What if the CEO asks me about... using Artificial Intelligence in our Retail stores?

The term "artificial intelligence" (AI) describes a range of technologies that enable machines to perform tasks that typically require human intelligence, such as the ability to comprehend, act and learn.

What are the upsides of AI in retail?

The retail and consumer industry is in an ideal position to capitalise on the benefits that AI can offer, with your customers providing you with vast amounts of information about themselves, their lives, their likes and shopping habits.

AI can enable you to pull together this information, often from disparate sources, and analyse and interpret the data to your advantage. It is therefore not surprising that the retail and consumer industry has already embraced AI with many successful use cases, including:

• improving demand forecasting;
• increasing accuracy in inventory management;
• automating CRM marketing activities;
• customising promotions and product recommendations; and
• optimising product placement.

Through these sorts of applications, AI can assist you in increasing profits by encouraging sales of products, improving customer service and strengthening customer loyalty. In addition, AI can lower costs through supply chain and operational improvements and optimisation, and potentially support your organisation's sustainability and environmental objectives through better demand management and waste management.

In these difficult economic times, adopting AI to generate revenue and lower costs seems like an easy choice to make, but before launching into an AI project, it is important to weigh up any potential risks from the intended use of AI.

What to consider before using AI?

1. Ethical use of AI

AI is still on the cutting edge of technology, and the ethical use of AI is a potential obstacle.

AI ethics is a set of values, principles and techniques using generally accepted standards of "right and wrong" to guide moral conduct throughout each stage of the development and use of AI. They are intended to not only motivate morally acceptable practices but also to detail the duties and obligations necessary to produce ethical, fair and safe AI technologies.

This is a very broad topic, but at a high level, one of the concerns is that AI should only use fair algorithms, and include ways of measuring and mitigating bias. Additionally, each step in the AI process should be transparent, justified and lawful.

The clear driver for ensuring only ethical AI is deployed is to avoid harm. In a retail and consumer context, it will also be important for you to maintain customer trust and to protect your brand and reputation. Where AI directly interfaces with customers, customers will be very aware when something goes awry, for example through inappropriate or insensitive product suggestions. Once broken, consumer trust is difficult to rebuild, so particular care should be undertaken when considering customer-facing AI.

2. Legislative Backdrop

Legislation will continue to struggle to keep pace with the rapid development and deployment of AI. In the UK, the Government has been reluctant to consider any overarching AI-specific regulation. To date, they have instead allowed sector-specific regulators to consider the impact of AI in their respective sectors and whether additional regulation or guidance is required.

So one of the first things you need to do before implementing an AI solution is to check that the intended use of the AI in question does not fall foul of the legislative framework. And where you are licensing or procuring AI from a third party, allocate responsibility for legal and regulatory compliance in respect of the various AI components.

Bear in mind that the underlying driver of AI is data, much of which might also of course be personal data in the context of your business, for instance your customers' personal details and financial information. Ensure that you have in place the systems and processes required in order to process all data (and in particular personal data) lawfully and in accordance with any other relevant contracts. As a good starting point, the ICO has published an AI and data protection risk toolkit which lists the key considerations of the use of personal data in the context of AI, and steps to take to assist with compliance with data protection legislation.

For a more detailed look at the regulation of AI, please check out this article.

3. What approach should you take to AI?

It is critical to do your homework on the vendor and the AI product in question and not simply be swayed by the sales pitch.

We have seen in the past a number of cases where AI has picked up data that was not specifically intended to be an input into the AI's algorithm, which resulted in some unintended consequences, such as variable pricing for different demographics. To address this:

• Ensure that you have a clear and agreed approach on data used to train the AI – this can be through controlling the data that is used or have a clear specification on the data that a supplier can use but in either case you will want to ensure as best you can that the data set is not biased itself, and doesn't contains gaps that might produce unintended results.
• Make sure that your supplier is clear about, and you understand, the approach your AI uses - so that you can explain what it does to your own customers and management.
• Consider how you can verify the data that has been used and what tools are available to confirm that the AI does what is intended and can identify any anomalous results. This might be through audit rights or verification tools provided by your supplier or right to have an independent review or using other third party tools to make that assessment.

It is also advisable to develop clear policies around AI and its appropriate use in your business – this can set out minimum requirements which any AI used by the organisation must comply with. Of course this will only work if you monitor these policies, and the AI in use, regularly to ensure that it is being used in the way that meets these requirements.

As with any procurement, look at the areas of your business which might be impacted, or which might impact upon, your use of AI. For example:

• Will your insurance cover apply where AI is used?
• Will the AI be used with other software, and if so, do you have the appropriate rights and licences to do so from relevant third-party software vendors? Is the use and/or commercial parameters of the other software still appropriate when being used with the AI given, amongst other things, automated processing?

One thing is for certain, AI can be a game changer but it does require more careful consideration on how it works than with traditional software products.
 
Disclaimer: The information in this publication is for guidance purposes only and does not constitute legal advice. We attempt to ensure that the content is current as at the date of publication, but we do not guarantee that it remains up to date. You should seek legal or other professional advice before acting or relying on any of the content.