The UG wants to buy AI chatbot licences as soon as possible to better protect data within the university.
‘We want to limit the amount of people putting sensitive data into public versions of chatbots’, explains director of the Centre for Information Technology (CIT) Nolda Tipping-Griffioen.
The university wants to ensure that the data will be saved in an European service and won’t be used on training modules. It had hoped to buy licences for the popular chatbot ChatGPT, but that proved impossible within the EU’s General Data Protection Regulation. ‘The data would still go to a service in the US’, says Steph de Gunst, business analyst at the CIT. ‘We wouldn’t have any control over the data, and that’s not what we want from a privacy perspective.’
Le Chat
The UG is now negotiating with the French company Mistral (Le Chat) as a safer and cheaper alternative. The aim is to implement the measure ‘as soon as possible’ because of the growing concerns about data safety.
The licences – up to a thousand – will be distributed equally amongst all faculties, and each one gets to decide how they will distribute them within their department. They will then also be made available to students.
AI literacy
In addition, the UG also plans to further develop its e-learning module on AI literacy to cover specific use cases or topics, in order to promote responsible use. ‘We hope to raise awareness to the fact that what they put into these bots is no longer private’, says Tipping-Griffioen.
UKrant is publishing the final part of its two-part series on the use of chatbots in education this week. While the first part featured students, this instalment focuses on lecturers.