Why Most Companies Are Hiring the Wrong AI Expert
The rush to hire AI strategists has become the latest corporate obsession, but industry veteran Koo Ping Shung warns that most businesses are putting the cart before the horse. After two decades in data science and artificial intelligence, Koo argues that the majority of companies lack the fundamental infrastructure needed to make AI strategic decisions meaningful.
The reality is stark: without established data management processes and optimised reporting systems, an AI strategist becomes an expensive consultant with little to strategise about. The solution isn't necessarily what you'd expect.
The Foundation That's Missing
Before any business considers hiring an AI strategist, three critical prerequisites must be in place. These aren't glamorous requirements, but they're non-negotiable for successful AI adoption.
Data infrastructure forms the backbone of any AI initiative. Companies need a centralised data warehouse where all datasets are stored and managed consistently. Without this single source of truth, AI projects become fragmented exercises in futility.
Data management processes ensure quality and security. This includes identity and access management protocols that protect sensitive information whilst enabling authorised access. Poor data governance kills AI projects before they begin.
Data quality measurement builds stakeholder confidence. Established metrics demonstrate that reports, insights, and analyses can be trusted. Without measurable quality standards, AI outputs remain questionable at best.
"AI is just one tool in the data toolkit. You can see AI as that smart shiny drill that can auto-change the drill bit based on your needs, but not all your business challenges are about drilling." - Koo Ping Shung, Data Scientist, Data Science Rex
The reporting prerequisite often surprises business leaders. Optimised reporting processes aren't just about generating monthly summaries. They represent organisational maturity in understanding data nuances, limitations, and practical applications.
The AI Strategist Paradox
Here's where the logic breaks down: AI strategists are incentivised to find AI solutions, even when simpler alternatives exist. This creates what Koo calls the "hammer and nail" problem applied to artificial intelligence.
Consider the typical AI strategist's mandate: identify areas where AI can be implemented, propose necessary changes, and integrate AI into business processes. The inherent bias towards AI solutions becomes problematic when a simple average calculation could solve the same business challenge more efficiently and cost-effectively.
The hidden costs of AI implementation are substantial. Short-term expenses include proof-of-concept development, design planning, data management setup, and model training. Long-term costs encompass maintenance, continuous monitoring, model validation, cloud computing subscriptions, and retaining essential skills within the organisation.
By The Numbers
- Over 70% of AI projects fail to move beyond the pilot stage due to inadequate data infrastructure
- Companies with optimised data reporting are 5x more likely to succeed with AI implementation
- The average cost of maintaining an AI model in production exceeds $50,000 annually
- Only 15% of businesses currently have the data management maturity required for strategic AI adoption
- Data strategists with comprehensive skill sets command 40% higher salaries than AI-focused specialists
Understanding what every worker needs to answer about their non-machine premium becomes crucial when evaluating whether AI solutions truly add value or simply automate processes that don't need automation.
What You Actually Need Instead
The solution isn't an AI strategist but a data strategist with comprehensive expertise across the entire data lifecycle. This professional understands that artificial intelligence represents just one option among many data tools available to solve business problems.
A qualified data strategist brings five essential competencies:
- Data management and quality assurance expertise
- Advanced data analytics and analysis capabilities
- Machine learning knowledge including model deployment and monitoring
- Understanding of generic business processes
- Change management experience for data-driven transformations
These professionals are rare because they require both technical depth and business acumen. Unlike AI strategists who focus narrowly on artificial intelligence applications, data strategists evaluate the full spectrum of data solutions to recommend the most appropriate approach for each business challenge.
"Data will be the new normal, and taking advantage of data will be what good business leaders will constantly be thinking about. There are multiple tools with varying value derived from data that businesses can leverage." - Koo Ping Shung, Data Scientist, Data Science Rex
The strategic advantage lies in having someone who won't automatically default to AI solutions when simpler alternatives exist. This perspective becomes increasingly valuable as companies build their own agentic AI systems without understanding whether automation is actually necessary.
| Role Focus | AI Strategist | Data Strategist |
|---|---|---|
| Primary Mandate | Find AI applications | Optimise data value |
| Solution Bias | AI-first approach | Best-fit approach |
| Technical Scope | Machine learning focus | Full data lifecycle |
| Business Impact | Innovation theatre | Measurable ROI |
| Long-term Value | Technology dependent | Process dependent |
Companies exploring how AI agents can transform their business should first ensure they have the foundational data capabilities to support such advanced implementations.
Making the Right Hiring Decision
The title "AI Strategist" has become a marketing tool rather than a meaningful job description. When evaluating candidates, focus on experience and background rather than impressive titles. The right data strategist should demonstrate practical success across multiple data disciplines.
Experience with small business AI adoption challenges often provides more valuable insights than theoretical AI strategy frameworks. Real-world implementation experience reveals the practical constraints and opportunities that academic approaches miss.
Beware of consultants who promise revolutionary AI transformations without first auditing your existing data infrastructure. These individuals often create expensive proof-of-concept projects that never scale because the foundational elements weren't properly established.
Do I need an AI strategist if my competitors are hiring them?
No. Competitive hiring decisions often reflect industry trends rather than business needs. Focus on whether you have the data infrastructure and reporting processes necessary to support AI initiatives before considering specialised AI roles.
How can I tell if a data strategist is qualified for AI oversight?
Look for hands-on experience with machine learning model deployment and monitoring, not just theoretical knowledge. They should understand both the technical requirements and business implications of AI implementation across different organisational contexts.
What's the minimum data maturity level needed for AI projects?
You need centralised data storage, established quality metrics, and optimised reporting processes that stakeholders trust and use regularly. Without these foundations, AI projects typically fail during the transition from pilot to production.
Should startups hire AI strategists or data strategists?
Startups benefit more from data strategists who can build scalable data foundations whilst identifying immediate opportunities for competitive advantage. AI-specific strategy becomes relevant only after achieving basic data operational excellence.
How do I evaluate whether my business problems actually need AI solutions?
Start with your desired business outcome and work backwards. If traditional analytics, simple automation, or process improvements can achieve the same result more efficiently, AI adds unnecessary complexity and cost.
The path to successful AI adoption runs through data strategy, not AI strategy. Companies that recognise this distinction will build sustainable competitive advantages whilst their competitors struggle with expensive AI theatre. Understanding why AI transformations keep failing provides crucial context for making informed hiring decisions.
What's your experience with AI versus data strategy roles in your organisation? Have you seen the hammer-and-nail problem play out with AI-focused hires? Drop your take in the comments below.
Latest Comments (5)
The idea of a single data warehouse for all datasets, that's actually a pretty traditional view. In practice, especially with the scale of data a lot of companies are dealing with now, you often see distributed data architectures. Think data lakes, lakehouses, or even federated approaches where data stays closer to its source systems. Trying to force everything into one centralized warehouse can become a bottleneck and introduce its own set of data quality issues if not managed extremely carefully. It implies a single schema and governance model that just isn't realistic for diverse data types and sources.
This point about established data management infrastructure being a prerequisite for AI adoption is crucial, especially in resource-constrained contexts. Many nations in the Global South struggle with this foundational data governance.
While a single data warehouse is ideal in principle, for many Indic language NLP projects, we frequently encounter distributed, siloed datasets that require federated learning approaches rather than consolidation, which presents its own set of strategic data management challenges before any AI application.
Koo bringing up the "single place for all datasets" hit home. We went through a nightmare trying to integrate a new NLP model only to realize our customer data was fragmented across like four different systems. Totally bottlenecked the project for months. If weโd nailed that first, the AI part would've been way smoother.
spot on about the data infrastructure being key. seen too many projects stumble because the "single place" isn't really single, or the costs for scaling it become astronomical fast.
Leave a Comment