Computer and information research scientists

Analyze experimental results from software and system testing

AI can process experimental data, perform statistical analysis, identify patterns, generate visualizations, and flag anomalies with high accuracy. Humans primarily validate interpretations and decide on next steps, making this task largely automatable with review.

BLS evidence: The duties section includes 'Analyze the results of their experiments.'

Design and conduct experiments to test software systems operation

AI can design test cases, execute experiments, monitor system behavior, and collect performance metrics with minimal human intervention. Humans primarily review experimental design validity and interpret unexpected results, making this highly automatable with oversight.

BLS evidence: Duties include 'Design and conduct experiments to test the operation of software systems, frequently using techniques from data science and machine learning.'

Write papers for publication presenting research findings

AI can draft paper sections, format references, generate figures, and even structure arguments from research notes. However, crafting novel contributions, positioning work in literature, and meeting publication standards requires researcher judgment, though AI substantially reduces the labor involved.

BLS evidence: Duties include 'Write papers for publication and present research findings at conferences.'

Determine computing needs and system requirements

AI can analyze usage patterns, benchmark requirements, and recommend system specifications based on stated needs. However, eliciting true requirements involves understanding unstated organizational context and future strategic needs, requiring human judgment alongside AI analysis.

BLS evidence: Listed in duties as 'Determine computing needs and system requirements.'

Create and improve algorithms to simplify complex computing tasks

AI excels at optimizing existing algorithms and can suggest improvements to known problem classes, but creating fundamentally new algorithms for complex tasks requires mathematical creativity and insight into problem structure that AI can assist with but rarely originates autonomously at research frontier levels.

BLS evidence: The description explains they 'work with algorithms' and 'simplify to make computer systems as efficient as possible,' leading to advancements in machine learning and cloud computing.

Design and develop new computing languages, software systems, and tools

AI can generate code, suggest language features, and implement specified components, but designing entirely new computing languages and software systems requires architectural vision, understanding of long-term ecosystem implications, and novel abstractions that AI can support but not lead independently.

BLS evidence: The page states they 'Develop new computing languages, software systems, and other tools to improve how people work with computers.'

Design computer architecture to improve hardware performance

While AI can optimize specific architectural parameters and simulate performance, designing novel computer architectures requires deep understanding of hardware-software co-design tradeoffs, manufacturing constraints, and anticipating future workload evolution—creative work AI can assist but not lead.

BLS evidence: The page states 'To improve computer hardware, these scientists design computer architecture' resulting in better networking, faster speeds, and improved security.

Explore computing problems and develop theories and models to address them

While AI can assist with literature review, pattern recognition, and hypothesis generation, exploring novel computing problems requires creative insight, identifying which problems matter, and developing genuinely new theoretical frameworks that current AI cannot reliably produce without substantial human direction and validation.

BLS evidence: The duties section states they 'Explore problems in computing and develop theories and models to address those problems.'

Collaborate with scientists and engineers to solve complex computing problems

Collaboration requires real-time negotiation of problem framing, integrating tacit knowledge across disciplines, building trust, and navigating interpersonal dynamics. AI can facilitate communication and provide technical support but cannot replace the human collaborative process in research settings.

BLS evidence: The duties section explicitly states they 'Collaborate with scientists and engineers to solve complex computing problems.'

Present research findings at conferences

Presenting at conferences requires real-time audience engagement, answering unpredictable questions, reading room dynamics, networking, and establishing credibility through physical presence. AI can generate slides and talking points but cannot replace the human presenter in research contexts.

BLS evidence: The duties section states they 'present research findings at conferences.'