This semester I am having my first experience with “AI-native” course design: I am working in close partnership with generative artificial intelligence (GenAI) as I design from scratch a new graduate-level course on educational theory and practice. As a result, my students have learned about behaviorism—which equates learning to observable and reinforceable behavioral changes—not through pre-class reading but via an interactive, web-based game that models behavioral reinforcement. In the classroom, I’ve incorporated several realistic case scenarios generated to reflect differing levels of difficulty or ambiguity, enhancing student discussions on learning theories and instructional approaches. I have been able to do things in my teaching that I’ve never been able to do before.
AI technologies, including GenAI, are transforming the neuroscience training landscape. They are challenging existing notions of “intellectual ownership,” changing the skills that students must learn and expanding the tools we use to teach, learn and conduct research. Graduate programs must adapt. Although it is understandable that many would feel overwhelmed as we face a seismic shift in our professional and educational practices, I want to encourage educators to view it as a time of opportunity. Just as technologies such as CRISPR, optogenetics and spatial transcriptomics catalyzed exciting new insights and lines of inquiry, AI is doing the same in both the research laboratory and classroom. In disrupting the status quo, AI presents an opportunity to critically reassess and redefine what neuroscience graduate training should look like.
As we reenvision graduate training, we can furthermore potentially address in novel and innovative ways long-standing training challenges—including how to train graduate students in both academic inquiry and workplace-relevant skills, and how to modernize training curricula to reflect current pedagogical best practices. As a training community that includes disciplinary expertise in AI, learning and cognition, neuroscience educators are uniquely positioned to lead the charge to reinvent graduate training in a way that maximizes the advantages and minimizes the disadvantages of AI in the realms of education, research and workforce development.
T
o begin, we must, as a field, collaboratively set forth—and commit to—a set of shared values to govern human-AI scholarly and research endeavors. To date, many conversations around GenAI and academic integrity in higher education focus on AI detection tools, even though these tools are unreliable and easily defeated by the ever-expanding capabilities of AI tools such as “humanizers.” Instead, let’s acknowledge that responsible AI use is an important aspect of academic and scientific rigor that must be taught. Many educational tenets to improve scientific rigor—including critical thinking, experimental design and metacognition—are similarly critical to prepare trainees to engage with AI responsibly and ethically. Recent results from a survey commissioned by The Transmitter on the future of neuroscience training prescribed “renewed emphasis on critical thinking and experimental design” to counter students’ increasing interests in applying AI to analyze results rather than applying their own expertise. Strengthening critical thinking and metacognition is important to counter “illusions of understanding” that can arise from the use of AI tools.We can leverage existing momentum around training in principles of rigorous research to identify opportunities for the integration of AI skills and practice. For example, as part of an educational grant through the U.S. National Institute of Neurological Disorders and Stroke (NINDS) Community for Rigor (C4R) initiative, my colleague Xiuqi (Jade) Li and I developed a metacognitive framework, AiMS, to scaffold experimental design instruction through structured reflection. Through interactive worksheets and a text-based narrative, we guide trainees through a sample neuroanatomical tracing experiment, prompting them to assess the relative strengths and caveats of aspects of the experimental system and consider possible experimental outcomes. Against this backdrop, we further address the relevance of GenAI to experimental design. We offer strategies to critically engage GenAI as a metacognitive partner—adopting the role of brainstorming partner, constructive reviewer or organizing assistant—as trainees use the framework.
As academic scientists and the broader neuroscience workforce define standards for responsible human-AI collaboration, we must also critically question and redefine the overall purpose and desired outcomes for neuroscience graduate training. What skills and knowledge—both in service of human cognition and effective human-AI collaboration—are essential for young neuroscientists? What skills must we incorporate into graduate training to adequately prepare trainees to enter an AI-integrated workforce? A vision of neuroscience training outlined a decade ago emphasized training students for diverse careers. Yet neuroscience graduate training remains misaligned to the neuroscience workforce. The rapid adoption of AI into various careers will accelerate this misalignment if graduate training programs do not provide opportunities for students to develop emergent and professionally aligned AI-relevant skills. We have an opportunity to creatively reimagine graduate training to better align the competencies developed during graduate school to the skill sets required for the range of careers—within and beyond academia.
Graduate programs should seek to identify, catalog and prioritize desired training outcomes —including AI-relevant outcomes—that ideally support both academic research excellence and preparation for at least a subset of training-relevant careers. These conversations should occur across graduate training programs; let AI be a catalyst for further connection and collaboration. For example, colleagues at Harvard Medical School, Morehouse School of Medicine and Spelman College and I are leveraging an existing cross-institutional neuroscience education partnership to organize a joint summit on AI in graduate education.
Finally, as we clarify what graduate training programs aim to achieve, we must equip faculty with the knowledge, skills and tools to effectively integrate AI into the curriculum and research laboratory in ways that enhance learning and support AI-related training outcomes. The COVID-19 pandemic demonstrated that graduate programs can adapt to new instructional constraints dramatically and quickly. We must treat the current moment as urgent, mobilizing resources to prioritize faculty development and curriculum and course support, as well as technology access and support. Modernizing our curricula and teaching methods for AI-native learning environments presents an opportunity to enact long-needed educational reform more broadly, as evidenced by the fact that less effective, passive forms of instruction, such as lecturing, continue to dominate STEM education.
Effective “active learning” approaches are also important for faculty development efforts. Specifically, providing educators with hands-on practice with AI is important to develop their knowledge of, and comfort with, the technology. For example, in a classroom context, I have provided biomedical educators with guided instruction in effective prompt engineering for instructional design that significantly transformed the quality and richness of the educators’ prompts to AI. I further developed educators’ insights on GenAI through metacognitive reflection exercises.
Though the field is moving fast, several resources to guide these efforts already exist. The Council of Graduate Schools, for example, published proceedings from a global summit on AI and graduate education held last year, which offers guidance on how to integrate AI into existing practices. The Society for Neuroscience (SfN) and journals such as eNeuro and Neuron have established rules emphasizing disclosure of AI use and human oversight of AI. Although none yet integrate AI, several science graduate training competency frameworks can guide discussions on what skills and knowledge neuroscience students should be gaining during graduate school. The SfN Neuroscience Training Committee’s pre-AI core competencies for graduate neuroscience training provide a discipline-specific foundation for these conversations.
The current moment offers an opportunity to drive broad educational reform. Together, we can harness the AI-induced urgency and energy to creatively reinvent neuroscience graduate training for the future.
