How advances in AI will transform online but won’t translate in-store

2017 was the year of AI. Even now, all you have to do is Google search ‘AI 2017’ to find headlines like these:

• ‘The biggest artificial intelligence developments of 2017’
• ‘2017 laid the foundation for faster, smarter AI in 2018’
• ‘All the creepy, crazy and amazing things that happened in AI in 2017’

AI took the tech industry by storm. Swarm AI correctly predicted TIME’s Person of the Year to be Donald Trump, AI moved into the household through the Amazon Echo and Google Home, and Google’s DeepMind AlphaGo Zero conquered the 2,000-year-old board game ‘Go’ through machine learning.

If you didn’t already know: AlphaGo literally recreated itself without the help of humans, using reinforcement learning to surpass the abilities of world champion Le Sedol and become the best Go player in the world.

In 2018, poker bot Libratus was the first to beat 15 top human players, and American technology company Nvidia created AI that could mimic your facial features, handwriting, and voice. They created ‘celebrities’ that don’t even exist.

Though it didn’t impress everyone, with comments like:

• “Aren’t they just mixing faces of celebrities? That doesn’t sound like “generating to me…”
• “Aren’t most celebrities on social media “fake” anyways?”

… It still showed us the tip of the iceberg. The iceberg that would later reveal the all-conquering and all-powerful force reckoned to control our entire lives – otherwise known as artificial intelligence.

AI might not take your job, but it will affect it

In July last year, Amazon pledged to invest over $700 million in upskilling one in three of its employees in the U.S. While this didn’t exactly confirm their jobs were on the firing line, it questions how much of our roles will be automated with AI. After all, with a revenue of approximately $334.7 billion, they probably aren’t struggling for new developments.

Figures like this weren’t, and still aren’t, helped by reports on advancements in technology. One report by Oxford Economics suggested that the global stock of robots will multiply even faster in the next 20 years, reaching as many as 20 million by 2030.

Even if reports like these were somewhat inaccurate, it’s no secret that AI and automation have the power and logistics to change the way we work. We might not see physical robots moving around the office, but virtual robots powered by machine learning guide our lives in more ways than we can imagine.

We wouldn’t know how much traffic there was on our commute home without it. When we order Deliveroo, we wouldn’t know the expected delivery time. AI isn’t just a part of technology companies and devices, but the way we function, eat, and consume information.

For businesses, advancements in AI have transformed the way customers make purchases online, and flipped the traditional buying cycle on its head. Where customers used to find out about a product, consider buying it, and pretty much end up typing in their card details, some brands have started giving them a taste of the product beforehand to encourage ease of service and more conversions.

We see this with brands like IKEA and Sephora, whose use of AI has provided such high-level customer experience that arguably cannot be replicated in real life.

Toshiba’s Fredrick Carlegren and Dunn-Edwards’ Rich Stefani touched on this concept in a recent talk on retail challenges with AI, both agreeing that stores need to be able to create experiences that will distinguish them from their competitors and enable them to meet customers’ expectations.

In particular, Stefani said retailers today have access to the data they need to accomplish that goal. “Stores today, for example, can know what the weather is like at any given location before the customer enters the store.” He continued, “Frictionless technology is exciting.”

It is exciting. The thought of gaining more customers with high intent, being able to successfully predict their next steps and potentially preempt their actual footsteps into your store. But data of this kind can’t be mirrored by humans alone.

For a brief example, take Nike.

Just last year, after estimating that three out of every five customers were buying the wrong-sized shoes, the multinational sports brand began using augmented reality to help customers find their perfect size. Now customers can use the app, otherwise known as Nike Fit, to find out exactly what size they should be buying.

For it to work, customers are asked to stand next to a wall and point their smartphone camera at their feet, which will prompt two AR circles to appear on the screen. Once it recognises your feet, it will simply scan them and tell you your ideal shoe size for Nike footwear. The process apparently takes less than a minute.

The problem with this is that shoes fit differently. And this kind of feels like a rip off of the classic foot measuring board used in Clarks and Schuh stores today. More expensive, rather, but pretty much the same thing. And neither of them are tailored to the shoes you actually want to buy – whether they’re Nike or not.

The same AR is also used in Nike stores across the U.S., this time assisted by the sales assistants.

Automation of this kind, like you see in marketing, will be excellent for parents who are panic-buying for their children or people who are used to wearing a harder-wearing shoe than Nike trainers. But if we’re being realistic, it’s rare you walk into a store and get served straight away. The sales assistants are humans too and unlike technology, we can’t use AI to increase our mobility: location, time of day, and number of customers in-store all make a difference.

Plus, technology is temperamental, and these handy devices don’t always work. From the off-set maybe, but the buying process becomes even longer for the customer when they’re waiting for technical assistance.

Despite his earlier convictions, Microsoft co-founder Bill Gates argued that a few years after the machines have automated some of our jobs, the intelligence will be strong enough to be a concern.

He didn’t get a $110 billion fortune under his belt making rash decisions. He knows what works and that’s what’s pushed him to where he is today. Does this prove the future of AI? Probably not. But it’s a damn good starting point.

The hare and the tortoise: move fast or finish last

For larger businesses such as Amazon, advancements in technology and machine learning have provided more opportunity to take over the marketplace. To emphasise the earlier point, just Amazon alone spent $22.6 billion on research and development in 2017, winning them the title of 2018’s largest investor.

Note: this research didn’t affect its pre-existing algorithm, otherwise known as the Recommendation Engine. If anything it enhanced it, keeping Amazon at the forefront of online shopping with clever recommendations for people all over the world.

The algorithm uses data from your purchase history and browsing history to create a list of products you’re likely to buy next, and shows these items on the homepage, your recommendations page, underneath a product you’re currently browsing or sometimes in an email.

However, it’s been argued that Amazon’s algorithm isn’t entirely accurate, with the “Amazon Choice” label receiving some pretty negative press. According to an article written by Buzzfeed last year, Amazon refused to answer questions about how their Choice listings are selected and are sometimes flawed.

It’s up for debate, but emphasises the fact that this sort of AI would never work in real life. Anyone who works in sales or copywriting will tell you it’s easier to sell over the internet if the product you’re selling has some sort of credibility. In this case, the Amazon’s Choice label. But we might not trust someone stood in front of us with an Amazon name badge, showing us a product with a “Best Recommended” label on it.

People trust Amazon, and so they can afford to gamble whatever they want in terms of algorithms. It’s up to us to decide whether we fall into the trap in future.

Scholars Davenport, Guha, Grewal, and Bressgott shine light on the future of AI in their theoretical paper, stating:

“With AI, online retailers may be able to predict what customers will want; assuming that these predictions achieve high accuracy, retailers might transition to a shipping-then-shopping business model. That is, retailers will use AI to identify customers’ preferences and ship items to customers without a formal order, with customers having the option to return what they do not need (Agrawal et al. 2018; Gans et al. 2017)”.

The line between these predictions and what businesses can actually achieve seems to be becoming increasingly blurred. Of course, global companies like Amazon have the resources to emphasise AI online, even if it would potentially fail in real life. But not all businesses can jump on the bandwagon and expect exceptional results.

Even without AI, some of the UK’s largest retailers have failed to provide the same service in-store as they do online. When footwear retailer Schuh launched a new store concept in 2016, the frictionless experience they once imagined didn’t translate. Replacing the main cash desk with integrated tills interrupted the customer journey of many, leaving questions unanswered, queries unresolved, and creating confusion in-store.

This likely cost them a lot of customers, but not as much as failed AI would have.

For technology entrepreneur and investor Masayoshi Son, artificial intelligence will serve us well if we use it in the right manner. He stated: “If we misuse it, it will be at risk, if we use it right, it can be our partner”.

IKEA’s Virtual Reality Showroom

Whilst Amazon is the global powerhouse allowing small businesses to sell products on their site, IKEA is the leader of all things furniture, accessories, and appliances. But you know that already.

All you have to do is click on the IKEA app to find some wardrobes for your new bedroom. It lets you drop virtual furniture into your home and view it through your smartphone camera, taking full advantage of Apple’s augmented reality framework. There’s also the option for these products to be delivered and built for you, with added peace of mind that each unit will slot into perfection.

IKEA makes your life easy.

The organisation has utilised AI in such a way that has removed the manual work of turning a house into a home. It’s how they’ve held the title of the world’s largest furniture retailer since 2008, which all began with 17-year-old Ingvar Kamprad self-assembling products in the 1940’s.

In fact, IKEA is so far ahead of other businesses that this AI was researched, designed, and brought to life in their own future-living lab Space10. The lab predicts and cultivates how we’ll be living in the next 10-15 years through innovative projects driven by artificial intelligence.

The augmented reality app, otherwise known as IKEA Place, was created here and launched on iOS in September 2017: another development to add to our imaginary yet powerful timeline. Supposedly shaping an automated way of living for us all, on the day of its release, Apple CEO Tim Cook said IKEA Place was the future of shopping.

But it isn’t all sunshine and roses.

It would be, if every single person found the AR to be successful and didn’t have to speak to an assistant in-store, re-jig their room, order new furniture and start again. But it isn’t all as perfect as it seems and people aren’t afraid to talk about it. For example, this extensive case study identifies 10 user pain points of using the app alone.

Even with a redesign, and some finding it helpful and easy to use, it’s nearly impossible for IKEA employees to mirror the speed, functionality, and convenience of IKEA Place.

To begin, the time slots given for IKEA deliveries often don’t take into consideration people who work. Items aren’t always the same colour as they appear on the app, particularly with black-brown furniture. And as YouTuber and influencer Elle Darby pointed out in her video, having someone from IKEA set up your furniture isn’t always a good thing.

She said: “The poor man made a mistake when he was drilling the holes in the door for the handles to go in, and he was so sorry about it and I could tell it was just a 2-minute mistake…”.

But when the whole point of utilising AI is to improve efficiency, it’s fair to say some customers will hold no prisoners.

Sephora’s Visual Artist

As of 2019, the beauty industry was a $532 billion market, built off millions of people spending their money on products recommended to them by supposedly reputable sources. It’s not quite the same as Amazon, but the impact content creators have on social media is often enough to make anyone try something new.

Beauty retailer Sephora observed this scene, and saw the detrimental link between ordering makeup products online and ending up with the wrong shade. The answer? AI. 

But not just any AI. An innovative and accessible AR tool that would allow customers to try on makeup virtually and use the AI-powered Color Match tool to find the perfect shade for their skin.

They could even sample a fragrance via touchscreen and scented air, and follow beauty tutorials digitally on their own face to learn how to achieve certain looks. They used AI to cut out the need for YouTube tutorials and reduce the chance of ever ordering an unsuitable product. That’s power.

Sephora’s Executive VP of Omni Retail, Mary Beth Laughton, said Sephora’s Visual Artist is a really good example of where there was a real customer need. She continued: “It can be overwhelming coming into our stores or shopping online, but this makes it easy to help you find your favourite shade and save you time”.

But that’s the exact problem here.

With Sephora’s Visual Artist, the only way to find a foundation shade that matches your exact skin tone is by uploading an image of your face into the app. But everyone knows the majority of pictures never look the same as they do in real life. Even in 2018, iPhone X users complained the new camera was automatically smoothing their skin.

More to the point, a picture you upload into an app is never going to be as accurate as testing different products in real life. And if you have specific requirements, like wanting your foundation to match the colour of your tanned skin rather than your natural face, you can’t exactly tell the app that. The communication is no longer there.

Plus, Sephora makeup artists won’t be able to keep up with the fast-paced service offered online. If people are finding products quicker through AI, the queues to pay will be out the door and they’ll need quick explanations if customers are recommended the wrong shades.

Even if AI supports customers in ordering their accurate shade, as data suggests it has, there will still be problems found with the app. There are endless complaints about the app not working from just last year. 

AI online might be a success. But for the people who expect such a flawless experience in-store, it will be a completely different story.

AI: A concept for the winners

The future of AI is talked about all over the internet. But the truth is, nobody knows for certain how much it will impact our lives. Cognilytica’s four-part AI-enabled vision seems pretty close, but that’s just my opinion.

What is clear is the power and speed of automation. Amazon’s algorithm generates 35% of their entire revenue, and the algorithm used by Netflix was estimated to be $1 billion per year in 2016. Even just two years later, Netflix crossed the $100 billion net worth mark, growing exponentially and surpassing everyone’s expectations. All through artificial intelligence.

But that is them, and this is us. There’s thought to be around 300 shops on London highstreet, and you can bet none of them are worth more than Amazon and Netflix. AI or not, they’ve got a lot of catching up to do.

Regardless, no amount of money put into human resources can compare with artificial intelligence. 

Long before Schuh improved their stores, employees would run up and down the stairs from stockroom to stockroom carrying as many shoe boxes as they could. But not all customers would be sat waiting for their shoes when they brought them up. The impatient left, anticipating the day of faster processes.

Now, customers will find something else to complain about, and the service just won’t be perfect enough. Particularly if they’ve experienced a fast-loading website and VR tools that basically prize the product into their hands.

As humans, we cannot replicate the same customer service we’re advertising online. Not until we are robots, but if that happened it’d make you question the whole of this article, huh?

Guest article written by: Post by Jess Kirkbride, Copywriter at Adzooma – the simple, quick & easy way to manage your online advertising.