Ask a hospitality operator whether they are using AI and they may start talking about the technologies they are considering for the future. But – for many – AI is already embedded in the vast majority of the systems they use every day.
From the demand forecasting tools, to the payment systems and marketing platforms, AI often comes in the form of another feature within existing software.
Therefore, hospitality leaders should be looking at the AI tools already influencing their operations, and whether they are fully in control of them.
Seeing what humans can miss
Used well, AI has enormous potential for hospitality, particularly when it comes to identifying patterns across volumes of data that would be impossible for an individual manager to process.
As an example, a site manager may know that Friday lunchtime is busy. Technology can potentially identify that the peak has slightly shifted, or that demand for a particular dish has changed.
It can also help with improving forecasting, reducing waste, identifying stock anomalies and enabling more relevant customer offers.
But there is an important distinction to flag. AI might identify what has changed, but it it is very unlikely to understand why it changed. That still requires people who understand the operation, the customer and the overall context.
The danger of the black box
One of my biggest concerns is not AI itself, but opacity. If technology recommends a decision is taken, how easily can an operator distil that recommendation and understand why?
Take pricing and promotions, as an example. If a system influences which customer receives an offer, operators should understand the basis for that decision. Otherwise, unintended outcomes may happen – without anybody consciously choosing them.
And the stakes become higher when technology affects vulnerable people. Hospitality technology increasingly operates across settings such as schools, hospitals and care environments. Decisions involving allergens, meal entitlement or payments cannot simply disappear inside a black box.
There is also the danger of creating operational issues. A forecasting model based on historic trading patterns can continue to produce what an operator will perceive as being a solid recommendation, even when customer behaviour changes. But because the output looks precise, they can stop questioning it.
A black box can hide mistakes, and by doing so give them authority.
Not every decision needs AI
Technology can help identify ingredients and alert staff to potential risks. But where a decision could affect someone’s health or safety, there must be clear human accountability.
Operators need to ask: Would AI make this better – and can we still understand and work with the outcome?
Sometimes conventional technology may be the better approach to take. Rules-based processes remain important where consistency, predictability and auditability matter. I’d argue that using AI simply because it is available isn’t innovation.
Being in control of AI is a design choice
AI shouldn’t be something you bolt onto a system afterwards. Operators should understand why recommendations have been made, have appropriate opportunities to intervene and know where the boundary lies between what technology can do independently and what requires human approval.
Our own ISO 42001 accreditation has reinforced the importance of accountability. Organisations need to understand what an AI system is designed to do, the information it uses, the risks involved and, ultimately, where responsibility sits. Increasingly, customers will expect that same level of clarity from their technology suppliers.
That thinking has helped shape LollySense, our framework for the responsible, strategic and sustainable deployment of AI. It ensures that every AI feature we introduce is purposeful, solving a genuine customer problem; practical, informed by both quantitative data and real-world insight; planet-conscious, considering the wider impact of the technology; and, importantly, ethical in how it is designed and applied.
For us, responsible AI isn’t simply about what the technology is capable of doing, but whether it should be doing it – and whether we can remain accountable for the outcome.
AI won’t be the competitive advantage
Over the next few years, AI capabilities will become commonplace across all forms of hospitality technology.
The competitive difference will be how intelligently businesses apply it. And asking themselves:
- Do we have the right data?
- Do our people understand what the technology is doing?
- Do I know which decisions should be automated and which shouldn’t?
- Can we challenge a recommendation and explain an outcome?
In this instance, there is a parallel with the early web. Having a website stopped being a competitive advantage remarkably quickly. Knowing what to do with it didn’t.
The future will be focused on designing solutions around the customer and understanding where technology genuinely adds value.
The strongest solutions will be those which bring together hardware, software, service, data and compliance – with people retaining appropriate oversight throughout.
For hospitality leaders, I would suggest one simple rule: Never allow a system to make an important decision that you couldn’t explain to the person it affects.
Because ultimately, the most important thing AI needs to earn in hospitality is trust.
