Research. A recently published report by The Royal Society looked at the different opportunities and challenges that come with machine learning advancements in the UK. Research of public opinion on the matter showed perceived opportunities regarding objectivity, accuracy, efficiency, economic growth and solving societal problems. Concerns were characterized by notions of physical harm, workforce replacement, impersonal and restricted human experiences.
The report describes the landscape painted by machine learning, with both opportunities and challenges. It starts by mentioning some of the most common daily interactions between users and machine learning systems such as image recognition systems (e.g. social media), voice recognition systems (e.g. virtual personal assistants) and recommender systems (e.g. online retailers). The Royal Society suggests that the UK needs to catch up with machine learning advancements. In this respect, three skills need to be addressed:
- basic understanding of general public on use of data and machine learning systems
- new mechanisms of informing users and professionals in the field
- increased support in building advanced skills in machine learning
A Public Dialogue on Machine Learning
The public’s opinion on this topic was researched as part of this project. In this respect, both quantitative and qualitative research was carried out. Firstly, a survey of just under 1000 respondents looked at levels of awareness and views on machine learning. Secondly, the general public was put in dialogue with Society’s machine learning Working Group to assess the implications of this technology. Debates were held in Birmingham, Huddersfield, London and Oxford. Machine learning was assessed through its practical uses in areas such as health, social care, marketing, transport, finance, policing, crime, education, and art.
The clearest of the findings as assessed by the Royal Society in this respect is that the public’s attitudes are diverse and highly dependent on the context. Technology was assessed on the basis of:
- perceived intention of those using the technology;
- beneficiaries’ profile;
- necessity to use machine learning;
- clear inappropriateness of activities;
- human involvement in decision-making.
Awareness of the term is low, but increases when practical applications are mentioned.
- only 9% of those surveyed recognized the term ‘machine learning’, yet majority is aware of some of its applications
- 76% had heard of computers that can recognize speech and answer question
- during public debates, participants mostly mentioned experiences with systems used by online retailers and offers from loyalty cards.
Attitudes are context specific. Even though some themes were common in public discussions, the nature and extent of concerns along with the perception of potential opportunities varied with each different application of machine learning.
Participants mostly associated the opportunities arising from machine learning with the term’s connection with big data and the capability to analyse data. The following opportunities were identified:
- Objectivity in comparison with humans’ and human error decrease – e.g. humans might get tired or emotional
- Accuracy increase- e.g. detection of medical images and making diagnoses
- Efficiency – e.g. public sector resources use
- Economic growth – e.g. increased business opportunities
- Large scale societal problems solutions- e.g. addressing climate change
Four themes were identified:
- Harm – e.g. autonomous vehicles accidents
- Replacement – in the workplace by machines or over-reliance on them
- Impersonal experiences or services – e.g. generalized predictions on individuals or changes in the nature of valued activities
- Restriction on human experience – e.g. directed marketing
In order to address the findings of the public dialogue research, The Royal Society report recommends:
- continuous engagement between researchers and the public – e.g. Governmental public engagement framework
- training in ethics as part of postgraduate students researching the field of machine learning
Read more on the topic:
Royal Society project presentation here.
Public dialogue summary here.
Full report here.