Will data science become less sexy? Before these AI breakthroughs arrived, responsibilities were neatly divided: business analysts gathered requirements and aligned them with company goals, data analysts turned raw information into useful reports, and data scientists developed predictive models and machine learning experiments. Now, specialized roles — like ML engineers and data product managers — tackle much of what data scientists once owned, while “citizen data scientists” build robust models using ChatGPT’s o1 models, Claude, or AutoML. That reduces the aura of “unicorn” data scientists, though advanced tasks still exist — creating new algorithms, analyzing black-box behavior, and developing more complex architectures. Many companies are reorganizing for GenAI, pulling data scientists away from complete ownership and into more specialized, high-impact contributions, but also opening up space for a new blend of domain insight, AI oversight, and collaboration.
Meanwhile, the rise of advanced GenAI tools is blurring the boundaries between roles even further. Recent industry data shows that 78% of organizations now use GenAI-powered analytics platforms, allowing business users to perform analyses that were once handled by technical specialists. The World Economic Forum’s Future of Jobs Report 2023 highlights analytical roles as shifting significantly, naming “analytical thinking” and “AI and big data” as the two fastest-growing skill areas through 2027.
Automated workflows can now handle coding, data visualization, and even design tasks that used to consume entire teams. In many ways, this is a welcome change. Who truly misses the late-night grind spent tweaking PowerPoint slides for investor roadshow, or hunting through Python scripts for a missing variable? Today’s leaders are less impressed by someone who can merely “crank out a spreadsheet” without using their mouse and more interested in professionals capable of weaving data insights into business strategies. Job descriptions are shifting away from technical query requirements toward interpretation and strategic application.
The rise of natural language interfaces has erased 60% of SQL tasks for BAs.
Rather than defining their value by how well they can wrangle Excel macros or keep a massive SQL database in check, they place a premium on bridging strategic questions and AI-driven answers. Stakeholder communication and conceptual AI literacy are becoming as essential as the ability to write optimized machine learning code.
Technical Debt and Automation
But is the pressure high enough (this time)? Seeing 50bn+ investment funds running thousands of positions in Excel sheets or major healthcare systems still using macros and siloed spreadsheets to manage patient data over the course of my career taught me that when outdated processes linger, teams risk pouring valuable time and resources into simply ‘keeping the lights on’ rather than prioritizing strategic growth and innovation. Despite the optimism that GenAI tools bring, organizations still grapple with mounting technical debt. Quick fixes — like patching an unwieldy Excel file for the hundredth time — might offer temporary relief but ultimately drive up long-term costs.
Technical Debt: Why Excel Costs More Than You Think — maintenance consumes 42% of analyst time
Meanwhile, GenAI is automating coding chores and standard data-preparation steps, making it easier for newly empowered users to bypass clunky legacy systems. These companies are actively involved in data democratization process that breaks down barriers that prevent people from accessing, analyzing, and utilizing data effectively. The new roles of data scientists will become the bridge between technical AI systems and actual business implementation, so investing talent and resources in upstream data assets and processes becomes more critical to stay competitive.
From Silos to T-Shaped: Startups Teaching Big?
Midjourney, valued at over $10 billion, launched in 2022 with only 11 employees and currently operates with just 131 people. Its meteoric rise demonstrates a broader shift in startup culture: fewer employees, accelerated growth, and heavy reliance on AI. Founders across industries once believed that bigger teams signaled bigger success — more engineers, more sales reps, more overhead. The new wave of AI-native companies is rewriting these rules. Small teams are attracting large funding rounds and automating processes that historically required entire departments. Large enterprises have taken notice and are increasingly concerned about losing their competitive edge to these lean, AI-driven disruptors.
ElevenLabs, operating with a team of around 155, exemplifies the same trend: rather than hiring dedicated teams for separate functions — like massive customer support or marketing departments — leaders tap AI to handle repetitive tasks. Their employees can then focus on strategic decisions and collaborative problem-solving. Other emerging startups have demonstrated similar success. Some operate with fewer than ten employees and still secure substantial funding while reaching multimillion-dollar ARR. They rely on advanced AI tools such as GitHub Copilot, Cursor, ChatGPT-based chatbots, and specialized AI agents to automate chat support, software development, data analytics, and marketing campaigns. These AI agents form on-demand “teams” to handle tasks like data gathering, strategic planning, and customer engagement, creating a new mode of knowledge work that streamlines processes once handled by entire departments. As a result, day-to-day operations run smoothly with minimal headcount.
Small, AI-powered enterprises are not the only success stories. Larger organizations are turning to AI to overcome entrenched inefficiencies, break down departmental silos, and accelerate decision-making. Rather than spinning up separate data analysts, data scientists, and engineering teams that rarely interact, forward-thinking companies combine these functions into cohesive “pods” of T-shaped professionals who can do everything from writing Python scripts to explaining model outcomes to stakeholders.
What Are T-Shaped Roles in the AI Era?
A T-shaped role describes a professional who possesses broad, cross-functional knowledge and deep expertise in one or two specialized areas. In AI, the horizontal dimension includes familiarity with data governance, prompt engineering, user experience, and core business strategy. The vertical dimension might involve advanced machine learning research, financial modeling, supply-chain optimization, or domain-specific know-how in regulated fields like healthcare or defense.
- AI Oversight As generative AI automates tasks — such as data cleaning, feature engineering, or even rudimentary coding — T-shaped professionals step in to validate results, interpret model outputs, and connect these findings to tangible business objectives. Instead of focusing solely on code efficiency or data wrangling, they delve into ethical usage, risk mitigation, and alignment with strategic goals.
- Domain Expertise AI seldom operates in a vacuum. Healthcare-focused teams must understand compliance and patient privacy; financial modeling projects require deep awareness of regulations, market behavior, and investor relations. T-shaped practitioners can integrate specialized domain expertise with a working knowledge of AI’s capabilities, leading to more relevant outcomes and smoother stakeholder buy-in.
- Cross-Functional Collaboration AI implementations do not thrive behind departmental barriers. A T-shaped individual speaks the language of data scientists, IT engineers, marketing strategists, and executive leaders, acting as a conduit for cross-team collaboration. This holistic engagement increases the likelihood that AI-driven solutions will be both technically robust and commercially viable.
While T-shaped professionals serve as the backbone of lean, AI-enhanced organizations, larger companies often grapple with how to integrate this concept without dismantling their entire operational fabric. Traditional Teams rely on separate, specialized departments that add overhead and delay decision-making. AI-Optimized Teams, on the other hand, empower T-shaped individuals who collaborate in real time and adapt swiftly to analytical insights. In the table below, you can find my attempt to visualize how these contrasting models stack up, illustrating how T-shaped roles reduce bottlenecks and support more flexible, outcome-focused work.
The traditional siloed structures introduce added layers of costly overhead and slow the feedback loop. AI-optimized teams address those inefficiencies by merging responsibilities into cross-functional pods that adapt quickly to new information. This structure encourages “Adaptive Decision-Making in Pods,” where AI-based Decision Support Systems (DSS) respond to fresh data in near-real time, reducing the need for lengthy handoffs. Through “Cross-Functional AI Integration,” T-shaped professionals combine broad operational awareness with specialized subject-area knowledge, bridging the gap between technical outputs and business objectives. Leaders who adopt T-shaped roles witness more agile decision-making and stronger cohesion between technical and strategic priorities. Many organizations discover that flexible team compositions lead to lower operating costs, faster pivots, and improved outcomes across their core business activities.
Redefining Roles & Responsibilities
Accessible GenAI platforms and the broader democratization trend challenge traditional, siloed structures by making data and analytical tools more widely available. Non-technical employees now have access to simplified analytics and can handle tasks that previously required extensive coding. This shift is changing established roles like Business Analyst (BA) and Data Scientist (DS), creating a new hybrid role: the Business Data Scientist (BDS). BDS professionals combine deep subject-area understanding with AI-oriented oversight and strong communication skills, stepping away from the elusive “unicorn” model and toward a structured blend of technical fluency and business insight.
Organizations are also seeing the rise of “PromptOps” as a critical function designed to refine model prompts, monitor AI outputs, and align analytical priorities with strategic goals. BDS roles extend the anchor classification from Fayyad & Hamutcu (2020) to cover tasks that span end-to-end AI development, from data preparation to stakeholder-facing storytelling. This broader ownership fosters stronger connectivity across teams, which helps drive more meaningful outcomes. Below are four main focus areas that define how BDS professionals integrate these responsibilities:
- Strategic Alignment They pinpoint where AI outputs can address real-world objectives such as new product launches or customer acquisition.
- Technical Know-How They possess enough Python and SQL familiarity to manage low-code AI tools, rather than relying on data engineering teams for every system adjustment.
- Cross-Functional Collaboration They translate complex findings so that executives and frontline managers can act on them, stressing ethical usage and long-term viability.
- Results Delivery They connect AI outputs to the organization’s immediate priorities, pitching concise recommendations to leadership or guiding teams on operationalizing model-driven insights.
AI Integration Pyramid
This comparative snapshot of BA-vs-DS-vs-BDS highlights only part of a wider transition that teams undergo when mundane tasks shift to Generative AI. In reality, many organizations progress through five tiers — from the “Excel Jockey” rapidly replaced by automated reporting, to the “Business Data Scientist” wielding AI for strategic impact. Not surprisingly, more than 7.5 million data entry jobs are predicted to disappear by 2027 (World Economic Forum) — the largest loss of any profession. Data entry clerks, administrative secretaries, and accounting roles rank as the most at risk. As repetitive workflows vanish, skills such as creativity, critical thinking, and oversight become more vital to interpret AI outputs, ensure relevance, and preserve diverse approaches.
Overly trusting polished AI suggestions can spark “mechanized convergence,” where teams converge on homogeneous solutions and lose breadth of insight. This may boost short-term efficiency but ultimately dulls deeper problem-solving abilities if everyone settles for “good enough” answers. It also goes beyond “overreliance”: even correct-looking AI outputs can limit alternative viewpoints if individuals never explore beyond the initial response. Balancing human self-confidence against AI strengths remains essential, lest confident reliance on AI overshadow the healthy scrutiny needed to identify hidden errors or biases.
In this new reality, skill development focuses on adapting to automation rather than competing against it. Emerging roles — like the AI-Assisted Analyst and Hybrid Data Scientist — bridge technical know-how and business context, tuning prompts, validating data, and leveraging explainable AI (XAI) and intelligent exploration to uncover insights in plain language. Their work helps organizations stay adaptable while machines take over routine tasks. Success depends on people who question assumptions, guide the responsible use of AI, and address challenges that technology alone cannot solve.
Implications and Training
More advanced AI architectures are prompting a shift from siloed teams to cross-functional pods that bring together domain expertise, data science skills, and strategic thinking. T-shaped professionals perform well in these pods by combining deep capabilities, such as advanced modeling or financial analysis, with broader knowledge of ethical AI use, stakeholder needs, and ongoing data integration. Instead of coding each module from the ground up, teams assemble solutions using connectors, AI endpoints, and dashboards that adjust to changing data without adding unnecessary overhead. Leaders who take this approach reduce delays, speed up iteration, and create more stable outcomes for their AI efforts.
Training and education must keep pace by emphasizing prompt engineering, interpretability, ethical risk management, and clear communication as core competencies. Universities, bootcamps, and corporate academies have already noticed the changing skill landscape, revising their curricula accordingly. Previously, data science programs focused on algorithmic complexity, data structures, and the mechanics of model building; today, they include foundational GenAI concepts — prompt engineering, interpretability, risk assessment, and real-time data pipelines — coupled with group projects that simulate realistic AI deployments. Students and professionals who refine their ability to simplify technical outputs, collaborate with non-technical teams, and adapt prebuilt AI platforms for specialized use cases will stand out. Meanwhile, organizations that encourage broad-based learning — merging domain know-how, problem-solving, and stakeholder alignment — will cultivate T-shaped talent ready to deploy AI responsibly and strategically in an evolving business landscape.
REFERENCE
- https://www.sciencedirect.com/science/article/pii/S2949948824000374
- https://hdsr.mitpress.mit.edu/pub/60qq269p/release/1
- https://www.weforum.org/stories/2024/08/how-ai-brings-corporate-might-to-small-teams-reshaping-business-for-all/
- https://www.microsoft.com/en-us/research/wp-content/uploads/2024/12/NFWReport2024_1.27.2025.pdf
- https://www.microsoft.com/en-us/research/wp-content/uploads/2025/01/lee_2025_ai_critical_thinking_survey.pdf
- https://research.aimultiple.com/risks-of-generative-ai/
- https://www.weforum.org/publications/the-future-of-jobs-report-2023/
- https://www.ibm.com/thought-leadership/institute-business-value/en-us/report/human-potential-genai
- https://www.gartner.com/smarterwithgartner/how-to-use-citizen-data-scientists-to-maximize-your-da-strategy
- https://arxiv.org/pdf/2401.11817
