Robots Are Better Than People home
Subscribe

Will AI Replace Data analysts?

Plain-language data tools now handle much of the SQL and dashboard work that used to fill an analyst's week. The evidence points to fewer routine analyst seats, while people still own the judgment, the business context and the blame.

Our readFewer seats, not zeroOur read is that AI now absorbs routine querying and reporting, but benchmarks and design write-ups show it still fails on messy company data without a person who knows the business.

The fear is simple: a manager types a question into a chatbot, gets a chart back, and stops needing a data analyst. The evidence says part of that fear is already true, and part of it is not.

A TechTarget columnist who argued in early 2025 that AI would not replace analysts revisited the question and wrote that "software does a large part of the analyst's work." The same TechTarget follow-up on AI and data analysts also argues that people, and only people, still answer for that work, judge the outcomes and train the next generation of analysts.

Our read: the job is shrinking at the bottom and shifting toward ownership, definitions and judgment.

Where AI is already pulling data and building reports

This is not a forecast. Named companies have deployed tools that do core analyst tasks.

OpenAI's internal data agent. In 2026, OpenAI described an in-house agent that serves over 3.5k internal users working across 600 petabytes of data and 70,000 datasets. According to StartupHub's coverage of the OpenAI agent, the goal is "allowing employees, not just dedicated data analysts, to pull data" and run their own analysis. A b-lab breakdown of the agent says it handles the full workflow: data discovery, SQL generation and execution, and report publication. Digital Watch Observatory reports it is built to move employees from complex questions to reliable insights in minutes.

Databricks AI/BI Genie. Genie lets business users ask data questions in plain language. It reached General Availability on June 12, 2025, after Databricks reported more than 4,000 customers adopted it during preview, per Refonte Learning's write-up on Genie. That figure comes from Databricks via a secondary source.

Snowflake Cortex Analyst. Snowflake's Cortex Analyst documentation pitches self-serve analytics via natural language queries for business teams and non-technical users. A Beri comparison of Snowflake Copilot, Genie and ThoughtSpot notes the split: Snowflake Copilot is built for analysts writing SQL, while Cortex Analyst targets business users.

Klarna's analytics team. In 2025, CNBC reported that CEO Sebastian Siemiatkowski linked Klarna's workforce reduction to slimming its analytics team into one "success team," with many people leaving through natural attrition. In December 2024 he said Klarna had stopped hiring a year earlier as it invested in AI, with headcount down 22% to 3,500, according to BankingNewsAI's Klarna tracker. Charter describes Siemiatkowski dropping his team's Google Sheet from a pricing strategy presentation into an AI tool to build his own view of the data.

Product managers doing their own analysis. In a Microsoft New Future of Work Report 2025 study of 885 PMs, 12% said they use GenAI for prototyping and coding and can now do tasks they used to rely on data science colleagues for.

Field note: The weak spot is business context. A data agent design write-up warns that without a context layer, "even strong models hallucinate column names and misinterpret business terminology." If your company's definition of "active customer" lives only in your head, the agent will guess. Writing those definitions down is analyst work, and it is the part the tools depend on.

Which data analyst tasks AI can handle now

TaskWhat AI does todayWhat still needs a person
Finding the right tableOpenAI's agent runs data discovery across 70,000 datasetsKnowing which source is trusted and which is stale
Writing SQLAgents generate and execute queries from plain-language questionsCatching wrong joins and invented column names on real warehouses
Building dashboards and chartsTools turn a spreadsheet into a visual on request, as Klarna's CEO showedChoosing what the business should track
Ad hoc questions from managersGenie and Cortex Analyst let non-technical users self-serveSpotting when the question itself is wrong
Defining metrics and business termsLittle: models misread business terminology without contextWriting and maintaining the definitions agents rely on
Checking resultsSome vendor tools add verification layersSigning off and answering for errors
Presenting a recommendationDrafts summaries and reportsDefending the call in the room and owning the outcome

What the numbers say about data analyst jobs

No official dataset in our evidence tracks "data analysts" as a separate occupation. The closest figures cover data scientists and related analyst roles, so treat them as neighbors, not a direct match.

  • Growth for the closest BLS category. The Bureau of Labor Statistics projects employment of data scientists to increase 33.5 percent between 2024 and 2034. BLS did not publish this figure for data analysts.
  • High overlap with AI. In Microsoft's study of real Copilot conversations from January – September 2024, described in the Working with AI paper, data scientists scored 0.36 on AI applicability and market research analysts 0.35, per R&D World. Forbes lists statistical assistants, data scientists and market research analysts among the most affected roles.
  • Overlap is not job loss. As Fox News reported on the same study, "This score reflects overlap, not guaranteed job loss." The researchers stressed that no occupation is fully performed by AI.
  • Users are not thrilled with AI analysis. R&D World notes data analysis activities received some of the lowest satisfaction scores among common tasks in the Microsoft data.
  • Real usage. The Anthropic Economic Index found AI use more prevalent for tasks tied to mid-to-high wage occupations like computer programmers and data scientists.
  • Talent still scarce. IBM's 2025 survey of senior data and analytics executives, cited by TechTarget, found 47% ranked attracting and retaining skilled people among their top problems.

On accuracy, the Spider 2.0 benchmark tested models on real-world enterprise SQL workflows. The most advanced model tested, o1-preview, reached a success rate of only 21.3%. The alphaXiv summary adds a fair caveat: "These are benchmark results. They do not measure performance in live enterprise deployments." The BEAVER benchmark, introduced in 2024, was the first enterprise text-to-SQL dataset built from actual data warehouses, which is exactly where analysts spend their days.

Which data analysts are most exposed

Report and query specialists

If your week is mostly pulling numbers on request and refreshing dashboards, you sit directly in the path of Genie, Cortex Analyst and OpenAI-style agents. Those tools exist to let the requester skip you.

Junior analysts at companies with frozen hiring

Klarna cut its analytics staff through attrition and stopped hiring, not through mass layoffs. Our read is that entry-level analysts feel that pattern first, because the open seats simply never appear.

Analysts serving product teams

The Microsoft PM study shows product managers already doing work they once handed to data colleagues. Analysts embedded with product teams should expect fewer simple requests.

Less exposed: context owners and senior reviewers

Analysts who own metric definitions, data quality and final sign-off are harder to replace. The tools need that context to avoid hallucinated columns, and someone has to answer when a number is wrong. The American Statistical Association's ethical AI statement frames statistical practitioners as responsible for the ethical use of AI and for keeping public trust in statistics.

Our verdict: AI takes the queries, analysts keep the accountability

We think AI will not wipe out data analysts, but it is already shrinking the routine end of the job. Self-serve tools are live at named companies, and at least one CEO has tied a smaller analytics team to AI. At the same time, models still stumble on real company warehouses, misread business terms, and leave people to answer for the results.

The analysts who last will be the ones who make the AI right: defining the data, checking the output and owning the decision.

The bottom line: Our read is that pure SQL-and-dashboard roles are shrinking now, while analysts who own business context, data quality and final judgment remain in demand.

Questions people ask

Will AI replace data analysts?

Not fully, in our read. Tools from OpenAI, Databricks and Snowflake already let non-analysts pull data and build reports, but a TechTarget columnist notes that only people answer for the work and judge the outcomes.

Has any company cut data analysts because of AI?

Klarna's CEO linked a workforce reduction to slimming its analytics team into one success team, with many leaving through natural attrition, according to CNBC. He also said Klarna stopped hiring while investing in AI, with headcount down 22% to 3,500.

How accurate is AI at writing SQL on real company data?

On the Spider 2.0 benchmark of enterprise SQL workflows, the most advanced model tested, o1-preview, reached a success rate of only 21.3%. Those are benchmark results and do not measure performance in live enterprise deployments.

Is demand for data jobs still growing?

BLS projects employment of data scientists, a close neighbor of data analysts, to increase 33.5 percent between 2024 and 2034. IBM's 2025 survey found 47% of senior data and analytics executives ranked attracting and retaining skilled people among their top problems.

Which analyst roles are most exposed to AI?

Microsoft's study gave data scientists an AI applicability score of 0.36 and market research analysts 0.35, and Forbes lists statistical assistants among the most affected roles. Researchers stressed the score reflects overlap, not guaranteed job loss.

Sources

Every number, quote and example on this page was checked by script against the sources below before publishing: quotes must match the source word for word, and figures must appear in a source or in BLS data. Our verdict is editorial judgment. Spotted an error? Tell us.

  1. Will AI replace data analysts: A year and a half later (techtarget.com)
  2. Companies are blaming AI for job cuts. Critics say it’s a 'good excuse' (cnbc.com)
  3. Artificial intelligence, information technology, and employment, 2024–34 : The Economics Daily : U.S. Bureau of Labor Statistics (bls.gov)
  4. Microsoft study: AI could transform desk work and data science more than lab bench (rdworldonline.com)
  5. Jobs that are most at risk from AI, according to Microsoft (foxnews.com)
  6. Working with AI: Measuring the Applicability of Generative AI to Occupations (arxiv.org)
  7. New Future of Work Report 2025 (microsoft.com)
  8. Are There Any AI-Safe Jobs Left? Microsoft’s Surprising Data (forbes.com)
  9. OpenAI's In-House AI Agent: Democratizing Data Insi… (startuphub.ai)
  10. OpenAI streamlines data analysis with in-house AI agent | Digital Watch Observatory (dig.watch)
  11. Data Agents: From OpenAI's 6 Layers to Open-Source Alternatives (vanducng.dev)
  12. Inside OpenAI’s in-house data agent (b-lab.team)
  13. Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflows (arxiv.org)
  14. Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflows (alphaxiv.org)
  15. Semantic Caching and Intent-Driven Context Optimization for Multi-Agent Natural Language to Code Systems (arxiv.org)
  16. American Statistical Association Statement on Ethical AI (amstat.org)
  17. Orientation, Technical Training, Placement & Continuity (refontelearning.com)
  18. Cortex Analyst | Snowflake Documentation (docs.snowflake.com)
  19. Snowflake Copilot vs Genie vs ThoughtSpot: Copilot Is the Wrong Tool (beri.net)
  20. Introducing the Anthropic Economic Index (anthropic.com)
  21. What Klarna learned from its ambitious AI rollout (charterworks.com)
  22. Klarna AI Strategy: CEO Statements, AI Products, OpenAI, Headcount (bankingnewsai.com)

The White-Hat Brief

Weekly: what the machines learned, and how to stay ahead

One email a week. Verified stories, plain-language defenses, no hype. Unsubscribe anytime.