21 February 2023
Written by Dr Cobus Oosthuizen - Dean: Milpark Business School
When ChatGPT was launched in December 2022, it quickly became the fastest growing app the world has ever seen. Over one million people signed up within five days of its release and only two months later, it had 100 million people all over the world. By comparison, TikTok and Instagram took nine months and two-and-a-half years respectively to reach these numbers.
In the beginning of February 2023, Microsoft founder and thought leader Bill Gates said that the developments around artificial intelligence (AI) were the most important innovation in the world right now. He said that the programme ChatGPT, which is run by the company OpenAI, will change our world. Some people can’t seem to stop talking about it – others haven’t even heard of it – and many feel too intimidated to even weigh in on the AI debate.
Even still, on 31 January 2021, the ChatGPT website noted a record 28 million daily visits, up 165% from the month before. Undoubtedly, a large number of these visitors were merely curious and trying out the new technology, which was initially made free to anyone on the web. It resulted in a flood of reports of how wonderful it is on the one hand, and how problematic it is on the other. Fear, confusion and uncertainty about risks and the rise of AI as well as its impact on human society have sparked debates on the web, newspapers, and television across the world.
Like it or not, ChatGPT is here to stay, says Professor Johan Steyn, Human-Centred AI Advocate, and faculty member at Woxsen University, as well as a research fellow at Stellenbosch University and founder of AIforBusiness.net. Professor Steyn says generative AI like ChatGPT increases productivity, saves time, and can improve efficiency across a range of different sectors.
“Over the last few years, the application of artificial intelligence has become increasingly commonplace in our day-to-day lives,” says Professor Steyn. But what makes ChatGPT different is its superior language processing capability. It is the first AI to understand human language and be able to generate responses based on huge data stores. He adds, “Generative AI is poised to revolutionise the world in ways we can't yet imagine.”
Professor Steyn will be talking about ChatGPT and how it can change our lives at a webinar on 23 February, hosted by Milpark Education’s School of Business. The webinar will explore what generative AI is, what its impact is on education and society and what the future holds; it will tackle the potential development of GPT-4, as well as ethical, privacy and security considerations.
Ok, but how does ChatGPT work?
OpenAI’s ChatGPT is an example of an AI language model that has been in development since 2015. It is primarily a tool for communicating synthesised information. Put one way, it provides condensed, plain-language answers to questions, whereas search engines like Google produce links to possible answers. Particularly compelling is ChatGPT’s conversation-like ability, which allows users to pose follow-up questions that build on the app’s previous responses. Users can, for instance, instruct the app to “add quotes by relevant academics”, or “find case studies relevant to this topic”.
The programme has been called a tipping point for AI or even “AI’s iPhone moment,” says Professor Steyn. The fact that it is always available, and able to communicate in a number of languages – including Afrikaans – has already made it useful to educators, academics and those in the media and creative fields. It poses immediate potential to sectors using customer service, for chatbots, automatic translation, recording meetings, providing transcription etc.
Ethical greyscale
But there are also many raising questions around plagiarism and duplication of content and the dangers that such a tool may hold for education. It is already posing challenges for educators, asking them to think critically about how they assess projects and test knowledge. In response to the threats, plagiarism and AI-detection software programmes like GPTZero have been developed to help monitor plagiarism in content.
For those in fields where content production is important, a tool like ChatGPT can save hours of work, which in our fast-paced world, means a significant improvement to productivity. It takes the heavy lifting out of a project and is especially helpful with initial research, drafting and creating content (which can then be edited by the author). Say it usually takes you three hours to write a blog post, but you now only need one hour – you have scored two hours in your day. In an academic context, the potential to help researchers was demonstrated at a recent DBA proposal defence, where a student lacked appropriate depth regarding theoretical frameworks befitting that context. Such a search could take weeks or months, but with the use of ChatGPT, examples of relevant and appropriate theoretical frameworks could be found in seconds.
Yet ChatGPT is not infallible and should not be used for mathematical equations. It is also particularly spotty on events after 2020-2021. A disclaimer on the homepage also states that not all information is 100% accurate and some users claim that it is not trustworthy or reliable at all. Increasingly, some users report having uncomfortable and even worrying conversations with chatbots on the site.
ChatGPT relies on algorithms that crawl the entire web and have access to millions of mega bites of data. It democratises information and speeds up so many processes. It can write legal briefs, screen plays, policy documents – for content generation at least, the possibilities are endless.
AI and deep fakes
But Dr Louis Rosenberg, an American engineer and AI expert recently warned that AI manipulation is a serious concern. Rosenberg suggests that the use of realistic virtual spokespeople or salespeople could manipulate people into buying products, believing fake news, or revealing sensitive information, like bank account details to hackers or criminals. Through your webcam, AI can monitor your facial expression, recording even your micro expressions, giving it more information about your preferences, interests, and feelings.
Professor Steyn agrees, adding that ChatGPT could produce convincing digital copies of a person’s writing style, which could allow others to impersonate them in emails or text settings. He says that when it comes to regulating this technology or imposing controls on it, there is a fine balance between innovation and regulation. While AI technology like ChatGPT could benefit us in so many ways, as with any powerful tool, it can also cause us much harm.
“There is a danger that our reliance on technology will become unhealthy as it continues to play an increasingly important role in society,” warns Professor Steyn. “If we place an excessive amount of faith in AI language models such as ChatGPT, we run the risk of becoming overly dependent on technology for our decision-making and of losing the ability to communicate meaningfully with other people.”