Does your business really need AI?

Before the invention of the internet, we didn’t know we needed it. Now we can’t live without it. The future impact and benefits of new, disruptive technologies like the internet, or for that matter AI, can be difficult to comprehend. This conundrum presents challenges for a business when they are looking to justify investment in technologies such as AI. So how do businesses address this dilemma?

I was recently meeting with a new customer, who represented a large multi-national organisation. It was clear to me that, despite their obvious interest in how they might use Natural Language Processing and other AI technologies, they were struggling with the question of how it will really benefit their business. I see this theme repeated frequently. In fact, with almost every organisation I meet.

I am not suggesting that these people are overly conservative or naïve. Far from it; by my assessment, they are forward thinking innovators who are looking at ways to disrupt the market.

Regardless, deep down, they are asking themselves healthy questions, such as “Do we really need to be looking at AI?”

As individuals, the people that I meet in these organisations are all acutely aware of how Artificially Intelligent systems are becoming omni-present in both their personal and business lives – Siri, predictive navigation systems that avoid traffic jams in real-time on the way home and facial recognition on Facebook, to name a few.

They are even more aware of the pressure placed on them as executives to keep up with their competition and current technology trends. In a recent publication by Boston Consulting Group,  almost 85% of respondents believed that AI would allow their companies to obtain or sustain a competitive advantage. But only half of those said that their organizations have a strategy in place.

This got me thinking. Why the hesitation, reluctance and, dare I suggest it, even fear?

At its core, I think is the notion that a good idea doesn’t make a great innovation.

Just because you have a great new idea, or shiny new technology, doesn’t mean that people will want to use it. Nor does it guarantee that it will really deliver material benefits to the anticipated users of that technology. Further, as an innovative and disruptive idea is being conceived, it is often difficult to imagine how that idea will deliver business value or even become indispensable. Will it ultimately change the world, or be left forgotten on the side of the road; a cautionary tale to those future business leaders who dare to innovate?

Facebook is another example. Before Facebook and others like it, I never heard anyone suggest that their life was somehow incomplete because they didn’t have access to social media. Now, somewhat embarrassingly, many of us would ask how we ever did without it! We had no frame of reference to understand the new technology, so we didn’t know what we were missing.

So, how do you know if AI will benefit your organisation?

Test your ideas early

The answer to this question is not simple, but what I do know from building countless innovative solutions over the years, is that testing your assumptions early minimises the risk of investing significant effort and funds into an idea that ultimately fails. I have heard this approach referred to with many different names, depending on flavour of the day: “quick and dirty”, failing fast, proof-of-concept, prototype… I imagine you could add many more to this list.

I can’t profess to have conceived this approach. This basic concept has been around for many years and has been refined over recent years through the application of methodologies such as agile project delivery. I’m no scrum master, but to my mind the basic tenets continue to hold true:

  1. Take a user-centric view
  2. Identify the most difficult problems that need to be solved, and the wildest assumptions that need to be proved, to achieve a successful outcome
  3. Solve these problems and prove these assumptions first
  4. Repeat until you reach your goal


The Minimum Viable Product – or MVP for short

The current flavour of this concept is the MVP – the “quick-and-dirty” implementation re-invented.

I was recently re-reading Eric Ries’ excellent book, The Lean Startup. In it he explores in great detail why we should use MVP’s to test our thinking early. Eric suggests that MVP’s should not necessarily be used to test product or solution features per-se. Rather, they support a process of experimentation, the goal of which is to accelerate our learning on how an idea can be executed successfully. And while Eric’s approach is aimed at Entrepreneurs, he posits that Entrepreneurial approaches can exist as well in established businesses as they can in a small start-up. If you haven’t already done so, I would recommend you have a read (I have no affiliation with Eric – I just like his book).

An example: NLP analysis of Scout Reports

I recently used this technique to great effect as part of a project where we were looking to use Natural Language Processing (NLP) techniques to extract insights from talent scout reports for a national league football club. This club employs a number of talent scouts to observe and assess the performance of many up-and-coming amateur players each week. The reports are used to help the club determine which players to draft. Each scout report consists of a scored assessment of the player’s abilities in a number of different areas and a written commentary on the scout’s opinion of the player more generally. Each year, the club must review and analyse thousands of these scout reports during the drafting period. This manual process is inefficient, time consuming and can lead to inconsistent analysis and potential confirmation bias.

Hypothesis: We can use NLP to extract insights from sports scout reports

Instead of developing a fully operational system, I devised an MVP that tested our key assumption that we could use NLP techniques to extract relevant insights from the written commentary about the players. We developed this MVP in less than a week, proving we could automatically extract key insights from the reports. This paved the way to a greater understanding of what was required to build an effective production system.

When formulated carefully, MVP’s can be the key to the successful adoption of any new, innovative, disruptive technology in a business – particularly where it is not clear how these systems will be used. Not only do they support the rapid learning of what is important for the success of an AI system in your organisation, but used correctly they also promote the development of transition plans and cultural acceptance. These will be critical to the success of your bright, shiny new initiative.

Now, I am not for a moment suggesting that this alone will ensure the success of your adoption of AI technologies in your business. You will also need to be thinking about other techniques, such as taking a business/user centric view, thinking about the data that you can leverage and building the right team (read more here).  But including the use of MVPs early will certainly help you successfully leverage AI to your business’ advantage.

So, do you really need AI in your business? … Yes!

There are almost certainly areas of your organisation where different AI techniques will improve your business outcomes. Now you just need to understand how … and prove it. No doubt I am biased in my thinking, but it is easier than you might imagine. So take your first steps, if you haven’t already done so. Before your competitors do!

About the Author: Paul van der Linden is an Owner and Director of outThought.coPaul has a passion for finding innovative solutions that deliver the best possible outcomes for customers. As an architect and engineer, he bridges the divide between technology and business. Paul’s professional expertise includes cognitive systems, Natural Language Processing, Artificial Intelligence, solution architecture, business development, technical sales, technical leadership, enterprise architecture, Information management, analytics, cloud and hybrid solutions and immersive simulation.