better decisions, faster
AltViz Peaks

An Executive's Primer on Artificial Intelligence

James Billot

Download this article as a nicely formatted PDF for printing or reading later.

When you mention the term AI during a meeting, a few things are almost certain to happen. The enthusiastic techies in your group get super excited about potential business applicatications. The cautious segment audibly murmurs about the inevitability of machines taking over human jobs. There may even be a terrified 80's movie fan prophesying the coming of Skynet, Ultron or HAL9000.

The majority, however, aren't sure exactly what you mean by AI, because there's no such thing as a universally accepted definition. Artificial intelligence is both a field of research and an umbrella term that covers processes like machine learning and natural language processing.

That's why in this briefing we are going to take a look at what AI really means in today's market, examine how its progress over the last decade affects various industries, and see it in action with a few case studies.

An Executive's Primer on Artificial Intelligence

A brief introduction to artificial intelligence

Artificial intelligence is the idea that a machine can learn over time and eventually develop the same decision making abilities as its human counterpart. The concept was first popularised by the mathematician and computer scientist Alan Turing who famously said that “a computer would deserve to be called intelligent if it could deceive a human into believing that it was human.”

The 50s was the first big era of AI enthusiasm. The problem was that in order to learn, computers needed data and there just wasn't enough of it. As tech struggled to come up with effective real world business applications, funding dried up. It wasn't until we hit the mid 2000s that AI finally became viable.

Advances in graphics processor units (GPUs), the creation of cloud storage, the spread of personal computers and the ubiquity of the internet meant that we finally had the tools and the data to start exploring AI's real potential. The data gave technology companies the fuel they needed to make huge leaps in machine learning, deep learning and natural language processing. After a couple of false starts, we have entered the era of artificial intelligence applications.

AI technology and the modern industrial revolution

Today's artificial intelligence applications can do a lot. AlphaGo, a computer running Google's DeepMind AI managed to beat the world's number one Go player in early 2017, a task previously thought impossible because of the game's sheer complexity.

AlphaGo isn't alone. Waymo's autonomous driving AI has been successfully navigating the streets of Phoenix, Arizona for the last few months and the company is getting ready to release a limited ridesharing service in the area sometime this year. IBM's Watson has become incredibly important to the medical community, as it can use global data to recognize tumors with speed and efficiency, allowing patients to receive better and faster care.

These machines are doing jobs that were previously thought to only be possible with humans. They use machine learning techniques to not only understand and interpret data, but to act on it. In many cases, they can do so much faster than we can, because they have access to extensive data that would take even the best team of scientists a very long time to interpret.

However, AlphaGo and Watson hardly resemble the terrifying AI overlords of science fiction.

That's because today's artificial intelligence is task-focused, and uses machine and deep learning to complete a specific task within set parameters, rather than expending processing power to enslave the human race. The scary world-ending tech we often think of is known as artificial general intelligence and, at the moment, we aren't anywhere close to developing it.

That means while Waymo's self driving AI has spent millions of hours learning how to drive and has arguably perfected the skill, it can't suddenly offer you break through medical advice, because that's not what it's been trained to do. It simply doesn't have the knowledge to make that kind of cognitive leap. Just like a data scientist and a neurosurgeon can't randomly swap jobs, today's AIs are hyper specialized.

Why we use artificial intelligence

AI applications are proving to be incredibly useful to businesses in multiple industries, but investment in this technology is currently dominated, rather unsurprisingly, by tech giants with deep pockets. The aptly named FANG companies Facebook, Amazon Netflix and Google are tearing into the AI space alongside their Chinese competitors Baidu, Alibaba and JD.

These behemoths have been collecting high-quality user data for years and have the money, know how, and corporate structure to use it effectively.

The current AI wave is poised to finally break through - McKinsey Infographic showing investement is high, but, adoption remains low

Sourced from McKinsey

In their global report on AI, McKinsey identified five key sectors of AI growth. These are robotics and autonomous vehicles, computer vision, language, virtual agents and machine learning. These sectors have the power to affect and disrupt every major industry and it's time for companies that want to stay competitive to consider the benefits AI can bring.

Intelligent automation and smarter decision making

AI can automate repetitive business processes and decrease the effect of cognitive bias by accessing insights hidden in the mountains of digital data most companies collect.

When people make decisions, we go through a process that uses our past experiences, learned skills, and current information to draw a conclusion. However, our memories, and therefore interpretations of data, are inherently subjective, and differs from person to person.

If you handed the exact same dataset to 10 different people and got their honest interpretation, you would end up with 10 slightly different accounts of what the data means based on individual background, experience, knowledge, leading hypothesis, and inherent bias.

However, if you gave the same dataset to ten computers that had software optimized for this type of machine learning, you would get the optimal answer every time. This is because machines have no interest in defending a worldview they do not have, nor do they make mistakes or attach personal bias. Your data center isn't trying to get promoted, impress you with its knowledge, or convince you it knows what it's talking about. It sees numbers and it interprets them in an optimal way, because that's what it's programmed to do.

When new companies emerge, they develop particular ways of carrying out tasks based on what works well in their sector. They determine what's most successful and eventually fall into a rhythm. Initially, this is important to group cohesion and moral, particularly in the early stages of growing an organization. However problems arise when best practices turn into a status quo and the workplace rhythm stagnates. Disrupting that status quo is hard because it means questioning the wisdom of your team- few individuals want to do that out of fear they might get it wrong.

This reluctance to take risks can stall innovation and lead to further cognitive stagnation. This in turn decreases profitability and opens up the door for a stealthy competitor or a newer, more disruptive company to swoop in and claim a large slice of your market share because they took a risk that you didn't.

Using AI applications to make data-based decisions can protect you from stagnation by effectively eliminating one key factor that can hinder smart decisions: office politics.

Better customer service

Artificial Intelligence

Whether you are in a B2B or a B2C environment, customer data is invaluable. Intelligent applications allow businesses to use data to better understand how their customers interact with their products, and influence future purchases.

Alexa is a perfect example of how to execute on collected data. Amazon's virtual assistant AI can now recognize the voices of up to ten different people and provide a personalized experience. From something as simple as remembering your media preferences and always playing your favorite song to making brand suggestions based on your shopping habits, it creates an experience that's unique to you.

Amazon is masterfully using Alexa to create a true Virtual Assistant that makes your life easier while making ordering from Amazon the simplest possible option. Businesses that aspire to true solution-based sales can learn a lot from this model.

Using AI tech to create a virtual assistant that helps customers navigate your product can elevate the user experience and cement brand loyalty. But not all AI has to be as pervasive as Alexa to enhance the customer experience. Nor is it always prudent to demonstrate the full potential of AI, particularly when it comes to an individual's finances.

Capital One, a major online bank based in the US, used natural language processing and AI to create Eno. Eno is a chatbot with a unique personality. Customers can text it and ask it to perform tasks like send money, pay bills, check balance, or explain where to find their routing number. This kind of text recognition and response has been around for awhile, but what makes Eno unique is its understanding of emojis, abbreviations and misspelled words.

Cap One released the bot in a closed pilot program that allowed the AI to get to know how customers talk in real life and learn from every interaction. This access to a large data set of vernacular speech allowed it to absorb how people talk to one another and mimic it, creating a seamless experience that feels human.

As tech continues to advance and further integrate into our day to day, the human touch is becoming essential. That's why in the US, there's been a huge move toward domestic-based customer service and personalization. It makes consumers feel valued and increases the chance they'll stick with you instead of packing their virtual bags and heading to the competition.

AI can help you use data insights to create a bespoke experience for each individual customer, allowing them to capture the feeling of a high-end luxury store or concierge service without the price tag. This next level of customer service increases retention rates and average spend without additional staffing costs to affect the bottom line.

AI can make us better

AI applications can help us make better, smarter business decisions across the board by understanding and interpreting the wealth of data we collect. The applications can offer personal recommendations based on hyper individual past experience.

The problems AI won't solve

The problems AI won't solve

The AI tech we've got right now isn't creative, and it certainly doesn't dream of electric sheep. As a result, it won't be able to revolutionize business practices or come up with brand new models by itself. That level of innovation will remain up to your organization's leadership. In fact, further advances in AI will make creative human problem solving more important than ever.

AI can deduce, regurgitate and make data-based decisions but it can't create a solution or reimagine the problem. To put that in context, AI can analyze a model of customer data that will predict when customers will become upset or discontinue using a service. But it can't use empathy to understand why customers act in a specific way, or decipher what you can do to change that. We still need a highly skilled human workforce to make cognitive leaps based on unbiased analysis of properly calibrated AI.

When you couple quick predictions based on big data with creative solutions-based thinking, you can get some pretty impressive results. That's exactly what McKinsey predicts will happen to C-level execs. Leadership skills, strategic thinking, and soft skills like empathy and human-focused problem solving will become invaluable.

Futurists Dan Wellers and Kai Gorlich agree and suggest that we need to strongly focus on developing these soft skills. In other words, we've got to work on the kind of skills that machines don't have. The more we develop various technologies that make lives and businesses easier around the world, the more important personal touch will become.

AI won't solve your data inefficiencies

Many of the companies we've mentioned are implementing AI into their business with substantially different approaches. The one constant is that effective machine learning requires clean, reliable and diverse data sets that can be easily accessed by multiple machines from multiple locations. The quality of the insights AI can give you is directly correlated to the “training” of the machine being used.

Waymo put their autonomous driving AI through millions of hours of real-life and virtual driving simulations. They've tested every conceivable situation multiple times in order to cement the learning. Capital One's Eno also spent months in a closed test environment learning different expressions and talking to thousands of customers in order to build up its vocabulary and expertise.

This kind of exhaustive testing takes a tremendous amount of discipline and work hours. While AI is at an early stage in its development, any real world implementation into a new or existing business will require extensive human oversight from a qualified team. For the project to succeed, it may also need continual management even after the process is capable of self correcting.

The quality of a company's data isn't the only hurdle to properly utilizing this technology. Many organizations that have been collecting customer information for decades still haven't fully digitized their databases, and will need to painstakingly review and input information into a usable model if they want to benefit from AI decision making. Without the right infrastructure, AI will struggle to work in your organization.

AI is not going to magically solve existing structural problems in your company in the same way adding the word blockchain to the name of your business won't make it a Fortune 500 company overnight. Without proper implementation, any existing weaknesses in processes or execution can be exacerbated by improper implementation of new technology. That's why support at every level of implementation is absolutely crucial for the success of an AI project.

Executives need to identify the areas that would benefit from AI systems, and then prepare each team so to use it effectively. It's up to managers and leaders to show their team members that AI is at its best when working alongside smart, dedicated employees.

How we can use AI

Artificial intelligence has potential applications in almost every industry, from the glamour of autonomous vehicles and the intrinsic practicality of fraud prevention, to efficiently processing insurance claims and improving the customer experience. Let's take a look at some of these applications in practice.

Democratizing healthcare

As an industry, health care is ripe for disruption and AI can help increase accessibility to treatment and decrease costs.

Hospitals keep data records on millions of illnesses. AI can potentially scour that data to identify large scale patterns in illness development and help early recognition and prevention efforts. IBM's Watson has already done this with tumor identification and it would be fascinating to see further implementation. But this specific application depends on different companies working together, so it is harder to realize on a large scale.

However, AI can have an immediate effect by supporting virtual healthcare. Adam Licurse, the medical director for telehealth at the Brigham and Women's Hospital worked with a number of partners to create a simple questionnaire that covered common primary care complaints. He tested the system by running a pilot with 700 e-visits that allowed patients to get quick treatment without a lengthy, hard-to-schedule doctor's visit.

AI can help create a better experience by identifying patterns and using natural language processing to communicate with patients.

Improving insurance

Processing insurance claims is neither easy nor quick. Just one look at the US healthcare market (even if you don't have personal experience with it) is enough to show that.

Insurance policies of all kinds are getting more complex and, as a result, more difficult to process. This means that insurance teams are spending more time going through and assessing claims, which leads to slow turn-around and increased customer frustration.

Intelligent insurance applications track customer behavior and use historic data to go through the forms, identify patterns, spot repeat claims, detect fraud, and help process payments in a fraction of the time. This frees up your team for more person-on-person interaction with your customers.

AI and e-commerce

As e-commerce becomes an integral part of how we do business, and Amazon continues to slowly stamp out (or acquire) the competition, AI will play a pivotal role in the industry's development. It's already got broad applications in aspects like inventory allocation and listing optimization.

One of the biggest use cases for AI in e-commerce is using past customer purchase data to suggest related items and create personalized shopping suggestions. You can add these to your site or use them to create a series of personalized marketing emails.

Why your team should use AI

Why your team should use AI

McKinsey surveyed 3000 C-level executives that have an above average awareness of what AI can do. Of the 3000, only 20% are using AI as a core component of their business because the value of using AI isn't always clear. So, should your business embrace it?

We are on the cusp of monumental change. The way we use technology is beginning to shift from passive interactions that satisfy an immediate need or automate a simple process, to entirely new ways of interacting with each other through machines that can make decisions in the same way that we do.

Where we are now is similar to another pivotal moment in history: the industrial revolution. Henry Ford's car assembly line made a lasting impact on manufacture. It saved time and made mass production possible on a previously unimaginable scale. However, once Ford's competitors saw what the process could do, it wasn't too difficult to replicate. This was true for many radical changes that happened during that period.

AI is different because unlike an assembly line, it's always learning. It analyzes data, makes errors and learns based on past interactions. It's constantly improving. Think of Google's search algorithm. When it first started out, it took it a long time to be able to identify backlink scammers and keyword spammers. However, each new advancement has happened faster than the one before because the AI is getting constant access to new information that drives rapid development.

Catching up to Google's algorithm now will be practically impossible if it keeps developing at its current pace. Every business that embraces AI successfully is likely to follow this pattern, due to the nature of AI. Those that get in on the ground floor, start developing methods for handling big data, and create clear goals for its use will be at an advantage.

Adopting AI across an entire business isn't easy. Results depend on the availability and maturity of high quality datasets, but the benefits of the tech can have lasting effects. When applied to specific business problems, it can help you increase your team's productivity and allow employees to spend more time on the problems only they can solve.

Executives need to get clear about how AI applications can help improve processes, increase productivity and serve customers better. We need to identify key areas for improvement and then look at existing solutions as well as options that can be customized or built on site.

Download this article as a nicely formatted PDF for printing or reading later.

We use cookies on our website for an improved experience & site analytics. 🍪
By continuing to use this website, you consent to the use of cookies in accordance with our cookie policy.