Most systems using artificial intelligence are reliant on data to be effective. In this space, that would include examples such as:

• Analysing the questions students ask to create better answers
• Recognising descriptions of disabilities and matching these to barriers or conditions
• Reusing data from past conversations or records to avoid repetition.

A further aspect of data sharing and privacy, raised repeatedly in our workshop and project discussions, is that the systems shouldn’t just be designed to help individual students overcome barriers, but also to guide educators and institutions to recognise and remove them.

People are rightly concerned (and legally protected) with regards to the use of their sensitive personal data.

These issues raise many questions, such as how do we create useful and representative data sets? How should data sharing be managed and clearly understood by actors and systems? And how can systems be managed such that inaccuracies or biases are recognised, and accuracy and currency be maintained and enhanced over time.