TL;DR

Analysts who embrace AI as a productivity tool and double down on skills like storytelling, prompt engineering, and strategic thinking will become even more valuable. The future belongs to analysts who combine human insight with AI efficiency.

Who should read this?

Data analysts, analytics leaders, BI professionals, and anyone interested in how AI is reshaping roles in data teams, from technical delivery to strategic decision-making and cross-team enablement.

Black and White Photo of a Woman Using a Vintage Computer. Pexels

The explosion of data in recent years, 90% of the world's data was created in the last two years and will grow 150% by 2025, means the need for humans who can interpret it has only increased. Despite alarmist headlines, generative AI is unlikely to make data analysts obsolete. AI excels at crunching numbers and spotting patterns, but it lacks the human insight, context and empathy that analysts bring. As technologist Donald Farmer observes, "Generative AI will not replace data analyst jobs"; it can automate routine tasks, but it "can't replace fields requiring human empathy and insight". In practice, AI becomes a powerful tool that lets analysts work faster and focus on strategy, rather than a true replacement.

In fact, the future role of analysts looks more critical than ever. The U.S. Bureau of Labor Statistics estimates that data analyst positions may surge 36% between 2023 and 2033, demonstrating the resilience of this profession in the face of technological advancement. With so much new data, organisations will double down on analytics talent to make sense of it. Analysts will still draw the charts and dashboards (often using AI-powered tools), but more importantly they will craft the story behind the data.

As Ganes Kesari notes, "data stories have a tailored narrative…to help users act" because "no one ever made a decision because of a number. They need a story.". In other words, while AI can generate charts and summaries, humans must interpret them for business meaning. Analysts know "exactly where data can help and crucially where it can't." A background in analytics gives leaders a "superpower" of critical thinking and business sense even outside the data team. Smart organisations are not cutting analysts they're embracing them as translators between AI and business.

AI as a Tool: Automating Tasks, Shifting Focus

Generative AI (ChatGPT, Claude, Copilot, etc.) can accelerate many technical tasks, but it usually needs a human in the loop. For example, AI can suggest SQL queries, write data-cleaning scripts, or assemble complex spreadsheet formulas in seconds. This ability frees analysts from manual toil. Donald Farmer explains that AI "can suggest code to extract, clean and analyse data," automating repetitive plumbing tasks. It can also recommend data models (e.g. star schemas) and even analytical methods, acting like a pair of extra hands.

However, the limitations of AI remain clear. The code AI produces often needs refinement: it lacks deep understanding of the specific data context, system architecture and business goals. AI-generated visuals or analyses come from patterns in its training data, not a true grasp of your unique situation. As Farmer notes, AI "does not have the same understanding of the data as a human analyst," so its first attempts can be wrong or irrelevant. In other words, AI can speed up development only if analysts verify and adapt the outputs. The burden of explainability and correctness still lies with the person. AI isn't accountable for errors: "responsibility and blame still rest with humans", so analysts must double-check models for bias and logical flaws.

In practice, analysts will differentiate themselves less by writing every line of code and more by applying judgment and storytelling. AI steps in "not to replace analysts, but to empower them", amplifying their capabilities. Rather than memorising syntax, analysts can focus on why the data matters. They will use AI-generated outputs as a starting point, then add domain knowledge, challenge assumptions, and frame insights in narrative form. In short, analysts must make the AI outputs their own by editing prompts, correcting flaws, and owning the results so that the final output truly reflects human expertise and business context.

An artist's illustration of artificial intelligence (AI). Pexels

The Art of Prompt Engineering

One key skill for modern analysts is prompt engineering. Because AI responds entirely to your instructions, crafting the right prompt is like writing the perfect query. Analysts quickly learn the old maxim: "garbage in, garbage out." A vague or incomplete prompt leads to meaningless output, just as dirty data breaks traditional models. Research confirms that "the way a question or statement is phrased can have a significant impact on the information returned by the model". In other words, analytic insight depends on how well you ask.

Effective prompts are clear, specific, and often step-by-step. As Matt Crabtree explains in a DataCamp tutorial, one should include detailed context and instructions: for example, specifying the desired format or asking the AI to explain its reasoning. Simply typing "give me an answer" rarely suffices. Instead, analysts should break problems into parts and iteratively refine the prompt. In practice, this means patiently dialoguing with the AI, as if guiding a junior analyst.

Best Practices for Prompt Engineering:

Provide clear context and instructions. Include any relevant background or constraints so the AI knows exactly what to analyse.
Be specific and structured. If you want output in a certain format (e.g. "Create a bar chart comparing X and Y"), say so. Asking the model to "explain" or list steps also yields more transparent answers.
Iterate and verify. Test and tweak your prompt based on the answers. Small changes or extra details can dramatically improve results.

Developing these prompt-engineering skills is non-trivial. It "requires understanding the model's behaviour" and often some programming know-how. Analysts who learn to write effective prompts will save hours of grunt work. But this is not a tool to rush through mindlessly; it's a new form of expertise. Prompt engineering is becoming so valuable that savvy professionals are commanding premium salaries for it.

Consider a data-cleaning task. You could tell an LLM "generate a script to clean this dataset." But better practice is to walk it through the logic: show examples of dirty values, outline your cleaning rules, and ask it to produce code with comments. The latter yields code you understand and can maintain. With careful prompting, the AI becomes an apprentice: it handles boilerplate, you confirm accuracy, and you document the reasoning together. This collaboration still depends on the analyst's knowledge to know what to ask and to catch any hallucinations or mistakes.

Ensuring Output Quality and Accountability

Even with great prompts, analysts must validate every AI output. Generative models are notorious for "hallucinating", i.e. confidently asserting false facts or flawed analyses. Without oversight, these errors can slip into reports. AI-generated insights "can contain logical gaps, biased perspectives and factual errors" if unchecked. In the worst case, blindly trusting AI can lead to seriously wrong business recommendations (or worse, data breaches).

Therefore, analysts stay personally accountable. If a manager asks how a chart was generated or why a model says something, the analyst must explain. This means understanding and often rewriting the AI's output. For example, if the AI suggests an unusual segmentation of your customers, the analyst should verify it with actual data and domain logic before trusting it. As Farmer notes, "AI is not accountable for its own errors: responsibility and blame still rest with humans."

In short, data analysts become the critical safety net. They use AI to rapidly generate hypotheses and code, but they apply skepticism and expertise to test each step. Any analysis involving AI should be accompanied by a reality check: comparing against known patterns, checking for data biases, and confirming that the result truly answers the business question. In our own work, for instance, we've seen AI produce plausible but meaningless datasets. Treat these as learning moments; they underline that the human in the loop is indispensable.

Two Silver Chess Pieces on White Surface. Pexels

Path to Leadership and Strategy

With routine tasks automated, analysts can shift toward leadership: strategy, communication, and decision support. In many organisations, analysts can now spend more time in stakeholder meetings, refining KPIs, or driving data initiatives. This is both easier (AI helps finish work faster) and harder (analysts must stay tech-savvy). Regardless, the shift is inevitable: the freed-up hours should go into higher-level thinking, not simply more analysis.

Effective data-driven leaders know both the numbers and the nuances. John Cook, a data analytics leader & strategist, highlights that leaders with analytics backgrounds "are thoughtful… stay curious… they know where data can help and crucially where it can't." These are the people who challenge assumptions, spot when the models are misleading, and blend data insight with business judgment. In fact, a leader who can say "that metric is irrelevant for this decision" is often worth their weight in gold. As Cook puts it, a data-savvy background is "a superpower balanced with good business sense", even in roles far from the data team.

In practice, analysts can prepare themselves by honing soft skill alongside technical ones. They should practice explaining insights in plain language, listening to non-technical concerns, and guiding teams through data literacy. Analysts who communicate well and question wisely will become natural coaches or product owners. Many companies actively rotate analysts into business roles to spread this perspective. In the age of AI, the most valued data professionals will be those who augment organisational wisdom, not just the tools.

Elderly Man Thinking while Looking at a Chessboard. Pexels

Democratising Data Literacy Across the Organisation

Finally, analysts will spend more effort teaching others to use data and AI responsibly. With AI tools becoming widespread, every team can potentially run analyses, so analysts become the data mentors. This might involve training "BI champions" in each department, holding workshops on AI ethics, or building guided analytics templates.

A successful strategy is to build a coalition of data champions across the business. When business leaders drive BI initiatives (rather than IT alone), it "has a much stronger tendency to democratise the data by creating a cross-functional coalition of BI champions". In practice, the analyst might identify a power user in marketing, finance, or operations and help them learn best practices. These champions then serve as first responders for analytics questions, cascading skills throughout the company.

Strategies for Building Data Literacy:

Educate colleagues on AI best practices. Show non-technical teams how to ask good questions of AI and how to critically read AI-generated outputs.
Train and empower "BI champions." Formally appoint go-to analysts in each unit who can guide peers. Provide them with advanced training so they can answer questions internally.
Offer ongoing support. Continuous learning is key. Regular training sessions, office hours, or an internal Slack channel keeps everyone up-to-date. A single onboarding isn't enough; employees need ongoing guidance as tools and needs evolve.

By spreading data literacy, analysts multiply their impact. Rather than hoarding dashboards, they build self-service ecosystems where AI-driven insights flow to every level. In that way, the analyst's role becomes part catalyst, part coach by ensuring that AI's benefits are realised responsibly across the organisation.

Evolution of Data Analysts with AI

To wrap up, data analysts are not an endangered species; they're evolving into AI-powered strategists. The advent of generative AI simply raises the floor on technical productivity, allowing analysts to focus on the most valuable parts of their job. While AI can write queries and summarise tables faster than ever, it still needs human judgment for context, relevance and ethics. Analysts who adapt will find new leverage in the AI era: they will use AI's incredible processing power while contributing the human creativity, curiosity and storytelling that machines cannot replicate.

The question isn't whether analysts will survive, it's how they will thrive. By mastering prompt engineering, emphasising narrative, and guiding their organisations, analysts can become leaders who harness AI responsibly. The data-driven future belongs to those who marry machine efficiency with human insight.

This article, Evolution of Data Analysts in the Age of AI, draws on industry analysis and recent research. All sources are cited above.

References