Bringing STEM to LIFE — an Aspirational Vision for Undergraduate STEM in 2040

The National Academies requested ideas for what undergraduate STEM should look like in 2040. My thoughts (999 words) are below. (Did this quickly the evening it was due so don’t tell me about typos).

Unfortunately, data science curricula teach “Hello World” before prompting students to reflect on their responsibilities in the world.

A biologist/high-school teacher’s aspirations for life-long learning that is inclusive, focused on need, and ethical — bringing STEM to LIFE.

1. Teach how to learn STEM

Problem: as STEM advances more rapidly than ever, students and educators will find it increasingly hard to keep up.

The National Bureau of Economic Research (NBER) concluded in its report STEM Careers and the Changing Skill Requirements of Work [1] that STEM workers are perceived to be in short supply because “new technologies replace the skills and tasks originally learned by older graduates, causing them to experience flatter wage growth and eventually exit the STEM workforce. Faster technological progress creates a greater sense of shortage, but it is the new STEM skills that are scarce, not the workers themselves.” STEM is racing ahead and our current approach to learning must almost certainly reach a failure point. In the life sciences for example, faculty believe computational skills should be included in their teaching but only about one-third of them achieve this citing their own lack of training as the bottleneck [2]. Alarmingly, new faculty better trained in these skills are less likely than senior faculty to bring them into the classroom and researchers across biological disciplines see training in computation as their most unmet need [3].

Idea: Train students early to be self-directed learners and build communities of practice that enable life-long learning.

Course-based undergraduate research experiences (CUREs) in STEM have been transformative in improving retention and graduation [4]. The next step is to go beyond teaching technical skills to teaching learning skills essential to STEM. We need to supplement the curriculum with course work that emphasizes evidence-based learning strategies that cultivate students as self-directed learners [5]. Doing so could allow us to better support the skilled STEM workforce we have already invested in. We must also create and institutionalize community of practice frameworks for learning post-terminal degree. A proposal for achieving these aims was a winning entry in the NSF 2026 Idea Machine competition [6].

2. Learn what STEM students need

Problem: STEM faces serious diversity problems — minoritized ethnicities are underrepresented at all stages of the pipeline. STEM has also failed to serve the skilled technical workforce which includes underserved rural students, veterans, and millions of Americans that find themselves un- and under-employed in our changing economy.

Organizations including the National Academies have extensively documented STEM diversity problems in reports such as Expanding Underrepresented Minority Participation — America’s Science and Technology Talent at the Crossroads [7]. While it’s impossible to thoroughly summarize the large body of work on the subject, clearly a system of education built intentionally and unintentionally to suit the needs of one group must be redesigned to accommodate the needs of every group.

Idea: Take an “all of the data” approach to deliver ethically microtargeted curricula spanning institutions.

Today, companies have few technical limitations in collecting any and all data about customers to deliver microtargeted goods and services. Centered on ethics and inclusivity, we should develop technologies that aggregate data about student learning and career outcomes to deliver the most personalized curriculum possible. As our ability to model and predict outcomes across the entire diversity of students grows, it must be matched with an equally powerful ability to intervene. By 2040, it will be difficult (and unethical) to think about universities as bounded by walls. When an institution does not offer the coursework or guidance an individual student needs in a given area, they can be guided by AI solutions towards supplementary open educational resources including online instruction from other universities, career centers, or industry. This entire approach can be executed ethically if it is transparent and governed by the voices and interests of the most vulnerable, marginalized, and minoritized [8] students.

3. Thinking (computationally and ethically) before doing

Problem: We know computational skills are more important than ever, but computational thinking and data ethics aren’t making it into the classroom early on.

The NAS report Data Science for Undergraduates Opportunities and Options [9] highlights the deficiencies we still are grappling with in matching the power and promise of data science with its disastrous capacity for bias and abuse. The major push for bringing data science into the curriculum at the earliest ages focusing on teaching how to code. Unfortunately, data science curricula teach “Hello World” before prompting students to reflect on their responsibilities in the world. This aspiration for computational preciousness resembles the desire to teach second languages as early as possible (and universities have suggested learning to code can substitute for language requirements [10]). As computers and artificial intelligence becomes more important in our lives, we need to account for ethical implications that are far more serious than learning to speak Spanish of French.

Idea: Prioritize teaching computational thinking and ethics before data science skills

We can enrich development of computational skills by shifting emphasis from technical capabilities and towards the conceptual and ethical concerns. Computational thinking in STEM focuses on decomposing traditional STEM educational problems (e.g. deriving rules of Newtonian mechanics, modeling chemical reactions) into ways that open them to computational exploration [11]. Students can explore this approach to thinking and related ethics before they type their first line of code. Focus on metacognition (i.e. how do I integrate this with my prior knowledge?) and personalization (i.e. how do the ethics of my community and profession govern the use of knowledge?) yield richer outcomes beyond raw technical skill.

Brining STEM to LIFE will make STEM more relevant and guide students towards an ethical and inclusive world I’d like to see.

References

1. NEBR. 2019. https://doi.org/10.3386/w25065

2. Williams et. al. PLOS One. 2019. https://doi.org/10.1371/journal.pone.0224288

3. Barone et. al. PLOS C Bio. 2017. https://doi.org/10.1371/journal.pcbi.1005755

4. Rodenbusch et. al. CBE L Sci Education. 2016. https://doi.org/10.1187/cbe.16-03-0117

5. How Learning Works. 2010. ISBN 9781119308683

6. NSF. 2020. https://www.nsf.gov/news/special_reports/nsf2026ideamachine/index.jsp

7. NAS. 2011. https://doi.org/10.17226/12984

8. J. Negro Education. 2013. https://doi.org/10.7709/jnegroeducation.82.2.0184

9. NAS. 2018. https://doi.org/10.17226/25104

10. Inside Higher Education 2017. https://www.insidehighered.com/admissions/article/2017/11/27/should-computer-science-fulfill-foreign-language-admissions

11. NRC. 2011. https://doi.org/10.17226/13170