ChatGPT - a disruptive tool we need to understand


This week on the Lisa Burke Show, Professor Valérie Schafer, Nathan Summers and Patrick Mischo discuss ChatGPT and AI more generally as it becomes part of our life.

OpenAI released an early demo of ChatGPT on November 30, 2022, and since then this chatbot has infiltrated conversation across many domains.

Top of the list, perhaps, was education, and so I turned to the University of Luxembourg to tell me who is working on how to figure out how to deal with Chat GPT at this level. Professor Valérie Schafer is a Professor in Contemporary History at the C²DH (Luxembourg Centre for Contemporary and Digital History) at the university. Her main research interests are the history of the Internet and the Web, digital cultures and infrastructures, and born-digital heritage (especially Web archives).

My other guests are Nathan Summers and Patrick Mischo. Nathan is an AI Research Analyst working for the Luxembourg Tech School. His work primarily focuses on ethical dilemmas associated with real-world application of AI systems. Patrick Mischo is the Office Senior Partner of Allen & Overy in Luxembourg. He talks about 'Harvey' - Open AI’s GPT for for lawyers.

As we enter this new world, with Open AI such as Chat GPT, we need to become even more aware of the way in which it works, how we can use it for good, and what the pitfalls could be. Naturally, with education (and other work), universities need to think about plagiarism and fraud. How can one use Chat GPT in a creative and pedagogical way? How should one word and reword questions to get the best out of Chat GPT?

Valérie explains that they use Chat GPT as a base, as a tool, with students and improve on one's answer from there. However, there is great risk of AI hallucinations, as Nathan Summers explains. AI systems can very confidently provide us with untrue information. There is also inherent bias. Despite Chat GPT being based on enormous amounts (but only with data up to 2021), this data is likely skewed as the internet has been used and populated with information predominantly in certain countries. It will become very important to constantly cross-reference any information given by Chat GPT.

As it happens, the mechanism of AI systems is a black box - we don't really know the internal workings of these neural networks, especially in proprietary systems. This raises a different issue, once more, the power of big tech companies in our lives? Valérie reminds us that if the information is free, we are the product, and with that we need to be increasingly careful of what information we input. This is especially true, and precarious for young people.

When it comes to delegating small tasks or generative roles, AI will definitely change the job market. Patrick Mischo talks about the use of Harvey within their work as a law firm. However, AI systems cannot be held legally liable for damages. As with education, in the legal field it is necessary to develop a critical eye to spot any potential inaccuracies.

We'd love to hear your thoughts on how Chat GPT is changing your work or life. Are you worried about your job? Do you think it's a wonderful addition to the tech world we co-habit?