On 11 October 2021, the Glasgow Local Group welcomed Marie Oldfield (Oldfield Consultancy) to deliver a seminar on Ethical Funding in Artificial Intelligence (AI) to an audience of around 30 individuals.
Marie started by highlighting the challenges in assessing the ethics of funding AI projects, particularly since ethics in AI haven’t been fully defined at present. This raised the question of what actually is 'fairness'? This is something at the heart of AI, and any analysis but which has a very subjective meaning. Marie then gave some examples of AI algorithms that have been unintentionally discriminatory, such as the Covid response to exam results.
The other issue that was highlighted is that AI is such a fast-changing subject area that it is difficult for professional bodies to keep up, never mind legislation. This means that it is difficult to prove that someone is an expert in AI, and it is therefore difficult to implement legislation. The solution requires digging in to the maths and stats behind the AI algorithm and looking at the modelling pipeline. However, it is difficult to know where to put this kind of education – can it be added at school level? In which case, does it align with the mathematics curriculum or the computer science curriculum? This makes training individuals to successfully implement AI difficult; AI draws from a number of different areas and typically requires a multidisciplinary approach, but companies can expect individuals to implement this sort of algorithm in isolation.
Marie then indicated that practitioners felt like there was too much guidance, spread a cross a variety of sources. This makes it almost impossible to have awareness of all of the guidance, which then means that practitioners tend to not use the guidance at all. To over come this there has to be a structured way of providing guidance and best practice. There has to be a set of frameworks that are jointly developed between academia and industry to ensure scalability.
Once these have been set in place, assessing the ethics of applications for funding in AI becomes a little more straightforward. It sets out the questions that can and should be asked, and allows questions around the detailed plans for dealing with ethical concerns to be asked. It also allows for things like audit processes, funding bodies and professional bodies to be set up to facilitate accreditation and training in ethical use of AI.
Author
Dr Kate Pyper (University of Strathclyde) who is the chair of the Glasgow Local Group.