Amalta Seevnarayan Dhanee | Attorney – Legal Division
What is AI?
Wikipedia defines Artificial Intelligence (“AI”) as follows: “intelligence demonstrated by computers, as opposed to human or animal intelligence. "Intelligence" encompasses the ability to learn and to reason, to generalize, and to infer meaning.”
Simply put this means that computers/robots who are controlled by computers have the ability to perform tasks that would ordinarily be done by human beings.
AI has been in use for many years and individuals have been utilising it without giving it a second thought. Google search, self-drive vehicles with enhanced safety features, google maps, Siri, and Netflix, to name a few, all run with AI applications (“AI apps”) in the background. With the release of ChatGPT on 30 November 2022, AI has generated a new buzz around it, reaching one million users in the first five days after release.
But can we rely on these AI apps and what are the pitfalls of using same in business or otherwise?
Some of the Benefit of using AI
AI has numerous benefits such as:
- It has the ability to process and analyse large volumes of data in a short period of time.
- It can automate monotonous tasks which humans don’t enjoy doing. Further, unlike humans it does not tire easily.
- It performs tasks with accuracy and works at a high speed.
- It reduces the cost to task ratio, meaning that AI can assist businesses and individuals in drafting documents and/or agreements and the like at minimal costs.
- It can assist in research.
- Assisting with risk assessments and fraud detection.
- It can identify risks in property transactions by flagging illicit transactions, identify money laundering and terrorist financing.
Some of the Pitfalls of Using AI
When using AI, one needs to bear in mind that one cannot blindly rely on the information provided, especially in drafting of Legal contracts, Last Will and Testaments and the like. Should one elect to do so there is a possibility that one could end up with a document that is not legally sound and enforceable. The reason being is that these AI apps do not have the ability to engage the user by asking the right questions. The AI apps are solely dependant on the information it is provided by the user. Further, when using AI apps to draft documentation especially legally binding documentation, the source must be credited.
Whilst AI apps are useful tools when it comes to assisting with research, the user must refrain from accepting the information received at face value. The information received must be verified first.
There have been a number of cases in which ChatGPT proved to be unreliable. In many instances it referred to case law which did not exist and provided incorrect case citations.
In another instance it referred to an article published in the Washington Post, when in fact no such article to that effect had ever been published. In response to the fabricated article, the spokesperson for ChatGPT stated that “When users sign up for ChatGPT, we strive to be as transparent as possible that it may not always generate accurate answers. Improving factual accuracy is a significant focus for us, and we are making progress.”
So, whilst AI is a wonderful efficient tool with information and being available at the click of a button, and businesses can create their own AI model to use within their organizations, this comes at a substantial cost. When used correctly it does still pose certain risks to businesses. Further the business needs to ensure that its AI model is fully compliant with the Protection of Personal Information Act 2013.
Firstly, if the incorrect information is entered onto the AI app, it can be harmful to users as the AI can’t currently prompt the user which will allow for errors based on the incorrect information inputted by the user. One cannot engage with the AI app in the same manner as one would with an attorney or other professional. Further, there is a potential for bias and reputational damage.
Secondly, the user must refrain from entering personal/sensitive information of clients onto the apps as AI retains this information and may use it in other instances.
Currently in South Africa there is no legislation to protect users against AI apps which is concerning. The most significant threat with using AI is cyber security and protection of personal information. It is essential one familiarize oneself with the terms and conditions of use on the respective app being used. Further, if you are utilizing an AI model in your business, your client’s consent to the use of same must be obtained. This applies if client’s personal information and/or sensitive data is being fed into the AI model.
Can one rely on AI?
In summary one cannot accept the information received from AI apps at face value. When using AI apps ensure the following:
- The information received must be verified first to ensure it is factually correct.
- Refrain from entering personal information on the apps.
- If creating an AI model specific to your business, ensure it is compliant with Protection of Personal Information Act 2013.
- Familiarize yourself with the terms and condition of the relevant AI app being used.
- Ensure the correct information is entered on the app as AI lacks the human element which will prompt the user to ask the right questions.