Bridging the Gap Between AI and Business Value

By: Elin Hauge, AI and Business Strategist

When OpenAI launched ChatGPT in November 2022, suddenly, 100 million users globally had first-hand experience with a technology that, for most people, appeared to be almost magic. Today, about a year later, a lot of people still struggle to understand how artificial intelligence (AI)  works with text and language. The outputs of ChatGPT and its peers (e.g., Bard, Bing, Ernie, Jais, etc) are often described as almost godlike, sentient, on the path towards super-intelligent machines that will take over humanity.

To bridge the gap between AI as a technology toolbox and the business value that may be reaped from utilizing this emerging technology, we need to call a spade a spade; AI is mathematics combined with computer science.

Fifteen years ago, big data was the hot topic. Then, about a decade ago, digital transformation came along. As a result, companies now typically have both the data and the digital processes in place. Therefore, the natural next step is to apply mathematical algorithms to this data and intelligently analyze, assist, augment, and automate digital processes. And we call it artificial intelligence.

AI Here and Now?

But what is different now? Fundamentally, three factors are different now compared to even only a year or two ago:

1. The world produces about 3.5 x 10^18 bytes of data per day. That's a million, million, million bytes in a day! These enormous amounts of data serve as training material for the Large Language Models (LLMs), which are the engines for ChatGPT and its peers.
2. We have access to unprecedented processing capacity in the many cloud data centres, powering the training and operation of AI algorithms.
3. The machine learning methods and architectures have significantly improved and are still improving month by month, increasing the capabilities and precision of the AI algorithms.

Large language models, such as GPT-4 powering ChatGPT, utilize the structures, correlations, and inherent semantic relations in languages to build prediction models for text. Although this may seem almost divine to many people, it is “only” advanced statistical models for language.

We may look at LLMs as the next-generation information models based on semantic language. For professional services firms, this opens a treasure trove, access to the vast knowledge buried in the chest of competence and knowledge collected over years, perhaps decades.

AI is More Than Just ChatGPT

Although ChatGPT, in particular, and LLMs, in general, have been the focus of the media and LinkedIn feeds for more than a year now, the AI toolbox contains so much more. It is important to keep in mind that AI has been around for decades, and many use cases are not about language processing or generation. Here is a very rough summary of the main categories of tools in the AI toolbox:

• Prediction models based on structured data (data in rows and columns), such as payment fraud detection, e-commerce and music recommendations, and predictive maintenance of machinery.
• Classification models, such as email routing, customer churn analysis, and spam filters.
• Computer vision models, such as CT scan analysis, facial recognition on your smartphone, and reverse vending machines. The latest developments in this field include the ability to generate images as well, such as Stable Diffusion, DALLE-2, and MidJourney.
• Natural language processing (NLP), such as sentiment analysis in marketing, text search, and ChatGPT (and all other LLMs). The latter is an example of the ability to generate text, not just interpret and analyse.
• Frequency models/time series, such as stock price forecasting, climate models, and error detection in industrial equipment. In this category, we can also find the capability to generate music.

Although simplifying what AI is (and is not) enhances the general understanding of the potential of these technologies, it should also be emphasized that the AI toolbox is extremely powerful. In fact, AI is potentially more disruptive than any other set of tools humanity has previously had access to. The big question for businesses remains, though: how to embrace AI in a value-creating and profitable way?

Embracing AI to Create Value and Stay Profitable

Let's start from the top; any company should maintain a laser-sharp focus on its overarching objectives. Although AI may change the dynamics of the market in which you operate, you still need to keep your spotlight on the purpose of the company and avoid giving in to the temptation of diverting your attention to shiny new toys. With that in mind, when you decide to leverage AI tools in your business, it may be wise to divide your strategy formation process into three parts:

• Refine your core business: How can you protect, improve, and strengthen your core business and with AI in mind, which data and digital processes do you have access to that can be utilized to increase revenue, improve efficiency, and reduce costs? Then, dig into the AI toolbox and decide which tool(s) may be most relevant to achieve this. It may not be the shiniest and coolest tool(s) that will help you the most, as precision and consistency will be of high importance.
• Grow the core: Ask yourself, how you can expand your core business with new products, services, geographies, business models, etc. The next question is, which data and digital process do you have access to that can be utilized to enable this expansion? Which tools in the AI toolbox may be most relevant to achieve this? You may perhaps be willing to apply tools that have more of an R&D nature in your specific context.
• Explore new business opportunities: When starting using AI tools, the investment is likely to be experimental, and data quality could be a challenge. To be successful in exploring and evaluating new business opportunities using AI, it’s important to keep this separate from the business’ core KPIs and key incentives, as they are likely not in alignment in the beginning.

This simplistic approach to the role of AI in strategy formation may seem naïve, but over and over again, companies find themselves in conflicting internal discussions between the need to streamline and optimize everyday business on the one hand and the need to innovate on the other. Innovation often carries risk, and in the space of AI, things can quickly become messy, complex, and risky in the exploration and pilot phase.

For example, in professional services firms, AI could raise ethical concerns around client information safety. If managed properly, AI proposes real efficiency opportunities and a competitive advantage, but if AI is not governed properly, the security risks are stronger and the benefits significantly less.

This triggers the necessity to redefine both delivery and business models, products/services, and business areas. AI is highly likely to play a key role in these developments in many shapes and sizes. However, trying to do radical innovation with AI in the middle of your operational business is most likely going to create a lot of friction, reduce your operational focus – and hence performance – and restrict your ability to truly explore new territory.

Instead, let your operational teams play it safe and integrate AI tools that are close to and enhance their core business processes. Put business development and AI resources to work on the growth scenarios that support your core business, with room to test out more novel approaches. Finally, keep the most disruptive innovation initiatives and their exploration of AI tools separate. Let them play with the AI toolbox in an environment where potential failures do not leak into the rest of your business and cause interruptions.

The bottom line is that strategy formation should take different paths for different purposes to balance operational excellence with radical disruption. The role of AI is ubiquitous, albeit some parts of the toolbox are more exploratory than others. Therefore, conscious considerations around the choice of AI tools may be of importance.

AI is the natural next step in the digital transformation journey, and it is a toolbox with a wide range of capabilities. AI is, therefore, not a strategy but an enabler of your strategy.