Programmatic Lunch

The business of AI

David Murphy

Is Artificial intelligence a force for good, or for bad and how should brands harness its power to improve the customer experience? Tyrone Stewart investigates.

Artificial intelligence. AI. We’re all well aware of it – whether it was first brought to our attentions through movies like A.I. Artificial Intelligence, I, Robot, Blade Runner, and Ex Machina, or when IBM’s Deep Blue supercomputer beat chess grandmaster and former world champion Garry Kasparov way back in 1997.

To date, AI hasn’t quite yet reached the levels seen in the above movies – though it’s likely to head there at some point in the not–so–distant future – but has come on leaps and bounds from beating the greatest chess player of all–time over two decades ago.

AI appears in much of the technology we use today, even if we take it for granted. It appears in obvious places like in our Amazon Echos and Google Homes, yes. But the technology is also powering the suggestions provided to you by Netflix, Spotify, and Amazon; the computer–operated ‘bad guys’ in video games like Call of Duty and Far Cry; in ride–hailing apps like Uber and Lyft to determine the price and match you with other riders; and it’s even used by social networks to help them moderate their platforms, alongside plenty of other uses of AI most of us come across every day.

“AI is on the rise in both business and the world more generally. AI will, in many ways, amplify and enhance human ability – particularly in business. Machines have been developed to take on highly sophisticated tasks and process vast amounts of information that humans can’t do at scale or speed. Automation can analyse vast amounts of information to produce meaningful insights quickly and accurately,” says Ben Plomion, global CMO at GumGum.

“AI can be used to see and solve difficult problems for humans, and free up time for us to take on more intellectually challenging tasks. In this sense, the appeal of AI for businesses is significant. As the technology becomes more sophisticated, organisations are increasingly leaning on AI for cutting–edge applications that satisfy the need for reducing costs, boosting efficiency and driving innovation.”

What’s the use?
This year, we have seen more controversial suggestions of how AI could be used by businesses to make either our or their own lives easier.


We’ve seen Google unveil its Duplex technology, which will one day enable people to get Assistant to make appointments and reservations for them over the phone at salons and restaurants. This wasn’t entirely well–received upon its unveiling because, in the demo, the AI did not inform the person on the other end of the line that it was indeed an AI – and people questioned the ethics of this. The internet giant did respond to the complaints and assures us that the final version of the technology will come with a disclaimer.

Uber also courted AI controversy with an interesting recent patent application. The transportation network filed for a patent for a computer model that can predict user states based on data from current in–app requests, as well as their previous behaviours – essentially being able to detect when people are intoxicated and requesting a ride. This information could then be used to alert drivers before they accept a passenger and may not give users the option to opt for shared rides. It’s certainly an interesting premise, especially for those (like me) who use the app while intoxicated 95 per cent of the time. (I still have a 4.6–star rating though.)

AI in the back
Though there are plenty of left–field uses of AI, like the above, many companies are utilising the technology in the back end to help power their platforms, rather than applying it for the consumer–facing elements of their business.

One such firm effectively using AI is Waze because: “Any company with decent data sources is, or should be, making the most of it,” according to Finlay Clark, UK country manager at the Google–owned firm.

“The combination of data with artificial intelligence – specifically machine learning – is fundamental to how Waze works,” says Clark. “Today we’re not only using the technology to help determine speed limits on a road, but also to help drivers navigate through dangerous zones, and to learn where our users plan to drive next and optimize their journeys accordingly.”

Waze is joined by companies like Trainline in ensuring all aspects of transportation benefit from what AI technology has to offer.
The digital rail platform uses “data to create smart solutions” for customers, with AI being “fundamental for any travel tech company looking to stay ahead of the curve and meet growing demand for a personalised experience,” says Fergus Weldon, director of data science at Trainline.

“At Trainline, AI trained with huge amounts of data provides bespoke contextual information for every customer, from helping them get the best ticket price through predictive analytics to helping them find a seat on a busy train through crowdsourced feedback,” continues Weldon. “Advances in AI mean that we can address some of the challenges our customers face when travelling and give them effective real-time smarts to help make their journeys more seamless.”

The robots are coming
Away from the positive commercial applications and the other more opinion-dividing ones, there is a long-running concern around the idea that AI technologies will eventually start working together in order to wipe out the human race (a notion that has been pushed in Hollywood by the movies mentioned earlier).

For now, there are two primary concerns around the technology. The first is around its potential to create mass unemployment, as computers become more capable of doing more of the things that humans currently do. The second is potential for bad actors to exploit AI by hacking it or by using it to create fake content, including video content, that could be used to manipulate the public.

Looking at how AI can affect jobs, many have speculated that the technology will result in the loss of thousands upon thousands, others have said the impact won’t be anywhere near as bad as a lot expect it to be because new jobs will have to be created to manage the robots. One thing’s for sure, we’ll get a better picture of which side is right within the next few years – but those within the industry seem to be of agreement that the impact on jobs probably won’t be anywhere close to as significant as the naysayers predict.

“As with any new technology applied to the advertising world, there are questions over how AI will impact creative roles and skills in the long term,” says Peter Day, chief technology officer at Quantcast. “Some are worried that it will kill off jobs. Others argue that technology cannot mimic true, authentic human creativity.

“The question shouldn’t be whether AI can do things better than humans, but how AI can help humans do better. The brands and marketers that first figure out how to complement their employees’ creativity with the power of AI will have a disproportionate advantage over their competitors.”

Looking more broadly, Stuart Templeton, head of UK at Slack, doesn’t “believe in a future where technology takes over all jobs”.
“Over time, with innovation, yes, some tasks may be replaced by technology, freeing us up as a society to do things that we were unable to do before,” he says. “For now, though, to free us up to be more productive, companies should create an environment where collaboration between employees is as frictionless as possible, and embrace innovative technologies to support this.”

When it comes to bad actors, there have been growing concerns surrounding the possibility that AI technology could be exploited.
Earlier this year, a report from seven organisations, including Cambridge University’s Centre for the Study of Existential Risk and the Future of Humanity Institute, warned that AI is in a position to be taken advantage of by rogue states, criminals, and terrorists. This notion was reiterated by the UK’s House of Lords’ select committee on AI, which highlighted the need for a cross-sector AI code to ensure AI is used correctly.

On top of all this, there have been several examples of researchers managing to hack smart assistants to spy on users or even order products via AI personal assistant devices without the owner being aware.

“There’s no question that malicious use of AI is a risk that needs talking about and addressing,” agrees GumGum’s Plomion. “Steps are already being taken to understand exactly what those risks are, what they could be in the future, and how we can prevent them or act on them when they do occur. With the technology still being in relatively early stages, it’s hugely positive to see these issues being raised and explored now so that we can continue progressing safely. AI has enormous potential to impact our lives for the better, and we need to learn how to work together with this technology in order to reap its rewards in future.”

Ethical issues
Given AI’s potential to be used to do harm, it is important to address the ethics of the technology. The implication that AI technology could be hacked by bad actors creates a whole new issue around the privacy of people’s data. And then there’s the issue of the potential biases that could either be ingrained in the technology or that it could learn.

“The main ethical issue surrounding the use of AI in my opinion is bias,” says GumGum’s Plomion. “It’s essential that AI be designed by a diverse group of individuals and designed to reflect the minutiae of society within which it hopes to fit.

“These issues will become increasingly important as new technologies powered by AI become more prevalent in society. Developers need to understand that they are no longer just building apps but technology that intertwines with wider social, political and economic systems.
“Likewise, issues of data privacy, product utility, creative execution and computing resources all come into play when building AI services and these require careful ‘human’ consideration in order to roll out smoothly and properly.”

The ethics behind AI were also a big part of the House of Lord’s select committee on AI’s recommendations. The group suggested that AI should be built for the benefit of humanity, with intelligibility and fairness, to ensure people’s rights to their data, to educate, and to not be used in order to harm or deceive humans in any way. It also highlighted the need for AI training in order to ensure all ethical standards are met.

Is the future br–AI–ght?
“AI is set to disrupt every customer interaction, every company, and every industry, including advertising,” says Quantcast’s Day. “Today, our industry is just scratching the surface when it comes to the potential for AI, and more specifically machine learning, to deliver more relevant and effective campaigns.”

Yes, that’s right, the AI technology we’re seeing now is still very much an infant in its development and “will have a bigger positive impact on our lives than the internet,” adds Day.

But with such an impact expected and all the negatives that are still swirling around the technology – whether its job losses, bad actors, or unethical use – has the technology already come too far, at least for the time being?

Looking back to the Google example or at a robot such as Sophia – if she’s real – and some will wonder if the scenes from the movies are a lot closer to becoming a reality than we ever expected. Especially when you hear people who utilise AI, like Tesla’s Elon Musk, warn of the technology having the potential to lead to a situation where you have a robot dictator.

“Voice technology and AI assistants could help to bridge that gap and deliver more intelligent and engaging customer experiences,” says GumGum’s Plomion. “When we think about AI going too far, it’s usually the concept of the uncanny valley coming into play – people tend to feel unsettled and threatened when androids resemble humans too closely. However, assistants of this kind have been designed with the intention of making life easier for humans – in theory, they will eventually be able to solve problems for us efficiently and effectively in a way that will feel natural and innocuous.”

Whatever fears there may be around AI, they have not stopped a huge number of people buying smart assistants like an Amazon Echo or a Google Home. Because of this, it’s important that businesses embrace the technology and use it to engage with their customers.

“The more we learn about the power of AI, the more we can understand its applications in the real world. Though there is always a bit of risk, I don’t think the risk of bad actors is greater than previous technologies,” says JJ Lopez Murphy, AI and big data tech director at Globant and one of the authors of Embracing the Power of AI.

“When AI was first introduced in commercial applications, we didn’t understand it as well as we do today. When previous technologies like the internet were introduced, many of its capabilities were unknown and feared. But the more we learned and understood the internet’s potential, the more we could see its benefits instead of its risks.

“As researchers and data scientists continue to advance and understand AI, the more organizations will see how using AI will benefit workforce. The bottom line is that AI is just another example of cutting-edge technology that will improve the lives of many people. And though there are risks, the rewards outweigh them,” Murphy concludes.

At the end of the day, we all know AI technology is here to stay and is only going to start popping up in even more places all around us. Whether businesses opt to utilise AI more on the front end than they are currently doing on the back end in the future remains to be seen. Whether we are set to be overrun by tyrannical robot overlords is another moot point.

For now, we can all just continue talking to our Alexas, Home and Siris, using our AI-powered apps and streaming services; and playing our AI-powered games while always hoping they don’t one day turn around and decide to control us instead.

This article first appeared in the September 2018 print edition of Mobile Marketing. You can read the whole issue here