, continuously questioning and scrutinising AI-driven results. Over-reliance on AI can lead to a decline in critical thinking and problem-solving skills, ultimately impairing the quality of legal services provided.

Conclusion: Although AI has the potential to revolutionise the legal profession, it is crucial to proceed with caution and carefully consider the potential dangers. By acknowledging and addressing the ethical, privacy, and employment concerns associated with AI in legal practice, we can strike a balance between harnessing the benefits of AI and preserving the essential human qualities that underpin the legal profession.

So, what is the problem?

ChatGPT’s reasoning is smart, superior, and logical. It neatly encapsulates the ethical, legal, and social dilemmas created by AI. The problem is that AI is so sophisticated and realistic that some vulnerable individuals may be at great risk when interacting with these intelligent machines.

In the March 2023 issue of De Rebus, Prof Michele van Eck illustrated the shortcomings of ChatGPT in making factual mistakes and giving incorrect answers (Prof Michele van Eck ‘Chatting with ChatGPT: Will attorneys be able to use AI to draft contracts?’ 2023 (April) DR 12). It tends to ‘hallucinate’ and invent answers or make statements with great certainty that are completely disconnected from reality. The Washington Post reported on 5 April 2023 that ChatGPT invented a sexual harassment scandal and named a real law professor as the accused while fabricating primary sources to back up its claims (Pranshu Verma and Will Oremus ‘ChatGPT invented a sexual harassment scandal and named a real law prof as the accused’ (www.washingtonpost.com, accessed 12-4-2023)). Another extreme example took place in Belgium where a chatbot became so manipulative that it convinced a user to commit suicide.

Even more concerning is the recent open letter by 1 100+ notable signatories including Elon Musk and Apple co-founder, Steve Wozniak calling on ‘all AI labs to immediately pause for at least [six] months the training of AI systems more powerful than GPT-4’ (Future of Life Institute ‘Pause giant AI experiments: An open letter’ (https://futureoflife.org, accessed 12-4-2023)). They raise valid concerns about the development of ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control.

In a recent article, Bill Gates also raises the possibility that AI may run out of control. He notes the long-standing fear that super-intelligent machines may well decide one day that humans are a threat. At present, nobody knows how far away we are from the development of Artificial General Intelligence (AGI). This will be a machine that can do anything a human brain can do but without any limits on the size of its memory or the speed at which it operates. Such AGI will be able to make its own decisions that may conflict with those of humanity. For the time being, AGI does not exist but the speed at which the technology is developing means that it cannot be predicted when this will become a reality. GPT-5, the newest rendition in the works at OpenAI is already a step in that direction.

What are governments doing about these technological breakthroughs?

We are at the foothills of the mountain when it comes to innovation in tech. Worldwide governments are responding. Italy’s swift move was to ban ChatGPT on 31 March 2023, citing data privacy concerns and the potential exposure of misinformation to young people.

The United Kingdom, on the other hand, has unveiled a world-leading approach to innovation, in publishing the first AI White Paper ‘A pro-innovation approach to AI regulations’ in March 2023, to regulate AI and build public trust in cutting-edge technologies. The United Arab Emirates appointed the world’s first and only Minister of State for Artificial Intelligence.

In the European Union, plans are underway for new legislation to regulate AI – the Artificial Intelligence Act. In Australia, law schools have held collaborative sessions to examine the impact of ChatGPT on legal education, academic integrity, and the legal ecosystem. Even though the United States (US) has published a potential AI Bill of Rights, there is no comprehensive federal legislation on AI in the US yet.

The situation in South Africa (SA) is troubling. At the moment, there is no comprehensive legislation which governs the use of AI and machine learning in the country.

Conclusion

AI is pervasive. It is penetrating every area of our lives daily, almost minute by minute and, without being alarmist, conspiratorial, or fearmongering, the dangers of such explosive technology are real. That should not, however, deter the enormous potential it has as a revolutionary tool.

This is an urgent call for a dedicated national AI strategy to address the ethical and social challenges highlighted in this article. We need a legislative framework in SA that will put the necessary guardrails in place, educate the public and develop expertise in government to manage this technological avalanche.

Without a legal regulatory framework, we will have to play catch-up with the rest of the world as technological advances race along at breakneck speed. The time for legislatures and policymakers to step in is now.

Diana Mabasa LLM (Wits) Legal Technology and Innovation Certificate (LT&I Institute, Dubai) is in-house counsel for the Office of the Legal Services Ombud.

This article was first published in De Rebus in 2023 (May) DR 17.

X
De Rebus