Reinventing Scientific Talent
My submission to NSF 2026 Idea Machine
What is the compelling question or challenge?
As the pace of scientific discovery accelerates exponentially, how will scientists, educators, and other STEM workforce professionals meet the demand for career-long learning?
What do we know now about this Big Idea and what are the key research questions we need to address?
NSF has made significant investments in developing a larger and more diverse pool of STEM talent. However, many of the practices of STEM preparation implicitly assume that professional preparation ends at a terminal degree. While being a life-long learner is an academic ideal, in practice this aspiration has not alleviated many of the skill gaps we see in science and science education today. No matter how much the training behind a degree is updated and improved, the shelf-life of skills is getting shorter in a rapidly changing, more interdisciplinary world. We must learn how to combine the deep and slowly acquired expertise of a degree with novel approaches to training and learning that can enable individuals in the STEM workforce to refresh and reinvent themselves over the course of their careers.
In biology, at least the last 10 years have been a transition into a “Big Data” science. While the technological obstacles of this transition were significant, the highest barrier continues to be building computational and data skills within the science workforce. Barone et.al. showed that currently, the top 3 of 13 unmet needs of investigators funded by NSF BIO were all forms of computational training. Computational and data science skills are a case study of how a discipline can quickly become dependent on methodologies only wielded successfully by the hands of a few. Disruptive but critical methodologies fuel progress, but at the cost of enlarging the skill gap scientist face. Will scientists in the 21st century have time to wait for transformative approaches to make it into the classroom? How will experienced researchers be on-boarded when skills emerge from outside of their discipline, and when the luxury of sabbatical study is inaccessible?
Only time (helped along by an interested and diverse community) can pinpoint all the relevant research questions about how we will learn what we have yet to discover. Casual observation suggests two exemplar questions:
1) What is the best way to disseminate a new skill into a STEM community of practice?
Again using computational skills in biology as an example, Feldon et.al. frames the current problem. Their analysis of 294 PhD students in life sciences from 53 US institutions concludes that despite more than $28 million in investment in training in the form of boot camps and other short formats, student outcomes don’t demonstrate sustained impact. While this claim may seem shocking, many of the conclusions of this paper would have been anticipated by experts in cognitive science and andragogy. How did we come to invest so much in a process that may have produced very little? We need systematic dissemination of evidence-based learning approaches and implementation science to address impacting the STEM workforce at scale.
2) How can curriculum for rapidly evolving methodologies be developed and delivered at scale?
It is plausible that there is one or a few optimal ways to teach a particular subject. Yet countless hours are spent creating and recreating educational resources that span a continuum — from low to high quality — in terms of their validity, accessibility, and assessability. How can we create the circumstance for the development of authoritative resources for the dissemination of skills? For example, the US Agency for Healthcare Research and Quality’s National Guidelines Clearinghouse was a single resource for physicians to obtain authoritative treatment protocols. An educational repository created by communities of practice in the sciences could save thousands of hours of curriculum development as new methods emerge. Interestingly, The Carpentries (Software Carpentry and Data Carpentry) (Teal et.al. 2015) presents the only global example of how a volunteer organization (of more than 1,600 researchers serving as instructors) can do this at scale. The Carpentries has reached more than 38,000 researchers in 46 countries and assessment data supports the conclusion they are achieving sustainable impact.
Why does it matter? What scientific discoveries, innovations, and desired societal outcomes might result from investment in this area?
Revolutionary approaches (e.g. CRISPR in biology, deep learning in data science) may appear and permeate a discipline in under a decade. Skill transformation and turnover is increasingly common and more rapid. It is uninformative therefore to guess at what scientific discoveries will be possible when the STEM workforce is more rapidly and more systematically able to acquire new skills. As a rule however, education well done is always a sound investment. Workforce re-training outside of STEM (e.g. individuals with jobs in fossil-fuel industries transitioning to jobs in renewable energy) has been common for some time now. Any illusion that STEM careers should be immune to these transitions should be dispelled.
As a society, we will benefit most if we recognize that scientists will not graduate with all the skills they need. As science becomes more interdisciplinary, the skill gap will make it less likely that mechanisms of informal learning associated with a STEM career will match the rapid rate of change. Skill gaps also amplify existing disparities for researchers from underrepresented groups and at under-resourced institutions. New and systematic approaches to career-long learning would create an adaptable STEM workforce able to respond to complex and rapidly changing scientific challenges.
If we invest in this area, what would success look like?
The research questions relevant to the educational problems outlined here and elsewhere are diverse. Impact however could probably be measured by a few convergent metrics. Success would look like:
1) STEM professionals are able to clearly identify with a community of practice — individuals within a discipline or related disciplines who share a common set of skills, approaches, motivations, and interest.
2) STEM communities of practice have well-defined ways to identify relevant skills and information. As new skills become relevant to the community of practice, there are also well-defined ways for individuals to be on-boarded to skills or to acquire the benefit of these skills through collaboration.
3) Training on skills and knowledge within a STEM community of practice operates as a Commons — a shared pool of resources managed for collective and individual good. Individuals are able to make informed choices about what skills they would benefit from acquiring, since expertise can only be achieved in a few areas. Skill gaps arising from inequities (e.g. lack of access to resources such as time and funding) or disadvantages (e.g. additional obstacles faced by underrepresented groups in STEM) should be addressed by the Common’s governance and access policies and principles.
4) Ultimately, all STEM professionals in all sectors (academic, professional, government, etc.) have a clear understanding of how to go about the process of acquiring new skills and are supported in their efforts by community and employers who understand the value of re-training.
Why is this the right time to invest in this area?
“The illiterate of the 21st century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn.” Alvin Toffler
The amount of information (and hopefully also knowledge) being produced is doubling at an inconceivable rate; things are changing and changing faster. The skills students need look less and less like the skills their advisors and educators have, and here too the amount of difference is increasing at ever shorter timescales. Along with these changes come enormous concerns — everything from mere missed opportunities, to serious ethical challenges. If NSF is looking to explore truly Big Ideas, now is the right time to transform the education of scientists and STEM professionals after their formal training. While such an effort will not be sustainable if it only funded and led by NSF, they are uniquely positioned to catalyze its implementation.
References
Tracy K. Teal, Karen A. Cranston, Hilmar Lapp, Ethan White, Greg Wilson, Karthik Ram, and Aleksandra Pawlik. Data Carpentry: Workshops to Increase Data Literacy for Researchers” International Journal of Digital Curation, 2015.
David F. Feldon, Soojeong Jeong, James Peugh, Josipa Roksa, Cathy Maahs-Fladung, Alok Shenoy, Michael Oliva. Null effects of boot camps for PhD students. Proceedings of the National Academy of Sciences Sep 2017, 114 (37) 9854–9858.
Barone L, Williams J, Micklos D. Unmet needs for analyzing biological big data: A survey of 704 NSF principal investigators. PLOS Computational Biology 2017, 13(10): e1005755.
