From information gathering to decision support
The online researcher’s job is changing. Generative AI now handles much of the first pass: collecting sources, summarizing documents and spotting patterns in large text sets. That does not make research less necessary. It moves the value of the work toward the steps AI handles poorly: framing the question, judging which sources deserve trust, verifying claims and turning findings into a recommendation someone can act on. The work is shifting from “I found it” to “I can defend it”.
This article looks at what labor statistics and workplace studies say about that shift, which human skills matter most, and how practitioners can choose training. Where the evidence runs out, the author’s view is labeled as such.
What the labor-market evidence shows
The data do not support either extreme: AI is neither making research work obsolete nor creating demand for every kind of researcher.
- Research roles are still growing. The U.S. Bureau of Labor Statistics (BLS) projects employment of market research analysts to grow 7% from 2025 to 2035, faster than the average for all occupations, with about 82,000 openings a year; the median wage was $78,760 in May 2025 (BLS market research outlook).
- Routine information work is more exposed. In case studies of how generative AI may affect its projections, BLS expects paralegals and legal assistants to grow only 1.2% from 2023 to 2033, against 4.0% for all occupations, and projects declines for credit analysts (−3.9%), claims adjusters, examiners and investigators (−4.4%) and customer service representatives (−5.0%). By contrast, financial and investment analysts are projected to grow 9.5%, and budget analysts 3.9%, partly because communication with decision-makers still needs people (BLS AI occupational projections). The authors note these projections reflect information available in June 2024.
- Analytical skills are in demand across jobs. In the World Economic Forum’s Future of Jobs Report 2025, seven in ten employers rated analytical thinking an essential core skill, and AI and big data topped the list of fastest-growing skills. Employers expect about 39% of workers’ existing skills to change between 2025 and 2030 (WEF jobs and skills report).
The author’s view: these figures suggest that research tasks consisting mainly of retrieving and restating information are the most exposed, while work that combines research with judgment, verification and communication is holding up. That is an interpretation of the data, not a forecast the sources make.
Where AI helps, and where it fails
Two workplace studies show how AI’s effects depend on the task and the setting. Their findings should not be treated as forecasts for all online research work.
In the published study of 5,172 customer-support agents, the staggered introduction of a generative AI assistant increased issues resolved per hour by 15% on average. Less experienced and lower-skilled workers improved both speed and quality, while the most experienced workers saw smaller benefits (Published AI workplace study). This was a customer-support setting, not a test of online researchers, but it challenges the assumption that AI necessarily eliminates entry-level work.
In a preregistered experiment with 758 knowledge workers, run with Boston Consulting Group, participants using AI completed 12.2% more tasks, worked 25.1% faster and produced higher-quality work on tasks within the AI’s capabilities. On a task deliberately chosen to lie outside those capabilities, AI users were less likely to reach a correct answer than those working without it (AI knowledge work experiment). The study’s title phrase, the “jagged technological frontier”, describes this uneven boundary between tasks AI handles well and tasks it does not.
Two further findings show why verification matters:
- AI can fabricate sources. About 55% of the citations GPT-3.5 produced in short literature reviews, and 18% of those from GPT-4, referred to works that do not exist (AI citation fabrication study). These figures concern those models and the study’s 2023 tasks, not the current error rate of every AI tool.
- AI increases the volume of unreliable content. NewsGuard had identified 3,749 AI content farm sites by 23 June 2026: sites that publish mostly AI-generated material with little human oversight while presenting themselves as ordinary news outlets (NewsGuard AI content tracker).
AOFIRS sets out a step-by-step method in verify AI answers and citations.
The augmented researcher
The term “augmented researcher” describes a practitioner who uses AI for speed and keeps human control of the steps where errors are costly. Four skills define the role.
Asking the right questions
Designing precise questions and prompts that steer AI tools toward relevant, checkable results, and recognizing when a question falls outside what a tool can reliably answer.
Critical analysis and synthesis
Connecting fragmented information, identifying patterns and supplying the real-world context that a summary leaves out.
Verification and ground truthing
Cross-checking AI output against primary sources to catch invented citations, outdated facts and misread documents before they reach a client or a publication.
Ethical and strategic judgment
Weighing bias, privacy and legal constraints, and understanding the business or social consequences of a finding.
Areas where verification is valued
Demand for research skills is strongest where a wrong answer is expensive. The author identifies four areas where verification can be valuable, although no single dataset measures their relative demand directly:
- Competitive intelligence: monitoring markets and translating signals into strategy.
- Due diligence and investigative research: supporting legal, financial and journalistic work with verified evidence.
- Scientific and academic research support: running AI-assisted literature searches while checking every reference.
- Business intelligence synthesis: turning raw data into briefings leadership teams can act on.
The BLS projection of 9.5% growth for financial and investment analysts (BLS AI occupational projections) is consistent with this pattern, but it is one indicator, not proof.
The research toolkit in 2026
Research workflows now combine several kinds of tools:
- AI agents and assistants, used for planning searches and summarizing retrieved material, with a human checking each output.
- Data visualization, to present complex findings clearly.
- Closed and specialist databases, whose records may require direct database queries, credentials, or licensed connectors beyond ordinary web search. The AOFIRS guide invisible web research methods explains how to find them.
- Documented hybrid workflows, in which people supervise, validate and interpret machine output and log each step.
Who needs these skills
- Core research professionals: researchers, analysts, librarians and academics updating their methods for AI tools.
- Adjacent knowledge workers: marketers, content strategists, founders, consultants, journalists and legal professionals who now do research as part of their job, often without formal training in verification.
- Students and early-career professionals: entering workplaces where AI literacy and source evaluation are expected from the start.
How to choose research training
Many providers now offer courses in AI-assisted research, including universities, professional bodies and private organizations such as AOFIRS. Whichever you consider, check whether the course:
- publishes a full syllabus and states how learning is assessed;
- teaches prompt design for research together with safeguards against invented facts;
- covers verification methods in depth: triangulation, contradiction tracking and source provenance;
- includes ethics, privacy and responsible AI use;
- trains you to communicate findings as decision-ready briefings;
- uses practical, assessed exercises on real tools rather than demonstrations only;
- is recognized by employers or professional bodies in your field.
Search and summarize
Search → Summarize → Deliver
Unverified claims can pass directly into the output.
Verified research workflow
Frame the question → Search with AI assistance → Verify primary sources → Document evidence → Brief the decision-maker
The researcher checks sources, uncertainty, and implications.
Checklist: skills of the augmented researcher
Use this list to assess your own practice or a training program.
Questions and planning
- I can turn a vague request into a specific, answerable research question.
- I know which tasks AI tools handle well and which they do not.
Verification
- I check every AI-supplied citation by opening the source.
- I confirm important facts with at least one independent source (triangulation).
- I record contradictions between sources rather than ignoring them.
- I check the provenance of images, video and documents before relying on them.
Evidence discipline
- I keep a log of searches, sources and dates so another person can retrace my work.
- I save or archive key sources in case they change or disappear.
Communication and ethics
- I state a confidence level for each main finding and explain why.
- I consider privacy, consent and legal limits before collecting information about people.
Conclusion
The available evidence points to a change in research work rather than its disappearance. AI speeds up routine retrieval and summarizing, helped less experienced workers in the customer-support study and fails in ways that are hard to spot. In the author’s view, the researchers who will remain in demand are those who can verify, document and explain what they find. The skills to do that can be learned, and the checklist above is a practical place to start. For career routes into the field, see become an internet researcher.
References
- Brynjolfsson, E., Li, D., & Raymond, L. R. (2025). Generative AI at work. The Quarterly Journal of Economics, 140(2), 889–942. Published AI workplace study
- Dell’Acqua, F., McFowland, E., III, Mollick, E., Lifshitz-Assaf, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality (Harvard Business School Working Paper No. 24-013). AI knowledge work experiment
- Machovec, C., Rieley, M. J., & Rolen, E. (2025, February). Incorporating AI impacts in BLS employment projections: Occupational case studies. Monthly Labor Review. U.S. Bureau of Labor Statistics. BLS AI occupational projections
- NewsGuard. (2026). AI tracking center [Web page, last updated June 23, 2026]. NewsGuard AI content tracker
- U.S. Bureau of Labor Statistics. (2026, August 27). Market research analysts. In Occupational outlook handbook. BLS market research outlook
- Walters, W. H., & Wilder, E. I. (2023). Fabrication and errors in the bibliographic citations generated by ChatGPT. Scientific Reports, 13, 14045. AI citation fabrication study
- World Economic Forum. (2025, January 7). The future of jobs report 2025. WEF jobs and skills report
