Physicists and astronomers

Perform mathematical calculations to analyze physical and astronomical data

AI excels at mathematical computation, statistical analysis, and data processing. Modern systems can perform complex calculations, fit models to data, and execute standard analytical pipelines with minimal human intervention, requiring only oversight to verify assumptions and interpret physical meaning.

BLS evidence: They 'do mathematical calculations to analyze physical and astronomical data, such as for new material properties or the existence of planets in distant solar systems.'

Develop computer software to analyze and model data

AI coding assistants can generate substantial portions of scientific software including data analysis pipelines, modeling code, and visualization tools. While domain-specific physics knowledge and validation still require human input, AI can autonomously write much of the implementation with human review of correctness.

BLS evidence: They 'develop computer software to analyze and model data.'

Write proposals and apply for research funding

AI can draft significant portions of grant proposals including literature reviews, methodology sections, and budget justifications based on templates and prior successful proposals. However, crafting compelling narratives, strategic positioning, and addressing reviewer concerns still require human oversight and refinement.

BLS evidence: Physicists and astronomers 'write proposals and apply for research funding.'

Make astronomical observations and collect data on celestial objects

Modern telescopes are increasingly automated and AI can control observations, process images, and flag objects of interest. However, deciding what to observe, adapting to weather and equipment issues, and making judgment calls about data quality still require significant human involvement in the observation process.

BLS evidence: Observational astronomers 'view celestial objects and collect data on them' using ground-based and space-based equipment.

Write scientific papers for publication

AI can draft sections, format references, generate figures, and improve prose, but writing scientific papers requires synthesizing novel findings, contextualizing within literature, making argumentative choices about presentation, and ensuring scientific rigor that demands substantial human authorship and judgment.

BLS evidence: Physicists and astronomers 'write scientific papers for publication.'

Design new scientific equipment such as telescopes and lasers

AI can assist with optical design optimization, simulation, and component specification, but designing novel scientific instruments requires deep understanding of physics constraints, fabrication limitations, and creative problem-solving to achieve unprecedented measurement capabilities that AI cannot yet fully autonomize.

BLS evidence: Physicists and astronomers 'design new scientific equipment, such as telescopes and lasers.'

Develop scientific theories and models to explain natural phenomena

AI can assist with pattern recognition, literature synthesis, and hypothesis generation, but developing novel scientific theories requires creative leaps, deep domain intuition, and judgment about what constitutes meaningful explanation that current AI cannot reliably produce independently.

BLS evidence: Physicists and astronomers 'develop scientific theories and models to explain the properties of the natural world, such as the force of gravity or the formation of subatomic particles.'

Supervise research team members and monitor their progress

Supervising researchers requires mentoring, providing scientific guidance, managing interpersonal dynamics, making personnel decisions, and offering career advice. While AI can track progress metrics and flag issues, the human judgment, relationship-building, and leadership aspects are not automatable.

BLS evidence: Senior astronomers and physicists 'may assign tasks to other team members and monitor their progress.'

Plan and conduct scientific experiments and studies to test theories

AI can help design experimental protocols and optimize parameters, but planning and conducting physical experiments requires hands-on equipment operation, real-time troubleshooting of apparatus, and adaptive decision-making in response to unexpected results that demand human presence and expertise.

BLS evidence: They 'plan and conduct scientific experiments and studies to test theories and discover properties of matter and energy.'

Present research findings at conferences and lectures

Presenting research requires physical presence, real-time audience engagement, answering spontaneous questions with deep expertise, reading room dynamics, and building scientific relationships. AI cannot replicate the embodied, interactive, and social dimensions of conference presentations and lectures.

BLS evidence: They 'present research findings at conferences and lectures.'