Artificial intelligence (AI) is emerging as a transformative force across industries, including the physical sciences. Indeed, physicists have been making fundamental contributions to AI development for decades, as evidenced by the 2024 Nobel Prize in Physics awarded to John Hopfield and Jeffrey Hinton for their research in artificial neural networks (AIP, 2024 ).
According to Insights for 2024: Attitudes towards AI Surveys of research professionals (including leaders and industry researchers) and healthcare clinicians found that while awareness of AI is high, usage remains generally low. Respondents were concerned that AI could spread misinformation, make serious mistakes, and cause chaos in society. At the same time, they were optimistic and foresaw various benefits from AI. Such benefits include accelerating knowledge discovery, rapidly increasing the amount of academic and medical research, and reducing costs for institutions and businesses (Elsevier, 2024 ).
To begin exploring the impact on the physics community, AIP has incorporated AI-specific questions into its annual degree recipient tracking survey. This report presents findings on how recent physics degree holders are engaging with AI, covering both the development of AI and its application in everyday professional activities.
Utilization and development of AI
New physics degree candidates will interact with AI in two different (and sometimes overlapping) ways: developing and training AI models, or using existing AI tools to support their work.
Across all degree levels, respondents were much more likely to regularly report using AI tools than to develop AI models. (Figure 1). New physics PhD holders were twice as likely to develop an AI model in their post-degree position compared to master's and bachelor's degree holders. Routine AI usage increases with degree level. 23% of bachelor's degree holders said they regularly use AI tools at work, compared to 40% of doctoral degree holders.
Bachelor's degree holder
Overall, new physics degree graduates employed in STEM (science, technology, engineering, and mathematics) fields are more likely to regularly use AI tools than those working in non-STEM fields, by 33% and 18%, respectively.Figure 2).
When users were asked about the purposes for which they use AI tools, STEM-employed bachelor's degree holders reported a wider range of uses than non-STEM respondents, with a median of 3.5 regular uses compared to 2.5 times for non-STEM respondents. Regardless of the field of employment, the most common uses for AI tools include: Write, debug, or optimize your code. The only category that non-STEM respondents reported using more frequently than STEM respondents was Generation of ideas, hypotheses, and research questions.
Examples of AI usage for new physics graduates
“Currently, I am working on computer vision models for various applications in the semiconductor industry.”
— Private Sector Applications Engineer
“We are creating AI models that predict the weather and applications that alert others to natural disasters.”
— Solutions Architect at a large technology company
“Development and training of convolutional neural networks for object identification using computer vision”
— Private university researcher
“AI software can help develop success criteria for students with respect to specific learning objectives.”
— High school physics teacher
Ph.D.
Approximately two-fifths of new physics PhDs reported regularly using AI tools in their daily work, regardless of potential permanent positions or postdoctoral appointments (Figure 3).
AI users with potential for permanent positions used AI for a wider range of purposes than postdocs, with a median of four regular uses compared to three for those appointed as postdocs. Similar to physics degree holders, the most common uses of AI tools were: Write, debug, or optimize your codeRegardless of employment type. PhD candidates with potential for permanent employment were more likely to report using AI tools than postdocs. Machine learning model development and Generation of ideas, hypotheses, and research questions.
Examples of AI usage by a new Ph.D. in Physics
“Broadly speaking, I develop AI models to accelerate progress in particle physics.”
— Postdoc at a private institution
“We are developing a camera-based detection model for self-driving cars.”
— Engineer at a private company
“I'm developing large-scale AI models to estimate distances to galaxies, and I've also worked to make the results of these models reliable.”
— Postdoc at a foreign public university
“Using quantum-inspired techniques to develop new AI models and algorithms.”
— Applied researcher at a financial institution
Research method
AIP conducts an enrollment and degree survey each fall, asking all degree-granting physics and astronomy departments in the United States to provide information on the number of students they enroll and the number of recent degrees awarded. At the same time, you will be asked to provide the name and email address of someone who recently completed a bachelor's degree. We then use this degree holder's contact information to conduct a follow-up survey the winter following the year the student earned their degree. The data in this report is based on a survey of physics degree graduates in the 2023-2024 academic year. The findings in this report are based on information from 439 bachelor's, 37 master's, and 263 doctoral students.
Due to an insufficient number of respondents, specific AI applications for physics masters are not included in this report.
References
American Institute of Physics. (2024). AIP congratulates the 2024 Nobel Prize in Physics winners. Where to get it
https://www.aip.org/aip/celebrating-2024-nobel-physics-prize
Elsevier. (2024). Insights for 2024: Attitudes towards AI. Where to get it
https://www.elsevier.com/insights/attitudes-toward-ai
