Don't Blame AI
In my work I meet leaders constantly, and the most frequent thing I hear from them is: 'I want to take advantage of AI', usually in the hope of reducing the time people spend on their processes.
It's understandable. Every conference, LinkedIn article, technology vendor and consulting firm is telling organisations that AI will improve productivity, reduce costs, accelerate decision-making and unlock new sources of value. In most cases, that's true. AI has enormous potential and there are undoubtedly opportunities for organisations to benefit from it.
My master’s degree is in artificial neural networks, the foundation of modern AI, and yes AI is doing amazing things that we never dreamed of back in the 90’s, but there is a fundamental misunderstanding that it is a panacea. We just put it in and all our troubles are gone.
My concern isn’t the AI itself, it’s the solutionising of the problem. We’ve jumped to the answer rather than understanding the question first.
Too many organisations are beginning with a technology and then looking for a problem to apply it to. They see competitors talking about AI, boards asking questions about AI strategies and vendors demonstrating impressive capabilities. The immediate assumption becomes that AI must be the answer, rather than wondering what question it would be solving.
AI won't fix poor data quality. It won't fix inconsistent processes. It won't fix unclear accountabilities, weak governance, conflicting business rules, a lack of organisational discipline. The list goes on. Yet these are precisely the issues that often emerge when organisations start discussing automation and artificial intelligence initiatives.
So, where to start?
Data
Almost every leader I speak to has legacy data they don't fully trust. Every AI solution relies on data, and a consistent version of the truth is rarer than it should be. The same customer exists multiple times across different systems. Asset information varies depending on where it is sourced. Financial and operational data are structured differently between departments. Even basic definitions can vary: is Period 1 January or July?
Organisations often assume AI will somehow overcome these issues. It won't. AI does not create certainty from inconsistency. It simply processes inconsistent information at greater speed and scale. If your data is unreliable today, adding AI merely increases the likelihood that poor information will influence more decisions more quickly.
AI can certainly help here. It can find duplicates, flag anomalies and surface errors much faster than a manual review but finding a problem isn't the same as fixing it. Governance needs to be in place to agree the definitions, decide which system is the source of truth, and clean up the legacy records. A practical starting point would be to look at your critical datasets, agree a single definition for each key term and name an owner for each. It's unglamorous, cross-departmental work, which can be time-consuming for internal teams already stretched by day-to-day operations.
Process
AI is often viewed as a way to make a process faster, but a faster process isn’t necessarily better. Checking to see whether it is producing the right outcome in the first place is the better place to start. There is little benefit in automating activities that are poorly designed, unnecessarily complex or no longer aligned with the organisation's objectives.
Consider a situation where two systems produce conflicting information. AI may be capable of identifying the discrepancy within seconds, but the real challenge lies in the response. What action should be taken? Who owns the decision? What thresholds apply? Are there regulatory obligations to consider? What level of risk is acceptable?
Those are business decisions not technology decisions.
Organisations frequently rush towards automation without first establishing the operating framework that governs how decisions should be made. AI can accelerate a process, but it cannot design one. If the business has not clearly defined how work should flow, who should make decisions, and how exceptions should be handled, introducing AI simply accelerates confusion.
People
The theory is that AI removes people from the process. The reality is that it removes some people from the production end but adds to the review end. Fewer people are needed to create the output but more senior people must review or decide on what to do. I've already seen AI-generated material produced faster than executives can find the space in their diaries to review it. Without an HR strategy to support this shift in skills, organisations risk replacing some people and swamping the decision makers. The bottleneck hasn't gone, it has moved up the organisation.
Do I hate AI?
Absolutely not, it has the potential to be amazing and change how we work in the future. The speed of change will, I’m sure, lead us into heretofore unexplored or unimagined ideas.
So don't blame the AI. It will do exactly what it was built to do, which was to process what it was given, faster and at scale.
The better question isn't how to speed up, it's whether what you're speeding up is what you actually want done.