Why Better Questions Matter More Than Bigger Models
Artificial intelligence has created a fascination with answers.
Larger models, faster processing and increasingly capable systems are expanding what machines can achieve. Across industries, organisations are exploring how AI can analyse information, automate tasks and support complex decision-making.
These advances are significant.
However, before any answer can be useful, there must first be a meaningful question.
Throughout engineering history, progress has rarely come from technology alone. It has come from understanding a challenge deeply enough to know what needs to be measured, what evidence is required and which tools are appropriate.
Artificial intelligence is no different.
The question is not simply:
What can AI do?
The more important question is:
What problem are we actually trying to understand?
Because the quality of any solution is shaped long before an algorithm is selected.
It begins with the question that comes first.
Engineering Begins with Curiosity
Every engineering project begins with curiosity.
A structure changes.
A material behaves unexpectedly.
A process becomes inefficient.
An environment presents a new challenge.
The engineer's first response is rarely to select a technology. It is to investigate.
Why is this happening?
What evidence do we have?
What information is missing?
How confident are we in what we know?
These questions guide the entire process that follows.
They determine what data is collected, how observations are interpreted and how success is measured.
This approach is fundamental to good engineering.
The role of technology is not to replace this thinking. It is to extend it.
Artificial intelligence is a powerful tool for identifying patterns, analysing complex information and supporting decisions. However, it cannot determine which questions are worth asking.
That remains a human responsibility.
Bigger Models Cannot Solve the Wrong Problem
Much of the current conversation around AI focuses on scale.
Bigger models.
Larger datasets.
More computational power.
These developments have driven remarkable progress, but scale alone does not guarantee useful outcomes.
An AI system can analyse millions of images, process vast amounts of information and identify patterns with impressive speed.
But if the information being analysed does not answer the right question, the result may simply be a faster route to the wrong conclusion.
Engineering has always recognised this challenge.
The objective is not to gather the most information possible.
It is to gather the right information for the decision that needs to be made.
For example, an inspection system designed to identify every possible change within an asset may produce large quantities of information.
However, an engineer may need a different answer:
Which changes indicate deterioration?
Which areas require intervention?
What should be prioritised?
The question defines the system.
It influences the data captured, the analysis performed and the decisions supported.
Better questions create better outcomes.
Questions Shape Evidence
A common assumption in AI is that more data automatically leads to better results.
In practice, the relationship is more complicated.
Data becomes valuable when it is collected with purpose.
A question determines:
what information is needed
how it should be captured
what accuracy is required
what uncertainty must be understood
how the results will be interpreted
This is particularly important when AI systems are applied to the physical world.
A photograph, measurement or 3D model is not simply a collection of information.
It is evidence about something real.
The usefulness of that evidence depends on context.
Where was it captured?
When was it captured?
Under what conditions?
How reliable is it?
What limitations does it contain?
These questions are not barriers to innovation.
They are what make innovation trustworthy.
The strongest AI systems are not built on the assumption that all data is equally valuable.
They are built on a clear understanding of what evidence is required and why.
Human Expertise Remains Essential
As AI capabilities grow, there is sometimes a perception that human expertise becomes less important.
In reality, the opposite is often true.
The more complex the problem, the more important context becomes.
A structural engineer understands the significance of a crack beyond what can be measured.
A heritage specialist understands why a material change matters beyond what an image classification system can identify.
An emergency responder understands the operational realities behind a dataset.
AI can support these experts by processing information at a scale and speed beyond human capability.
But it cannot replace the experience, judgement and responsibility that make information meaningful.
The most effective systems are therefore not designed around replacing expertise.
They are designed around augmenting it.
Human understanding and machine capability are strongest when they work together.
Better Questions Enable Better AI
The most successful AI projects are rarely those that begin with the technology.
They begin with understanding.
Before developing a model, organisations need to consider:
What decision are we trying to improve?
Who will use the information?
What evidence is needed?
How will uncertainty be communicated?
What does success look like in practice?
These questions influence every stage of development.
They shape data collection.
They shape system design.
They shape evaluation.
They shape deployment.
This is why hybrid approaches are becoming increasingly important.
Combining AI with domain expertise, physical understanding and human oversight creates systems that are more adaptable and more trustworthy.
The future of AI is unlikely to be defined by technology alone.
It will be defined by how effectively we combine technology with understanding.
Curiosity as a Competitive Advantage for SMEs
For smaller organisations, the AI landscape can sometimes appear dominated by companies with access to enormous datasets, extensive computing resources and large research teams.
However, SMEs have a different advantage.
Understanding.
Many SMEs possess deep knowledge of specific industries, processes and customer challenges.
That expertise is invaluable.
It enables organisations to ask better questions.
It allows them to identify meaningful problems rather than simply applying technology because it exists.
It helps create focused, practical solutions.
In many cases, the advantage is not having the largest dataset.
It is knowing which information matters most.
Curiosity, therefore, is not simply a creative quality.
It is an engineering capability.
Engineering Better Understanding
At Aralia, we believe technology is most valuable when it improves understanding.
This principle sits behind work across AI, 3D imaging, digital heritage and disaster resilience.
The goal is not simply to create more advanced systems.
It is to create systems that help people better understand complex environments and make more informed decisions.
Engineering better understanding begins with engineering better questions.
Final Thought
Artificial intelligence will continue to evolve.
Models will become more capable.
Computing power will increase.
New approaches will emerge.
But one principle will remain constant:
The quality of the solution depends on the quality of the question.
The organisations that gain the greatest value from AI will not necessarily be those with the biggest models.
They will be those that understand their challenges deeply, gather meaningful evidence and ask the questions that truly matter.
Technology can provide powerful answers.
Engineering determines which questions are worth asking.
