A Health Tech convention: Opening up the shiny box
By: Francisca Grommé, Justien Dingelstad, Eliana Bergamin
Structural issues in healthcare are partly being addressed by start-ups and technology companies that offer AI-enabled solutions. Such companies propose practical interventions to mediate personnel shortages and a growing demand for care. For instance, by using natural language processing to automate nurses’ administrative tasks, or by predicting the risk that an elderly person will not drink enough water. The diversity of start-ups and products however does raise questions of coordination: can such diverse, and often fragile, market solutions contribute to structural issues? And how can the companies that offer them be supported in doing this? To start answering these questions we need to know more about how this market is steered in practice, who is doing the steering, and how the actors in the field relate to each other. While the AI market can appear like a shiny and glittery box full of promises, it is equally important to investigate what lies within it.
Tech conventions are a unique opportunity to start developing such an overarching insight into an otherwise seemingly opaque field of activity. In this blog post, we will therefore narrate our experience at a health and ICT convention, which was an elaborate showcase of the latest, sometimes futuristic, items of technology. It offered visitors (among whom health practitioners and policy makers) the opportunity to get acquainted with new tech inventions, for example through showcasing virtual reality applications that visitors can touch and experience for themselves. Visitors spent their day(s) walking past and through an overwhelming number of stands, where an abundance of personnel was ready to answer their questions.
Organised in a large, square hall without any windows, it was like a sealed box containing its own reality [see Figure 1]. Within this box, the stands presented themselves as smaller boxes with each in turn contained their own small reality. All of this created an overflowing boxed-off world full of colour, sound, and smell. Sealed-off and sensory-rich experiences — social studies of conferences show — are exactly what contributes to realising a sense of unity in an otherwise fragmented field (MacDonald, 2010). What is more, the elaborateness of such events can justify existing priorities and relations in the field. In this way, it is argued, events like tech conventions bring a reality into being that does not necessarily exist apart from it. In other words: they are ‘performative’. How AI in healthcare is steered and who takes part in this steering is therefore not set in stone in advance of the event. Instead, the events at a tech convention contribute to the organisation of the field (Hajer and Versteeg, 2010; Grommé et al., 2021). Relevant mechanisms are the degree of ‘spectacle’ that is operated, among which material displays and demonstrations, and how the event is orchestrated to present and draw together actors in predesigned ways (MacDonald, 2010; cf. Goffman, 1959; cf. Möllers, 2016).
Field trip
Following this line of thinking, trade fairs in ICT and health offer an opportunity to study the setting and everyday interactions that take part in shaping the role of AI in this field, along with the roles and responsibilities of public and private actors. We therefore set out for a one-day visit to find out what we could learn from the event, knowing that our observations would be exploratory at this stage. Two of the authors registered for the convention, followed the preparatory and follow-up e-mails, and visited for a day [See Figure 2]. In advance we agreed on observing the performative elements of the fair, as well as the specifics of the products that we on display. To some extent our aims were similar to those of many visitors: we also visited to extend our network in the field. In our interactions, we presented ourselves as academic researchers. Following such introductions, we often experienced practiced sales talks, after which we attempted to engage in a more in-depth conversation. Besides this, we took notes on our observations, experiences, as well as photographs. We reconvened together during our visit, combing our observations, and we reflected upon it afterwards as well.
A guided tour: the power of orchestration
At the convention, there were several tours available, organized by the convention organization. These tours were thematic tours at which visitors were guided from stand to stand. There were for example tours about robotics, data exchange, and care platforms. An overview of the tours was provided beforehand through an online platform where visitors could sign up. At first glance, all these tours seemed equal. They were listed in a chronological timetable and accessible for all visitors without prerequisites. However, a closer look shows how one tour had a prominent position amongst other tours. This was ‘The’ Artificial Intelligence Tour. It was the only one labelled ‘the’ tour and the only tour held twice a day every day. It was also the only tour highlighted in the program sent to visitors before the event. Besides the prominent positioning of the tour, its importance also stemmed from the fact that it was organized by the Netherlands AI coalition (NL AIC) [See Figure 3]. This is a public-private partnership between all levels of government, business, research organizations and civil society, that aims to put the Netherlands in a front-runner position in terms of AI knowledge and applications (nlaic.com, 2022). It has a national research agenda, organizes AI living labs around the country and connects AI developments and initiatives. Because of its composition and activities, it has an authoritative position in the Dutch AI ecosystem, which it actively performed during the convention.
Having signed up for the tour, we first received a short introduction to the convention, after which a consultant guided us past ten different organizational stands. At every stand an employee or representative explained what the respective organization did with AI. This ranged from automation of administrative tasks to sentiment analysis for psychological care, to post-graduate AI courses, to legal support for AI in health. There was no explanation given for why these companies were selected. This is interesting, as the curation has dramaturgical importance. First, the stands that were selected for the tour were (literally) put in the spotlight, whilst the other stands remained out of sight. This means the NL AIC coalition, as director of the tour, had the power to orchestrate what the visitor — or spectator — got to see and what not. Second, this power was reinforced by the authoritative position of the NL AIC. If a stand had made it into the tour, it had passed the selection of the NL AIC. This seal of approval created a hierarchy between the stands that were inside and outside the tour: inside meant quality guaranteed by the coalition, outside meant quality to be determined by the visitor. Lastly, a curated tour in a convention setting is of extra importance in terms of dramaturgical positioning, since visitors are limited to their body (MacDonald, 2010). The body can only be at one place at one time and process a limited amount of information. In a space that is so overwhelming in terms of colour, sound and smell, a visitor can simply not take in everything that is out there to experience, and a tour makes an authoritative selection.
Chipsoft and start-ups: the spectacle of power
As part of the tour, we were introduced to a start-up promoting an AI-enabled system for care workers. The system presented a dashboard that would inform the care worker about the patient’s status just before consultation. For instance, by providing a visual of a sentiment analysis based on consultation notesv [See Figure 4]. After the tour, we approached the stand and met the data scientist who designed the system. “What is the status of this product?” we asked, “are organisations using it?”. The dashboard was still only a demo but, as the data scientist added, several care professionals at this conference commented that they think it can be useful. However, to actually be implemented he would need access to electronic patient file data (EPD) stored in the national EPD system. We asked whether this would be possible, and in response, the data scientist pointed across the central pathway to a neighbouring stand: Chipsoft. It depends on “them” because they hold the patient file data.
This short exchange first demonstrates how the relationship with health practitioners takes shape through the demo and is performed in subsequent conversations (“they think it could be useful”). As it is the case with many start-ups, a product’s potential takes shape through a range of interactions
(Callon, Méadel, Rabeharisoa, 2002), among which these short exchanges. In this sense, this is also part of a spectacle constantly performed at this convention that, in this case, brings into being a relationship between the technology and potential users.
Another part of the spectacle is the pointing of the finger to another participant. This denotes another, hierarchical, relationship relevant to steering the direction of AI in healthcare: between this start-up and Chipsoft. With 70% of hospitals as a client, Chipsoft is the dominant market player in providing the IT infrastructure (HiX) for sharing patient files (M&I/Partners, 2022). In order to access patient file data needed to develop AI-enabled health applications, almost every start-up would need to collaborate with this company. At the stand, a Chipsoft employee confirms that he is indeed very often approached by start-ups (“at this moment I have 6100 unread e-mails”). However, many of these ideas are not viable because they lack integration with practice or with hospitals’ IT infrastructures. In addition, AI is not the company’s “core business,” Chipsoft’s employee noted.
Chipsoft’s position is also expressed in the placement and design of the stand. The company is symbolically positioned in the centre of the conference room [See figure 5], at a position more central than its main competitor, Epic. This is also a sensory experience; Chipsoft’s relevance did not need explanation. In terms of its design, large, red, ‘columns’ hanging from the ceiling imitated the columns from information systems diagrams, thereby symbolising its central role in health data and innovation.
Conclusion
The actors presented by the NL AI coalition; the start-up and its potential users; and the dependence of start-ups on Chipsoft are only three relevant relationships that partly take shape during the convention through spectacle and orchestration. In a field that develops ad hoc and in which participants change quickly, the everyday performances that we witnessed contribute to its coordination. Overall, we can understand the instances discussed in this blog as the realisation and justification of an overarching norm: market actors in the ICT health sector will largely steer themselves when it comes to the development and implementation of AI, with public and third sector actors predominantly positioned on the sideline, watching the spectacle go on. Considering that much of the funds spent on purchasing these innovations is public, the naturalisation of this state of affairs deserves attention.
The scope of this one-day fieldwork visit is of course limited. Across the year, a number of health conferences is organised by public sector actors and provide opportunities for public and private actors to engage, and AI-enabled technologies are also on their agendas. Moreover, university and other public sector researchers also develop AI-enabled technologies, often presented at academic conferences. Yet, these events seldom allow for the presentation of finished products to a diverse, professional audience.
In conclusion, then, tech conventions could play relevant roles in openly discussing, negotiating and questioning AI-enabled technologies. They provide insight into an otherwise rather opaque sector. What we are wondering is, however, if these settings can become accessible to other types of thinking, for non-scripted interaction. In our experience, the convention’s spectacle and orchestration took many aspects regarding the role of AI in healthcare for granted. While it was interesting to notice all of the different stakeholders coming together, as visitors we often felt like guided actors that were given more or less passive roles in the performance. In the shiny and glittery box of the spectacle of AI, is there a place where visitors are not blinded by its brightness? A place where to slow down and reflect? Critically analysing this, using a performativity perspective, can help putting these assumptions up for discussion. It is conceivable that conventions could provide more of such spaces, in addition to providing a place for market actors to promote their products.
Sources
Callon, M., Méadel, C., & Rabeharisoa, V. (2002). The Economy of Qualities. Economy and Society, 31(2), 194–217. https://doi.org/10.1080/03085140220123126
Goffman, E. (1959). The presentation of self in everyday life. Garden City, New York: Doubleday.
Grommé, F., Cakici, B., & Takala, V. (2021). Statistician Subjects: Differentiating and Defending. In E. Ruppert & S. Scheel (Eds.), Data Practices: Making up a European People (pp. 237–268). Goldsmiths University Press. https://www.gold.ac.uk/media/documents-by-section/about-us/goldsmiths-press/publications/Data-Practices.pdf
Hajer, M., & Versteeg, W. (2005). Performing Governance Through Networks. European Political Science, 4(3), 340–347. https://doi.org/10.1057/palgrave.eps.2210034
M&I/Partners. (2022). EPD-marktinventarisatie ziekenhuizen 2021: Consolidatie EPD-markt zet door. M&I/Partners.
MacDonald, K. I. (2010). Business, Biodiversity and New ‘Fields’ of Conservation: The World Conservation Congress and the Renegotiation of Organisational Order. Conservation and Society, 8(4), 256–275.
Möllers, N. (2016). Shifting in and Out of Context: Technoscientific Drama as Technology of the Self. Social Studies of Science, 0306312716638951. https://doi.org/10.1177/0306312716638951
nlaic.com (2022). ‘About the NL AI Coalition’, assessed on 30 June 2022, through: https://nlaic.com/en/about-nl-aic/.
