Current State of Automation in Education
Overdue for modernization
The following is based on the presentation I gave at the inaugural Automated Science Education meeting at SLAS2025. We're now organizing quarterly Zoom meetings on this topic, open to anyone who would like to participate. If you're interested in joining us, please complete this survey and we'll send you calendar invites.Over the last couple of years, I’ve had the unique opportunity to work with dozens of educators around the world working to bring automation into their classrooms. From world-ranked universities to rural public high schools, I have developed a unique vantage point into this space.
Something I’ve found time and again is that most people bringing automation into the classroom are figuring it out in near-isolation. This makes the task much harder, as they lack the benefit of knowing what are common learning objectives, how skills can be mapped to existing learning outcomes, and what’s most valuable for their students to learn. Too many educators are forced to reinvent the wheel, starting from scratch to build something that has already been built by someone else.
Automation skills are career-critical
We’ve reached the tipping point in which over half of life science labs use some form of automation. In most labs, automation is in the form of task-specific automation; liquid handling robots, and microplate handling for high-throughput screening. At the other end of the spectrum are self-driving labs, which combine robotics and AI to design experiments, analyze results, report data, and select and initiate subsequent experiments.
There are a range of skills needed to implement various forms of automation. Technicians need to know how to set up and clean the equipment for experimental runs, scientists need to know how to translate their manual protocols for automation, engineers need to understand the biological necessity underlying the robotic programming, statisticians need to know how the system functions to understand how the data was generated, administrators and executives need to understand the problems solved and the problems left unaddressed by automation. At minimum, everyone needs a basic working knowledge of automation, and individuals in each role benefit greatly from deeper knowledge.
Unfortunately, regardless of which role you look at, the majority of those educated for that role were not taught how to use automation or even how to think about the implementation of automation. Put another way, neither the hands-on skills nor the theory of automation is being taught to most trainees.
Defining automated science education
Automation refers to a broad spectrum of tasks and equipment. There are automatic plate readers, which involve putting a plate on a tray and clicking a mouse a few times. Then there is automated liquid handling, which takes care of your pipetting tasks. And there’s the full-blown self-driving lab. What needs to be taught, and to who? What even qualifies as automated science education (ASE)?
At the interest group meeting, Dr. Josh Kangas broke it down into three main components: robotics, machine learning, and artificial intelligence. The full value is derived only when all three are implemented, but each of these pillars brings significant value to a lab and can be broken into clear corresponding skill sets.
This is a great starting point for defining ASE. Education which includes robotics, machine learning, and artificial intelligence, in part or in combination, falls under the umbrella of ASE. The specific learning objectives are dependent on your students’ current knowledge, skills being sought, and level of proficiency being assessed.
ASE is taught across a wide range of ages and skills
Although overall penetration is low, ASE is currently being taught across a wide range of ages, including high schools, community colleges, universities, and graduate programs. There are even a small number of programs that are approaching this outside of traditional institutional structures, such as the Genspace Break into Biotech program.
High School
It is rare to see ASE at the high school level, but the examples that do exist are exemplary. If you’re looking for general patterns then you’ll find that more public than private high schools are implementing ASE, there is solid geographic distribution (NYC to rural Louisiana), and there are programs like BioBuilder that are attempting to make it easier for high schools to implement skill-based curricula.
Undergraduate
It’s quite uncommon to see ASE at the undergraduate level. Nearly exclusively automation is found in elective coursework for science majors, and not in the core sequence of coursework. It is a big lift for professors to implement, and school administrations are mixed in their level of support.
Graduate
Masters programs are the most common curriculum entry point for universities. The pattern is to start by implementing in a masters program, where you have the benefits of smaller class sizes and a stronger knowledge base among your students. The learnings taken from this are intended to be applied to the undergraduate level, but this transition seems to be a significant hurdle that only some are beginning to overcome.
Community Colleges and Workforce Training Programs
There are a large number of community college programs that are incorporating automation into the core training sequence. This surprises many but is very obvious from the perspective of the community colleges. Their focus is to provide education and training to their local population to meet the needs of local industry. As industries have increasingly incorporated more and more automation, that need has been communicated to the community colleges who have been steadily updating their curriculum. Because of this model, these programs are often very industry-specific, and due to the lack of education standards for automation, it is hard to directly compare programs across different regions.
Community colleges are leading the way; universities are dropping the ball
While community colleges have established reputations for developing hands-on skills, the reputation of universities is for developing theoretical knowledge. The problem is that employers want practical knowledge. Increasingly, it’s not just the employers who are dissatisfied: a recent survey of university (1) graduates points to a scathing review of university education. Of the students surveyed, 85% wish college had prepared them for the workplace, and less than a quarter of students reported having all of the skills they needed for their roles. If you think on that for a moment, nearly half of the students who felt they had all the skills needed still wish that they had been better prepared. Worst of all, overall of the students say that they weren’t prepared for the jobs at all.
This is damning enough, but it’s going to get worse in coming years. Automation and AI are bringing significant changes to life science careers, and while community colleges are talking about universities are going ostrich mode. And that’s a shame because students with four-year degrees are going to be competing with community college students for roles that are largely hired based on skill.
This may sound shocking, but hear me out. I’ve talked before about the impacts of AI and Automation on bio careers, but I didn’t address how it was going to impact education and workforce training.
Today, lab employees today are skill-heavy - so much so that across the industry there are more scientists than laboratory technicians.
From an education perspective, this has meant that the majority of the life sciences workforce have at least a bachelors degree. By some estimates, over half of the total workforce have a masters or doctoral level degree.
AI and Automation mean that fewer senior scientists are needed to plan experiments to be performed, and fewer people are needed to run the experiments. Most of the roles dedicated to running experiments will do so with automated systems. Consequently, the workforce is likely to shift, with a collapse of mid-level roles.
Instead, a smaller number of senior scientists will collaborate with PIs to guide research direction, while a larger number of technicians will operate automated platforms to conduct experiments. Those with bachelors degrees will increasingly be unable to compete for scientist roles, and instead will seek technician roles. Yet they will struggle to land those roles as well, as employers are increasingly hiring based on skills rather than degrees - an area in which universities fall short of community colleges.
Automation needs be integrated into the core curriculum from HS through workforce development
The only true solution is to modernize science education, starting no later than high school. Students of biology today learn more about the methods used 50 years ago than modern methods. The argument that automation undercuts the ability to teach foundational knowledge simply isn’t true. Mathematics courses long since found a balance with calculators; it’s time for the sciences to figure it out too.
The current tendency to reserve automation-based skills for elective coursework simultaneously fails to train a sufficient number of students and undermines the value proposition of paying for education. Students enter programs trusting that the program has the expertise to train them well; it's a betrayal when programs fail them. Even worse are all of the students taking on debt for the promise of improved job opportunity, only to discover after graduating that they weren't taught the requisite job skills.
Denial gets in the way of productive change
This implementation gap persists largely because of widespread denial within educational institutions. Although many educators agree that change needs to happen, they cite various forms of institutional and personal resistance as challenges. Denial comes in many flavors: denying that it’s their job to teach it, denying that it’s feasible to do so; denying that what they really are blocked by is their own lack of knowledge about automation.
But the most impactful denial is this: when you don’t know how to adapt for the sake of your own future, it’s impossible to contemplate how to adapt training for your students futures.
Denial is the nucleus of the snowball of barriers. There are limited training opportunities for educators to learn automation. There are very few teaching resources for automation, like lesson plans and textbooks. Education suppliers sell chemicals for experiments, but they are all packaged for manual workflows.
Education standards are the key log
These systemic educational barriers—from denial to resource limitations—create a complex challenge that no single institution can solve alone. The key to unraveling this tangle is the establishment of comprehensive education standards that can guide implementation while ensuring consistent quality across programs.
In the absence of these standards, developing a new program requires reinventing the wheel—a time-consuming and inefficient process that diverts resources from educational innovation. Standards can provide a blueprint, reducing redundant efforts and promoting more strategic curriculum design.
For employers, education standards would resolve the opacity around training quality, and alleviate the pressure to maintain direct relationships with numerous educational institutions. This transparency particularly benefits non-name brand institutions, which often struggle to demonstrate their educational value without established relational networks. But it’s students who stand to gain this most.
This is how students can finally receive employable skills commensurate to the rising cost of tuition.