AI can write code and run experiments, but formulating genuinely novel research questions, designing studies, and advancing the field's understanding still require a trained researcher. Here is what the research says about the computer and information research scientist profession in 2026, and what you can do about it.
Get My Personalised Fossil ScoreFossil Score
68
Species
Archaeopteryx
AI can write code and run experiments, but formulating genuinely novel research questions, designing studies, and advancing the field's understanding still require a trained researcher.
Task Automation Risk
28%
of current computer and information research scientist tasks are automatable with existing AI tools
AI tools are transforming the mechanics of computer science research — GitHub Copilot and Cursor accelerate implementation, Papers With Code surfaces related work automatically, and large language models can draft literature review sections in minutes. Roughly 28% of the research workflow (routine coding, literature synthesis, boilerplate writing) is being streamlined. What remains irreducibly human: identifying which problems are worth solving, designing experimental methodology for a novel claim, peer reviewing work rigorously, and making the creative leaps that move a field forward. Computer and information research scientists are working on the frontier of AI itself — which means they simultaneously study automation and deploy it in their own work. The researchers who understand both the theoretical foundations and the practical deployment realities are building careers that sit at the most valuable intersection of the field.
Task Autopsy
🦕 Class A — At Risk Now
🦅 Class C — Protected
Your AI Toolkit
You don't need to learn all of these. Pick one, use it for a week, and see how it fits into your work. Most have free options so you can try before you commit.
The dominant deep learning research framework — used at virtually every major AI lab for building and training neural networks, required knowledge for ML/AI research
Try it ↗Experiment tracking and model visualisation — logs training runs, compares experiments, and tracks model performance across sweeps, now standard in ML research labs
Try it ↗Research aggregation site that links papers to their open-source implementations — the fastest way to find the state-of-the-art on any benchmark
Try it ↗AI coding assistant — accelerates implementation of research algorithms and reduces time spent on boilerplate infrastructure code
Try it ↗Open-access preprint server for computer science, mathematics, and physics — essential for staying current with the field, as most significant results appear here before journal publication
Try it ↗ACM and IEEE are the primary professional organisations for computer science — membership provides access to digital libraries, conference proceedings, and professional standing in the field
Try it ↗Extinction Timeline
AI coding assistants are already standard in research labs — most researchers report using GitHub Copilot or Cursor for implementation work. Experiment tracking with Weights & Biases is becoming standard practice for reproducibility.
AI-assisted paper writing is accelerating review cycles. Researchers who can evaluate AI-generated text critically and use it to iterate faster will publish more. Those who rely on it without deep domain knowledge will produce work that fails peer review.
Computer science research will remain a high-value, human-led discipline precisely because it defines the frontier of what AI can do. The researchers who identify the most important problems and design the most rigorous experiments to address them will remain in short supply regardless of the tools available.
Not for the research work that matters. AI accelerates the mechanics of research — coding, literature search, drafting — but it cannot identify which questions are worth asking, design valid experiments for novel claims, or produce the creative insights that advance a field. If anything, the tools that automate routine research tasks raise the bar for what constitutes genuine scientific contribution.
PyTorch and JAX for deep learning research; GitHub for version control; Weights & Biases for experiment tracking and reproducibility; Papers With Code and Semantic Scholar for literature search; Overleaf for collaborative paper writing. arXiv preprints make it essential to track the field in real time.
AI safety, formal verification, cryptography, and systems security are high-demand areas where the gap between current capability and what is needed is large. Human-computer interaction research that keeps humans meaningfully involved in AI-assisted work is growing as organisations deploy AI at scale and need to understand the consequences.
Essential at the top research labs (Google DeepMind, OpenAI, MSR, academic tenure-track positions). For applied research roles at technology companies, a strong publication record and demonstrable research skills matter more than the degree specifically. The PhD remains the default credential for anyone aiming to lead research programmes.
Take the free Fossil Score assessment at DontGoDinosaur.com. It looks at your specific daily tasks — not just your job title — and gives you a personalised risk score with practical steps for the next 6 months. It takes about 4 minutes.
More in Computer & Mathematical
Computer Network Architects
Automated provisioning and network monitoring have absorbed the routine work, but designing networks for complex security requirements, hybrid cloud environments, and resilience under failure is still an expert job.
Actuaries
AI is automating actuarial modelling and data work rapidly. The regulatory requirement for human sign-off protects the credential — but the work behind it is changing fast.
Operations Research Analysts
AI helps operations research analysts do their jobs better and faster, but it can't replace the human skills at the heart of this work.
Statisticians
AI helps statisticians do their jobs better and faster, but it can't replace the human skills at the heart of this work.
Web and Digital Interface Designers
AI helps web and digital interface designers do their jobs better and faster, but it can't replace the human skills at the heart of this work.
Architectural and Engineering Managers
Project dashboards and resource planning tools run automatically. Deciding which projects to take, resolving a dispute between the structural and mechanical teams under a deadline, and managing a client whose expectations have shifted — those decisions are still with the manager.
Further reading
Your Personal Score
Get a Fossil Score built on your actual daily tasks, not a category average. 4 minutes. Free.
Calculate My Personal Fossil Score