Many SMBs are already enthusiastic adopters of artificial intelligence, but others are taking a more cautious wait-and-see approach. Very often, it is anxiety about AI safety that is holding SMBs back – and even early adopters have their concerns. The good news is that there are simple steps every business can take to move past these issues with confidence.
By David Prosser
Concerns about safety, privacy and security are preventing many SMBs from embracing the potential of AI. Here’s how to address them:
What’s stopping small and medium-sized businesses (SMB) making use of artificial intelligence-powered tools? Research published recently by YouGov identified the biggest blocker as fears about safety and security. The market research group found 46% of SMBs in the UK were already using AI-powered tools or planned to; of the remainder, 49% said they were put off by nervousness about data privacy and security, while a further 19% cited ethical concerns.
That data chimes with Sage’s own research: it recently found that among SMBs that say they trust AI, 85% are now actively using the technology; among those where trust is lacking, that statistic drops to 48%.
They’re right to be worried. While AI has huge potential to benefit your business, the technology also comes with risks. The number of horror stories about organisations getting into trouble following their use of AI continues to grow, from the fast-food chain whose AI-powered ordering system served customers bacon-topped ice-cream to the police chief forced to step down after his force used erroneous AI conclusions to inform its approach to football safety.
The good news is that you can tackle these challenges head on. Here are seven steps SMBs can take to protect themselves while still taking advantage of AI opportunities.
1. Check the data you're putting in - is it accurate?
Technologists have a useful bit of shorthand for the problem with poor data: rubbish in, rubbish out. Most AI tools depend on information from your business to function effectively – anything from data on the sales you’re making to a set of instructions about the steps in a business process. If that information isn’t accurate, don’t expect the outputs from the AI tool to be so either.
That means you need to check you’re supplying your AI tools with high-quality data that is timely and correct. If you’re not confident in your information, better to leave it out of your AI model until you’ve verified it.
Aaron Harris, chief technology officer at Sage, thinks finance leaders in SMBs have a critical role to play in ensuring that any data going into an AI model is accurate and up to date. "Finance leaders simply won't be able to rely on blind trust, because in finance, 'almost right' is wrong,” he argues.
2. Interrogate the AI - do you understand how it works and what it does?
To be confident in the answers your AI tools produce, you need to understand how they were arrived at. That doesn’t mean you have to become an overnight expert on computer algorithms, but you should take some time to get familiar with the way that AI tools work.
"If a model is going to support financial decisions, it needs to be as transparent as a spreadsheet - and you should always be able to see how it got there," adds Harris. “Leaders should expect AI to earn trust the same way their teams do: by showing how it got there."
Some generative AI tools are often criticised as “black boxes”, where you can’t see exactly what’s going on inside the model. But with more basic AI tools, often SMBs’ first port of call, it should be much easier to understand the process; and even with GenAI, there will still be an explanation of process available.
Moreover, services providers should be willing to explain how they’ve embedded AI in products. For example, Sage’s AI Trust Label provides transparent, accessible information about how AI functions across Sage’s products
3. Set clear rules - run your AI in a way you're comfortable with
If there’s an area of your business where you have a particular concern about safety or security, don’t prioritise it for AI implementations; these can come later once you’re feeling more confident about the technology.
In the meantime, invest in training for someone in your business – you or another leader, perhaps – so they develop the skills to take control of AI implementations and operations. With the right training, they’ll know how to set guardrails for the AI tools your business uses; these provide clear rules for what such tools can and can’t do.
4. Think critically - apply some human oversight
Keeping humans in the loop is the best way to use AI safely. Your business will be full of people who know your organisation and the market in which it operates; they’re well-placed to question AI findings, particularly surprising conclusions that are clear outliers.
Indeed, it makes sense to treat every AI output as potentially incorrect until it has been verified. Double-check the conclusions and recommendations that tools reach, just as you would keep a close eye on the work of a junior employee you’ve just hired.
5. Keep privacy front of mind – only share information you're comfortable with
“Leaders must act as stewards, prioritising the responsible and ethical deployment of these technologies,” says Tomoko Yokoi, an AI and digital transformation expert who lectures at Swiss science university ETH Zurich. “Build stakeholder trust by addressing concerns such as data privacy, transparency, and ethical implications.”
Remember, data you share with AI tools – particularly GenAI models – will flow through into databases and processes managed by third parties. You need to understand who will have access to data you supply and in what form; how will it be anonymised, encrypted and secured?
You may also be dealing with personal data and other information where strict legal rules apply. If you’re not sure how those rules affect your business, the UK Information Commissioner’s Officer operates a dedicated SME hub and a telephone helpline offering advice.
6. Monitor returns on investment - are you getting payback?
One element of safety often overlooked is the financial impact on your business of AI investment. All implementations cost money, so you need to be sure you’re getting good value in return. That means monitoring data on the savings AI is generating for your business, or the revenues it is supporting.
Paula Whitehouse, professor of business growth and innovation at Aston University, points to “the availability of technologies enabling business leaders to access dashboards which show them how the business is performing on key measures.” She explains: “This enables leaders to not only plan for growth but monitor on a day-to-day basis whether they are on track to achieve it.”
7. Recommit to the cyber security – is your business risk-resilient?
Finally, it’s important to think about AI safety and security in the context of your business’s broader cyber safety. Now is a good time to review the protections your business has in place, from IT tools to employee training.
“Businesses with good cyber security practices better protect their operations, intellectual property and customers from cyber-attacks that can cause financial loss and business disruption,” points out Mia Haffety, policy manager for the digital economy at TechUK.
