AI Use Case AI Ethics Other Industries Perspectives

Published:
Est. reading time: 8 minutes
Author: Steph Organ

Artificial Intelligence (AI) is already beginning to transform several industries, and it continues to divide opinions on whether it will transform our lives for good, for bad, or exaggerate existing divides by benefitting some and neglecting others. One of the areas where AI has been applied with extremely effective results is in retail, both online and off, and is a fine example of why AI can be simultaneously good and evil, highlighting the need for regulation.

Support productivity

Work smarter with low-code tools

Get our automation toolbox

Steph Organ

Ambitious digital marketer with a love for words and history of content creation.

More

Artificial Intelligence (AI) is already beginning to transform several industries, and it continues to divide opinions on whether it will transform our lives for good, for bad, or exaggerate existing divides by benefitting some and neglecting others. One of the areas where AI has been applied with extremely effective results is in retail, both online and off, and is a fine example of why AI can be simultaneously good and evil, highlighting the need for regulation.

Shopping is already an activity that can be easily abused i.e. with “retail therapy” whereby buyers purchase items as a distraction to boost their mood in the short term, but due to short-lasting effects, this can become addictive. Another serious issue associated with shopping is Compulsive Buying Disorder, which can have many different causes. But the majority of us go through the ritual of buying things we don't need, often in relation to that hit of serotonin. So how do AI and shopping come together to create a runaway experience? It has partly to do with why AI has been so effective in this industry.

Why is retail such a prime setting for AI?

The current surge in AI and accompanied hype can be equated largely to two things; a huge increase in computing power that is required to run and train machine learning algorithms, and an exponential increase in rich, real-time data to feed said algorithms. Such data is particularly rich in the world of retail, where AI systems are able to analyse the data, recognise patterns and make predictions far more effectively than a human ever could.

But an additional factor that makes retail such a thriving environment for AI, is the margin for error. A retail algorithm can be deployed and tested in action long before it is has been perfected, as an error would not be detrimental. It can learn on the job, so to speak, recognising what has worked, and constantly improve itself. Someone not clicking on an ad is a small lesson, not a costly mistake, whereas other AI systems may have far more at stake over the outcome of just one decision (take self-driving cars as an example), making progress in other domains seem much slower.

Several sections of retail lend themselves to AI, from handling customer interactions and brand reputation to creating an engaging user experience, perfecting marketing, and more.

Customer service

Retail companies have been able to jump on general AI trends like chatbots to improve the customer experience and build trust by offering 24-hour support in addition to cutting costs, as well as using social listening and social media scraping to monitor how customers are feeling and resolve any issues that they are expressing to protect their brands reputation. In fact, most companies can make use of these tools to strengthen their brand and protect its reputation, but the retails industry has a few more tricks up its sleeve that we’ll look into now.

User experience

Keeping your users happy has always been key to improving that customer lifetime value, and e-commerce brands are approaching this by creating a more engaging experience in a number of ways. AI can be applied to create more intuitive search functions that bring up more relevant results through the application of natural language processing, or even employing visual search, such as Pinterest’s Lens feature or ASOS’s style match, that can recognise items in photos and suggest similar products.

Being able to search and actually find what you want is great, but accurate recommendation systems could be even better. You don’t even have to search to be shown things that you ultimately want. Many types of companies use recommendation systems, from retailers through to media. Machine learning algorithms feed off rich data that reveal patterns such as what actions a person is likely to take based on the actions that other users with similar data took. As the algorithm collects more data and adjusts, accuracy of personalisation, and hence satisfaction, increases.

The user experience can be optimised in other ways, too. For example, online clothes brand Misguided responded to their customer data and after noticing that most of their users shopped via mobile, specifically using an iPhone, they decided to improve their mobile app and added unique and relevant features like Apple Pay as a function. Artificial Intelligence can play an important role in optimising the user experience assisting UX designers and increasing creative possibilities.

Users feel closer to a brand when optimisation and personalisation are so fine-tuned, as if the brand really knows and understands them. This brings us on to our next topic.

Personalised marketing

Marketers are well aware of how personalisation increases engagement, with 72% of consumers now only responding to personalised adverts. So just as with optimised user interfaces, marketing thrives on plentiful data to create hyper-personalised deals. Everywhere we shop, whether its online fashion websites or grocery stores with loyalty cards, or through location tracking on our phones, our behaviours are logged and turned into data that can be used to make powerful predictions and personalise marketing efforts.

For example, your grocery loyalty card can pick up patterns in your buying behaviour, like how often you stock up on your cupboard staples, and present you with personalised offers, perhaps when it calculates you are due to run out, and personalised rewards that keep you buying with the same company. It might also combine similar customer’s data and suggest new products that other customers go for when purchasing some of your regular items. Store data can also be used at scale to streamline the inventory processes by picking up subtle locational patterns or up and coming trends and stocking for them.

This seems pretty useful when it comes to groceries, but these methods are reflected throughout the rest of the retail world, where purchases are generally far less vital. We’ve all conduct a Google search to find that suddenly all our online adverts are trying to sell us that specific thing. Since many websites and apps now share data, your online activities can be traced and manifest themselves in highly targeted offers that are difficult to resist.

There is some debate about whether or not Facebook listens to your conversations, applying Natural Language Processing (NLP) to process the topics you discussed and generate explicitly targeted adverts. It can seem questionable that something so specific could come up by chance, and many experiments have been done to support this theory, however Facebook has repeatedly denied this accusation. They claim that their predictive models are just highly accurate, which is also plausible.

Predictive pricing

Another influential factor in someone’s buying decision is the price. Predictive pricing can help marketers push the right promotions to the right people at the right time in order to yield the biggest benefits. Studies show that more half of mass promotions do not break even. Predictive pricing ends that, ensuring that offers are temping enough to certain people, without giving away more than necessary.

Analysts must take into account multiple factors at once to get pricing dynamics right, which is why a good machine learning algorithm can do it so much better than a human. An algorithm’s ability to analyse multiple factors at once, such as seasonal trends, competitor’s prices, and individual buying habits of a customer, means it consistently gets it right, or at learns and improves from the cases where it got it wrong.

What about the ethics?

We’ve seen the many different ways that retailers are capitalising on artificial intelligence and the amazing ways that it can boost their revenue, but what about how it affects the population of consumers? Data has become a powerful weapon, collected from the masses and now used against them to influence their own decisions.

With the abundance of data now available around our shopping habits and preferences, marketing campaigns become painfully relevant and hard to resist, resulting in huge wish lists, or in some cases an accumulation of repeat buying “offences”. Consumers get tricked into buying more things that they can’t afford and don’t really need, getting carried away and not realising the overall impact when spending is split across multiple smaller purchases.

In China, shopping addiction is already a trend, and Gartner predicts it to become so much of an issue in western civilisation that retailers will soon be forced to take more responsibility around these exploitative practices, just as casinos and gambling sites must issue warnings around playing responsibly.

What is your take on this? Is AI in retail working wonders in connecting brands to their target customers and helping consumers easily find the things that they want, or is it adding to the problems of society by serving up easy fixes to unsuspecting consumers? What kind of regulations do you think should evolve as AI and retail integrate to create such a personalised and addictive experience?

Don’t forget that Nightingale HQ can help you find data and AI consultants to work with to help improve your business, guide you through processes and help you comply with the laws that are slowly adapting around data.

Support productivity

Work smarter with low-code tools

Get our automation toolbox