In its 2025 Future of Jobs Report, the World Economic Forum projects that AI will displace around 92 million jobs by 2030. Data science itself – the very discipline that gave birth to modern machine learning – now appears on the list of professions highly vulnerable to the impacts of automation.
For professionals who have spent years mastering code and building predictive models, this creates the uncomfortable irony that the tools they helped create may now be coming for their jobs.
Admittedly, that’s a little dramatic. The real question isn’t whether AI will replace data scientists; it’s whether data scientists have the ability to evolve fast enough to remain indispensable in the face of AI advances.
It’s an important question when one considers that, today, building a machine learning model has become largely a “recipe” exercise. In fact, not only are there textbooks with clearly defined steps; there are platforms that can train, tune, and deploy these models with minimal human intervention.
Routine data cleaning, basic feature engineering and standard model development are increasingly being automated. Tasks that once defined junior data science roles are now handled by AI systems in a fraction of the time. So, if your primary value proposition as a data scientist is executing a predictable sequence of technical tasks like this, you potentially have a problem – because that is precisely what AI does exceptionally well.
Of course, this doesn’t mean that data science careers are doomed – just that, like every other career impacted by AI, the nature of the work that data scientists do must fundamentally change. And the good news is that it can.
AI may now be able to generate model documentation, write code, and produce initial analyses faster than any human; but
- Can it fully understand the subtle context of your organisation’s unique business challenges?
- Can it navigate the political dynamics of presenting findings to sceptical executives?
- Can it figure out that the data anomaly that emerged last quarter was caused by a system migration glitch, not a customer trend?
The answer to these questions is mostly still “no” – and that paves the career path for any data scientist that doesn’t want to go the way of the door-to-door encyclopaedia salesman.
The most valuable data scientists in 2026 and beyond will spend less time producing outputs and more time reviewing them. Of course, this fundamental shift from execution to oversight demands a skill set that traditional education rarely emphasises. Technical proficiency remains necessary but is no longer sufficient on its own. The data scientists who will thrive in the age of AI are those that possess, or are willing to acquire, a different combination of capabilities.
For starters, deep business understanding is essential. For example, knowing how to build a fraud model is a textbook exercise. Understanding the future of fraud detection in your specific industry – and taking into consideration regulatory changes, emerging payment technologies and evolving criminal tactics – requires strategic thinking that AI cannot replicate.
Communication skills, too, have moved from the realm of “nice to have as a data scientist” to indispensable. The ability to translate complex technical findings into actionable business recommendations – and impart those insights persuasively to others – will separate those who lead in the data industry from those who merely participate in it.
Then there’s ethical judgement. This has become a genuine and vital job requirement for the data scientist. As organisations embrace AI, questions of responsible use, bias detection and algorithmic transparency have moved from academic discussions to practical daily decisions. There’s a real difference between letting AI produce outputs unchecked and thoughtfully reviewing and taking accountability for those outputs – and that’s a key job of the modern data scientist.
Importantly, securing your resilience as a data scientist requires you to act fast, given the eye-watering pace of technological change. When Excel transformed accounting, professionals had years to adapt. When generative AI emerged, the disruption happened in months. The World Economic Forum estimates that 50% of employees will need reskilling by the end of 2026. Data professionals are no exception. In fact, given their, exposure to AI, their transformation is arguably most urgent.
Ultimately, the question of job security for any data scientist comes down to value creation. If your contribution can be summarised as “I build models,” you are competing directly with increasingly capable AI systems. If your contribution is “I understand complex business problems, deploy appropriate technical solutions, communicate findings to stakeholders and ensure ethical implementation,” you are offering something that remains distinctly human and highly valued by business.
Understand that while the tools have changed, the human capacity (and business need) for judgement, creativity and contextual understanding hasn’t. So, while the robots aren’t going to take your data science job, complacency might.
Written by Prof. Yudhvir Seetharam, Head of Analytics, Insights and Research at FNB Business