How to plan for a capability that does not yet exist

How to Think About AI — Edition 5

There is a dangerous assumption sitting underneath most AI strategies in surveying. That the AI of today is roughly the AI of tomorrow. Faster, perhaps. More accurate. But fundamentally, the same thing is doing fundamentally the same work, with the same limitations.

The assumption is wrong, and the strategies built on it inherit the error.

The AI that firms are planning around is the AI that exists now, in mid-2026. It is genuinely impressive. It can also do a fraction of what it will be able to do in three years, and a fraction of a fraction of what it will be able to do in ten. Planning a firm’s response to AI by studying what AI can do today is like planning a journey by studying the road behind you. The information is real. It is also the wrong information.

The trajectory matters more than the snapshot

There is a temptation, when looking at any new technology, to focus on its current limitations. AI today still hallucinates. It still gets context wrong. It still cannot consistently interpret a complex building from photographs alone.

These are real limitations. They are also not strategically useful.

Every honest assessment of AI right now should carry a footnote. Today’s version is the worst version that will exist.

The model running in your browser today is already obsolete in the lab. Limitations being discussed in podcast interviews are often solved by the time the episode is published.

The useful question is what AI is on a clear trajectory to do, and what the firm’s position should be when it gets there.

Planning when the future is genuinely uncertain

Here it gets harder.

No one knows exactly where AI will be in five years. Not the people building it. Not the people predicting it. The capabilities arriving next are unlike anything that has come before, which means historical patterns can guide thinking but cannot deliver certainty. Anyone selling a precise forecast of what AI will be doing in 2030 is either misleading you or has not thought hard enough about the question.

This is the part firms find paralysing. If the future is genuinely uncertain, how can a firm plan for it?

Strategic planning under genuine uncertainty looks different from planning under known conditions. It does not produce a five-year plan with milestones and targets. It produces a posture. A way of being positioned that can respond to multiple futures, rather than betting the firm on a single forecast.

Four principles separate the firms doing this well from the firms doing it poorly.

Build optionality, not commitment

Invest in capabilities that pay off across multiple possible futures.

A firm that retrains every surveyor around a specific AI tool is committing. If the tool turns out to be the right one, the firm has won. If it turns out to be a transitional product, the training budget is gone.

A firm that builds general AI literacy across its team, how to think about AI capabilities, how to evaluate tools, and how to integrate them while maintaining professional judgement, is building optionality. Whichever tools dominate, the firm has people who can use them well.

The same principle applies to client positioning, service design, and pricing. The strongest position holds up under several possible futures. The weakest depends on a specific future arriving as planned.

Watch the trajectory, not the current state

The default approach to evaluating AI tools is asking what they can do today. Useful for procurement. Useless for strategy.

The better question is what these tools are on a clear path to doing within eighteen months. That is the strategic planning horizon that matters. Long enough to make decisions count, short enough that the trajectory is visible. If a tool is 70 per cent reliable today and the underlying capability is improving by 10 percentage points a year, plan for the world in which the tool is 90 percent reliable.

Identify the parts of the work that are trajectory-proof

Some of the professional values are uncopyable. The judgement is built from years of work with specific buildings in specific local conditions. The presence is required when a client has just discovered something serious about a property they were about to buy. The accountability that comes with being a regulated, insured, named professional whose name sits on the report.

These are the load-bearing parts of professional value. They are not affected by the trajectory of AI capability, because they are not the kind of thing AI is on a trajectory to provide. A firm that has identified these parts of its work and built its identity around them is in a strong position, whatever the technology does. A firm that has not done this work is in a weaker position, because when the routine work contracts, the firm has nothing distinctive left.

Act before the picture is clear

The hardest one.

Firms are trained to make decisions when they have enough information to be confident. AI does not let firms do that. The information needed to be confident will arrive too late to be useful, because by the time the picture is clear, the strategic options have already narrowed.

The discipline is making smaller bets more frequently rather than large bets occasionally. Learning from each one. Adjusting. Acting with imperfect information now is the alternative to acting with even more imperfect information later, at higher cost.

What this means in practice

If strategy conversations in a firm are dominated by questions like “what AI tool should we adopt?”, or “when will AI be ready for our work?” or “how much should we invest in AI training this year”, the firm is asking the wrong questions.

The right questions:

What capabilities should we build in our people that pay off regardless of which specific tools win?

What is on a clear trajectory to change in our work over the next eighteen months, and what would our firm look like if it had already changed?

What are the parts of our work that no foreseeable AI capability will be able to provide, and have we built our identity around them?

What can we start doing now, with imperfect information, that gives us a better position when the picture clarifies?

These questions do not produce a five-year plan. They produce a posture. A way of operating that is responsive to the trajectory rather than dependent on a specific forecast.

That is what strategy under genuine uncertainty actually looks like.

This is the fifth edition in a six-part series on how to think about AI in surveying. Each Monday, I set out one principle that is challenging the profession, and what to do about it.

Next Monday: why Governance is the strategic question, not the compliance one. The final edition.

Nina

The next AI in Surveying Briefing – The vendor questions that every surveyor is entitled to ask.

My next free AI briefing for surveyors is on 15th July, 1-2pm. A practical hour on where your data goes when AI tools handle it, what the RICS Professional Standard requires of your firm, and the questions you can put to any supplier in writing.

Register here →

Ready to put a framework in place?

The RICS Standard on Responsible Use of AI in Surveying Services sets 30 mandatory requirements. Most firms know the Standard exists. Far fewer have a working framework for meeting it.

GUARD is the only AI governance framework mapped directly to those 30 requirements. It has now been delivered to over 550 surveyors. The next workshop is on 14th July.

The session takes you through Governance, Use, Accountability, Risk and Documentation, shows you exactly how each element maps to the Standard, and leaves you with a working approach you can apply to your own firm.

Book your place →