The pharmaceutical industry, the Higher Education sector and the world of the arts may seem like three very disparate areas but, along with almost all arenas of modern life, all three are heavily entwined with technology. Machine learning, data mining and AI in particular are hot topics in the information and technology spheres right now, and each of the aforementioned sectors has their own way of interacting with it.
The AHRC-funded project ‘Patterns in Practice’ - which is led by Senior Lecturer Dr Jo Bates from Sheffield’s Information School, working with Professor Helen Kennedy from the University’s Department of Sociological Studies and Dr Erinma Ochu, Associate Professor of Immersive Media at UWE Bristol, as well as Dr Itzelle Medina Perea and Dr Monika Fratczak working as postdoctoral researchers in the Information School - aims to explore how the beliefs, values and feelings of practitioners from the pharmaceutical, education and arts sectors shape how they engage with machine learning. The project began in 2021, and will run for 2.5 years total. It is currently in its data collection stage, with Dr Medina Perea covering the pharmaceutical and education sectors and Dr Fratczak taking the arts.
“In the three sectors, practitioners are using machine learning in very different ways”, says Dr Medina Perea. The research team are conducting interviews with people active in each field, discussing how they interact with machine learning in their work, following these up with focus groups to facilitate discussion between the individuals. There are then ‘observations’ - which take very different forms in each sector, reflecting the variety of activities undertaken in the disparate disciplines. Practitioners are also invited to contribute diary entries on their everyday experiences. The idea is to understand what is happening in each sector individually, and then find connections and overlapping issues arising between them.
“In the pharmaceutical case, practitioners use machine learning in the form of predictive models during the drug discovery process”, explains Dr Medina Perea. Interviews and focus groups in this area have been conducted with managers, as well as both medicinal and computational chemists, looking at their work in ‘dry labs’ (i.e. work with computer models), as well as how this drives work in traditional ‘wet labs’.
The research on the pharmaceutical industry has been going on for the longest, with some preliminary results already shared at the EASST and Data Power conferences. The findings show a desire at management level to increase the use of predictive models, as this is considered to be a future worth investing in. Dr Medina Perea reports that managers acknowledge scepticism amongst some chemists and while this was quite normal and valued, outright resistance was not
“Our participants tend to distance themselves from the hype around machine learning. They don’t feel it's sustainable; there’s a lot of excitement, but they don’t want to be part of it.”
“Some people we spoke to feel excited about using predictive models”, says Dr Medina Perea, “but at the same time they feel pressure to keep the promise alive because there has been so much investment in it.”
As with any conversation about AI, one criticism of predictive models is centred on a concern that machines might take jobs away from humans. However, the research in this area so far shows this to not be the case, with chemists feeling that their creativity and uniquely human insights will keep their positions safe.
“Medicinal chemists are confident that their work is not at risk, and they feel assured that the skills they have are valuable”, says Dr Medina Perea.
“Our participants tend to distance themselves from the hype around machine learning”, Dr Medina Perea adds. “They don’t feel it's sustainable; there’s a lot of excitement, but they don’t want to be part of it.”
This links into the next domain.
“In the arts, we’re looking at different ways of using machine learning; for example as a tool, or as a topic”, explains Dr Fratczak of her research domain within the project. The team are engaging with performance artists, musicians and combinations of these, as well as art curators and commissioners, but the largest proportion of their participants are visual artists. Some of these creatives are using data mining techniques and machine learning to produce art completely unrelated to the concept of machine learning itself, whilst others create art which in some way comments on or even criticises machine learning and its impacts on society. Now more than ever this feels like a prescient time to be looking into this area, as the merits and potential pitfalls of ‘AI art’ are being debated heavily in the public discourse right now, particularly how they impact what can and cannot fall into the very definition of ‘art’.
”The goal is not only research, but understanding why machine learning matters to society from a critical perspective.”
The research into the Higher Education sector is the last to fully kick off, but involves practitioners who use predictive learning analytics as part of a teaching practice.
As if to illustrate the commonalities between the domains of pharmacology and the arts before the research has even reached that stage of comparison, Sheffield’s Festival of the Mind brought about an unlikely interdisciplinary collaboration when the team wanted to share insights from the project.
“Recently a lot of academic research has involved collaboration with animators and visual artists, which is a really valuable way of sharing findings”, says Dr Medina Perea, “but we wanted to do something different.” Working with an artist seemed like a good way to drum up some interest in the topic at hand, whilst matching Festival of the Mind’s scope for bringing research to life in innovative ways.
In conversation with co-investigator Dr Ochu from UWE, Dr Medina Perea realised that if you want to collaborate with an artist to disseminate your research insights to a wider audience, you shouldn’t just get them to translate the results into a new format - you want them to bring their own ideas and creativity into the process, engaging with the materials in a completely novel way, without too much of your direction as a researcher. The idea of collaborating with a poet came up as a particularly unusual and interesting way of doing this, and after approaching Sheffield film makers ENON films, the name Otis Mensah was suggested.
“The collaboration was based on finding common interests, coming from a place of respect for the role of the artist and the role of the research team”
Otis Mensah is an abstract poet and alternative hip-hop and spoken word artist from Sheffield. Under Magid Magid’s tenure as Lord Mayor of Sheffield, Mensah was named Sheffield’s Poet Laureate, and has gone from strength to strength, with a string of underground EPs (including 2022’s self-produced effort Things I Should Have Said a Year Ago) landing him a prestigious performance slot at Glastonbury Festival and opening for the likes of Benjamin Zepheniah and The Sugarhill Gang, to name just a few of his achievements.
 |
| Filming 'Data/Opium' |
The research team approached Mensah about writing a poem responding to the initial findings from the pharmaceutical case in the Patterns in Practice project, and he agreed.
“He’s from Sheffield, he was the first Poet Laureate of the city, and knowing some of his work we knew he would come from a critical perspective”, says Dr Medina Perea on why Otis Mensah seemed the perfect choice of collaborator.
Mensah didn’t undertake the work lightly; he had lots of questions, asking the team if they wanted him to take a stance, praising what machine learning is doing for the pharmaceutical industry, or if he was more free to make his own statement.
“The collaboration was based on finding common interests, coming from a place of respect for the role of the artist and the role of the research team”, explains Dr Medina Perea. “We tried to find a balance between artistic freedom and sharing insights from our work, also bearing in mind the ethics of sharing findings that are based on confidential interviews with people.”
Once the poem, ‘Data/Opium’, was written, Mensah performed it on camera, captured by ENON Films, with additional background sounds and imagery relating to the topic completing the artistic statement.
One audience member at the event asked Mensah how he found the collaboration, and he explained that to him it wasn’t just about responding to the findings, but also reflecting on his own experiences as someone in the music industry and how machine learning is impacting that sphere. This scope for thinking beyond the confines of the project is exactly what the team hoped to achieve through this collaboration.
Read a Twitter thread about the event here
The video was shown at the event and then displayed throughout the Festival. It is currently on display in the foyer of the Information School, with plans in motion to share it at other events and locations. Otis Mensah himself shared it on his social media channels, immediately reaching far more people than it had before, showing the power of interdisciplinary work in bringing in people from outside your immediate community.
 |
| The team at Festival of the Mind |
This isn’t the end of the non-academic engagement, either, with the team planning a project with an artist in residence later in 2023.
Comments