Professor Rix Groenboom interviews ChatGPT
Rix Groenboom still has a vivid memory of the goosebumps he got in 1993 while building one of the first websites in the Netherlands, then as a PhD student at the University of Groningen. The same adrenaline rush the energetic New Business & ICT lecturer got from ChatGPT, the chatbot that has been rocking education land since late last year. We put it to the test and let the chatbot generate and answer a number of questions.
What might be good questions to ask an expert for an interview about using ChatGPT? Could you also answer these questions?
'Yes, I can generate and answer these questions.'
Please, explain what ChatGPT is?
'ChatGPT is an artificial intelligence model developed by OpenAI, an artificial intelligence research organisation. The model is trained on huge amounts of text and is able to understand human language and respond in a human way.'
What can you use it for?
'ChatGPT can be used to automatically generate texts such as summaries, articles and product descriptions. For analysing conversations and identifying trends, emotions and intentions. The model is also useful for other applications where human language needs to be processed, such as translations.'
Can you explain to us how ChatGPT works and the technologies behind it?
'ChatGPT uses transformational neural networks (Transformer Networks) and is trained with billions of text fragments from different sources. The goal of ChatGPT is to help users find answers to questions and tasks by using text input. ChatGPT analyses the input and then tries to generate an appropriate answer through pattern recognition and predictions based on the trained data.'
What are the benefits and challenges of using ChatGPT?
'The benefits are quick response time. The chatbot can answer questions quickly, efficiently and accurately, and can take over tasks that would otherwise have to be performed by humans. A challenges is the limited knowledge. ChatGPT is only as good as the data that trains it, so the quality and amount of information available determines the accuracy of answers. ChatGPT is trained on existing data, which can lead to giving incorrect or biased information. Moreover, a human being is still needed to ensure that ChatGPT is used correctly and ethically.'
ChatGPT has an impact on education. What are the implications?
'AI technologies can offer students a more personalised education by determining which subjects to study based on their achievements and interests. In addition, students can achieve a better understanding of what they are learning by being provided with interactive visualisations, simulations and clarifying examples. The technologies can help teachers grade and review assignments more efficiently, giving them more time for other important tasks. There are also negative consequences. By using AI technologies in teaching, students may experience less human interaction, which can have negative effects on their social-emotional development. Implementing AI technologies in education requires technical skills and infrastructure, which can be challenging for schools and universities. Overall, it is important to deploy AI technologies in education responsibly and ethically, considering the pros and cons and protecting students' human interaction and socio-emotional development.'
How many Hanze UAS students have ever used ChatGPT?
'It is difficult to determine the exact number of students at Hanze UAS, as ChatGPT is a popular tool, it is likely that some students at Hanze use it as part of their studies or projects. However, as ChatGPT is a tool often used by companies and research institutions, it is likely that the number of students at Hanze University using it directly is limited.'
What are the main challenges ChatGPT still faces?
'It is important to ensure the accuracy of the information generated by ChatGPT. Because ChatGPT is trained on data from the internet, it can have a biased view of the world. It is also very important to ensure user-privacy and the security of the data ChatGPT works with.'
Professor Rix Groenboom is particularly fascinated by the possibilities of ChatGPT. 'Besides human language, the chatbot can also code programming languages. It passed the Master of Business Adminstration final exam. There are all kinds of programmes around that claim to recognise whether a text was created by ChatGPT, but I have not yet been able to find one that actually does that. So far, Microsoft has put one billion into the development of such programmes; they say they will invest another 10 billion. That's more than our defence budget. This development is definitely not just for nerds, but for the general public,' Rix Groenboom concludes.
The makers of chatbot ChatGPT themselves have come up with a solution to recognise artificial intelligence-generated texts. The AI Text Classifier programme tries to estimate whether a text was created by a human or a machine.
The illustration was created by DALL' E the artistic sister of ChatGPT. Keywords: chatbot in the style of Mondrian.
How satisfied are you with the information on this page?