In November 2018, the UK Parliamentary All Party Group on Data Ethics launched a call for evidence about the use of data and machine learning in four areas:
- Autonomous vehicles
With the help of some excellent colleagues, I submitted a paper on behalf of the institution (Nottingham Trent University) outlining our great interest in the potential for learning analytics, but also our concerns.
Our conclusion was
Learning analytics (and big data more generally) is starting to change the nature of how universities support students. There are enormous potential benefits for the way that institutions do so. However, in addition to retaining healthy skepticism about some of the claims made by technology vendors, this response ends with the following recommendations.
- Institutions must have a clear and justifiable rationale for using data and algorithms.
- This rationale must be encoded in policy and the policy must be enacted and periodically reviewed.
- The rationale must be communicated to staff and students. Furthermore, training should be part of any communications strategy, in both the resource and data literacy more generally.
- Institutions have a moral responsibility to ensure that any algorithms used to make decisions about students are understood and that the data sources used are agreed.
- Using demographic data in any algorithm requires careful consideration. It may be better to avoid using it altogether and instead use more-actionable data sources.