If you have asked an app like ChatGPT or Claude for a recipe recommendation, travel tips or tech support lately, you are not alone. Around seven in ten adults in South Africa have used an artificial intelligence (AI) chatbot, with many of us relying on the technology to help us with everyday advice and tasks.
Yet many people have a healthy scepticism about AI when it comes to trusting the technology with high-stakes issues such as legal advice or financial planning, generally delivered by experts such as attorneys, tax practitioners or debt counsellors. These matters are personal and consequential, so most people understand that they cannot risk acting based on the wrong information.
There is the nagging worry that the machine cannot understand your circumstances or empathise with your concerns. You may also wonder if you even know enough to ask the AI the right questions about a complex issue such as dispute with SARS or your employer. And as a layperson, how will you even know if the AI has given you incorrected or outdated info?
In short, how do you know if you can trust AI about big, important decisions?
But what if that is the wrong question? Perhaps we should ask how AI can help us to become better informed so that we are more empowered when we make decisions or seek help from a human professional. It is a tool that empowers and educates us, which in turn helps protect us from being misled or exploited. When we think of AI in this way, it becomes a tool that supports rather than replaces human advice and decision-making. We don’t have to trust blindly anymore, and we might be able to be of more help to the professional by providing relevant information and having more realistic expectations.
Imagine, for example, if you receive a formal written warning at work. You can upload the letter and your company’s disciplinary code into an AI tool. The AI chatbot can then explain any unfamiliar terminology, outline your rights under labour law and company policy, and suggest questions to raise with your employer, legal advisor or trade union representative.
The same principle could apply to other facets of your life. You could ask AI about the basic principles of saving for your retirement or managing your personal cash flow before you approach a debt counsellor. Or you could ask for some tips about what to ask a personal trainer before you go to meet them at the gym.
The AI tool is not a replacement for human advice.
It gives you a starting point, so that you can determine whether you need professional advice and what you should ask when you speak to an advisor. It closes the gap between knowing that you have a problem you need to address and knowing what you need to ask to resolve your challenge.
With the correct instructions from humans, many AI tools excel at tailoring their advice to your level of expertise. You can, for example, ask for a plain language explanation of legal or financial jargon. You can also get advice about which information is important as well as which questions you should be asking before deciding or seeking professional advice.
Knowing your rights before signing an employment agreement, accepting a payment arrangement or responding to a legal demand can change how you approach the situation. Understanding the basics of a financial decision can make it easier to question an assumption or seek appropriate advice.
Using AI in this way is not about treating it as an infallible source of information.
We all need to be responsible about how we use these tools because without proper prompting, they can produce convincing answers that are incomplete, inappropriate to your situation, or based on older information.
Using a specialised tool rather than general purpose AI platforms can often improve the quality and reliability of the information you receive. Specialised platforms overcome some of these limitations by focusing on a smaller use case and using a focused dataset that is specifically applicable to that field. For a legal question, a system built around South African law can provide a more relevant starting point than a general chatbot trained on data from around the world.
Having human experts in the loop to curate and double-check that information is accurate and up to date is an important part of this approach. It also means you can contact a human adviser when a question is too complex, specific or consequential for AI alone.
Help with high-stakes decisions
When it comes to matters like the law or your finances, the consequences of the wrong advice are far more serious than a bad restaurant suggestion. The higher the stakes, the more important it is to double-check information and know when to involve a qualified professional.
As such, ‘trusting’ AI is not about giving up your agency to a machine. It is about understanding how AI can give you a useful starting point. The point is not to let a machine make your decisions or replace human experts. It is to make sure you understand enough to ask the right questions and to know when you need the help of a professional.
