Automation, expectations, and laboratory work. A robot in every lab?
The impact of automation and robots in our labs
March 23, 2016
There is, so it is said, a coming wave of automation, with effects on everything from cancer surgery, to hotel concierge services to self-driving cars. As this most recent story about 550 job-losses through automating financial advice at RBS demonstrates, these issues are now more consistently hitting mainstream media thanks to their tangible real-world effects on people and institutions. This puts automatizing work in science and engineering in a much broader context, and leads me to the question: is everyone talking about the same thing when they talk about the rise in robotics and laboratory automation?
This was the backdrop to my participation at a recent synthetic biology workshop, ‘Automation and Robotics for synthetic biology’. I went with the intention of finding out what automation meant to different people, what they were most excited about, and the problems they believed automation solved. Perhaps surprisingly, researcher-led in-house lab automation (within Universities) was seen as ‘old school’ in the opening presentation. Subsequent speakers, in contrast, envisioned the deployment of robots in every lab and focused on the challenge of changing the culture of scientists as labs shift to a new paradigm of automated laboratory practice. A move towards Cloud-based and out-sourced lab services was presented as the coming trend in synthetic biology. Most of the workshop participants agreed that the future of biological research and engineering would be more automated, even if the scale, speed and location of those changes were yet to be decided.
Attendees ranged from academia to process management specialists, and included a significant component of industry representatives. The latter were particularly well represented (in comparison to perhaps other kinds of event), because the workshop was embedded in a 'scale up for synthetic biology’ series, scoping out how to move biological research from lab to industry, and the roles for robots and automation in that process. Scale-up and automation were presented as having considerable overlap, as was also recently seen in the recommendations of the UK strategic plan for synthetic biology.
The role for robotics in science, however, covers more ground than academia-industry partnerships. This view was expounded in a presentation by a team working to develop the Robot Scientist. For this group of researchers, the automation of laboratory work is essential to understand the huge complexity of biological systems – they argue that human minds alone will not be sufficient. This is a fundamental problem in biological research, and it is one that is independent of delivering industrial synthetic biology products to market.
There is consensus among all these groups – committed industrialists or not – about the value of automation for increasing productive capacities within labs. What is still not settled is whether increased productivity should be applied to the expansion of experimental space in biological research (i.e. for increasing the capacity of empirical work to understand biological functionality) or if attention should be turned to re-aligning academic laboratory productivity to match up with the aims and expectations of industrial biotechnology. To ask where an academic field or set of technologies is headed is not idle speculation, because making promises about the future power of a technology, including lab automation, legitimises actions in the present.
One example of promises about technological futures legitimising actions in the present is the case of US Defence Department (USDD) budgeting after the Second World War. Applying the ‘Programming, Planning, and Budgeting (PPB)’ approach, the so-called ‘Whiz Kids’ of the RAND Corporation implemented operational research approaches into the USDD. These ‘total systems analyses’ (which consolidated the future budget requirements for all army, navy, and air force operations) put an end to the then established practice of top-brass commanders playing budgetary ‘games’. That is, commanders would first get sign off for the ‘thin edge of the wedge’ and build a number of aircraft, then, subsequently, during the planes’ construction, inform budget holders of the need for additional funding for bases, training, and tankers.
This example demonstrates how the promise of an aircraft for strategic defence planning (a speculation about future technological requirements) had significant implications for actions in the present (the construction of bases in the mid-20th Century). It also shows that to understand how large systems are organised – be that the budget of the US Air Force or capital investments for large-scale Research and Development – requires attention not only to the technical specificity of the system in question, but also to the social dimensions that shape decision-making around technology adoption. This is because when applying the total systems approach the Whiz Kids were not just envisioning a more productive and efficient way of organising the US military but also, in deploying the PPB method, these new entrants to the USDD top brass disrupted established hierarchies of authority. We can also therefore ask: what claims are being made about automation’s future, and for whom do these futures help in the present?
For the participants discussing automation at the recent workshop, robotics represented more than simply an extension of human capacities in the lab. The ‘new paradigm’ mentioned earlier invoked not just a change to the hands-on, craft nature of biological research at the bench, but also a change to the forms that such research can take: “question all existing assumptions” was the plea from SynbiCITE’s head of automation. This paradigm shift in the nature of laboratory work seems to coincide with the scale-of-change envisioned through automation in society more generally. In this respect robotics and automation are not only positioned as a necessary next step for enhancing biological research, its reproducibility, and commercialisation, but also as a major disruptive technology for years to come.
In addition, according to some workshop contributions (especially during Q and A sessions), the desirability of automation in biological research is currently eclipsed by the challenges that remain in understanding complexity in how biology functions. For these contributors, understanding biology is still the main goal in an academic research laboratory. All this is to say automating even simple lab processes is tricky. So why bother? Some of the presenters at the workshop clearly see commercialisation as the way to ensure synthetic biology has a future in times of substantial economic challenge. Others seemed more concerned with the opportunities that automation may bring for expanding the experimental space of biology.
In many ways the workshop raised more questions than answers. For example, how are laboratory users situated in workflows that incorporate automation technologies and commercialisation? Is this future of fully automated, technological biological research an inevitable one? And, do those who resist such changes risk being labelled as modern-day Luddites?
I know from experience working alongside current researchers in labs that automated technologies are everyday tools in their work; they provide data points and help scientists to construct empirical cases of biological function. It is in this potential (for expanding the experimental space of biology) that the lab users I’ve encountered see as having most value when thinking about the adoption of automated workflow technologies. Whether or not such potential becomes entangled with industrialisation and commercialisation efforts does not seem settled, even if documents like the UK Strategic Plan appear to reinforce this narrative. The story of laboratory automation, therefore, has multiple protagonists, each with their own back-story; just as the Whiz Kids were able to plan, programme and budget their away around the military’s budgetary games, will there be a system-wide overhaul of the nature of laboratory work, or will the promises of automation being made now seem quaint to future laboratory users?
The current community of researchers in biology seem well placed to offer perspectives on the issues raised here. I would be grateful for views and comments, so please get in touch if laboratory automation is a relevant issue in your work. Thank you.
Chris Mellingwood is an EPSRC-funded PhD student in Science, Technology and Innovation Studies (STIS) in the School of Social and Political Sciences at the University of Edinburgh.