A Leader's Framework for AI Application in Business

In this post, I will explain artificial intelligence concepts, benefits, applications and uses in business beyond just the technological aspects. I will also dive into a leader's framework that serves as a guide for building an AI-enabled, innovative organization in a digital transformation era.

Artificial intelligence (AI) is increasingly becoming a must-have technology for many organizations. As a consultant myself, leaders ask me a lot of questions around the future and adoption of AI in a business context. I think the hype around AI recently is being driven by the diffusion of consumer generative AI applications, such as ChatGPT and Google Bard, that are putting pressure on many leaders to think about adopting AI solutions for their organizations. Apart from the hype in consumer AI applications, AI is considered to be a general-purpose technology meaning that it can be applied in many business areas and in a wide range of industries, and it has the potential to alter an entire economy and the society at large. AI is currently being deployed in many industries, including retail, healthcare, transportation, manufacturing, and even in oil and gas, agriculture, arts, and many more. Studies show that AI's value creation will reach $13 trillion by 2030, and the job market for AI-related jobs is on the rise. These stats indicate that AI is here to stay beyond the current hype, which explains why a growing number of organizations are deploying in-house AI projects or are being interested in adopting some form of AI solutions as part of their digital transformation plans. But the challenge for many leaders is how and where to get started in this uncharted territory.

Through my work in advising midsize companies in the Middle East on innovation adoption and transformation projects, I have noticed that some organizations struggle in answering key questions when it comes to AI adoption and application in their business area. Based on my discussion with middle and top managers within these organizations, I can safely say that 'How?', 'What?', 'When?', and 'Who?' types of questions come up frequently, and these more often kickstart the AI project machine.

Typically, these leaders are skeptical and ask specific questions such as:

'What can artificial intelligence do to help us innovate, outcompete, and improve business performance? Is AI even good for our business and industry?

What do we need for this type of project? What capabilities?

How and where to start? Is now a good time to deploy AI?

How much do we need to invest and how to justify ROI?

What is the impact on our organization and transformation project? Who will be impacted?

What are the expected risks and challenges given the controversies around AI?'

At this point, hesitation is understandable because of the potential serious investments, work disruptions, and business changes that a new tech could bring to the way of doing things. The AI inquiries that follow are usually meant to understand usefulness, clarify uncertainties, overcome challenges, and more importantly maximize business benefits as opposed to adopting AI for the sake of adopting a trending tech.

Let's first understand AI concepts and uses in business, and then discuss the ways it can be integrated and deployed in organizations to help businesses derive greater value.

Understanding Artificial Intelligence

What is AI? Artificial Intelligence is about getting computers to do tasks that normally require human intelligence. It's usually done through analyzing vast amount of data using machine learning to augment human intelligence and help people do a better job.

There are three types of AI: Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI), and Artificial Super Intelligence (ASI). We can think of these as ANI or narrow machines that can do specific tasks with close to human intelligence capabilities, AGI or general with equal capabilities to human intelligence, and ASI or super that exceeds human intelligence capabilities. At this stage, we are still in the ANI phase and most AI applications available today in the market are of the ANI type meaning that they can do specific narrow or weak tasks, such as self-driving cars, chatbots, smart speakers etc.

What is Machine Learning? Machine learning (ML) is a subset of AI that uses algorithms to learn from datasets and reach certain conclusions. The machine maps the path from X (input) to Y (output) points based on training and learning without being explicitly told or programmed to do so.

What is Deep Learning? Deep Learning (DL) is a subset of ML that uses artificial neural networks to learn from datasets and extract more features from data than ML would.

Learning can be supervised, unsupervised or reinforced. Supervised learning means that we have to train the machine by labelling input data (X) and output data (Y). Unsupervised learning means that we do not have to label output data (Y), and the machine can learn from identifying patterns in the input data (X) to reach a conclusion. Reinforcement learning means that we have to tell the machine if the course of action taken is correct or wrong through reinforcement techniques.

Now, let's discuss the business implications and uses of AI rather than just discussing the technology itself.

'No technology is good on its own if not put to good use.'

What AI can and cannot do?

Contrary to the belief that AI is a super power that can do magic, AI's capability is limited to learning from the data that feeds into it whether structured or unstructured data. This can be image, video, audio or text type of data. As it stands now, AI is still in the narrow (ANI) stage, and this means it can do specific, single tasks. So, think of tasks that humans can probably do and transfer that to an AI system. For example, chatbots or searches can be done by AI simply because we can feed the machine historical data from previous interactions between humans, and the machine can learn to converse or return search results just like what a human would do but in a much efficient and faster manner.

What AI cannot do? AI cannot currently converse with empathy, for instance. This is because we cannot feed the machine implicit information and datasets on every occasion where a user asks questions and receives empathetic answers from a real person on the other side of the conversation. At least for the time being, AI systems cannot do things that a human being can do perfectly. It can augment human tasks and can do a pretty good job with speed and efficiency to help humans make better decisions, but it cannot replace a human being especially with complex and detailed jobs, such as research and analysis.

AI Applications in Business

When thinking about applying AI to business activities, we have to go back and think about the data. Data quality and availability are paramount in training an AI system to do specific business tasks traditionally done by humans.

Data comes first, whether it's visual, audio or text. If we have a fair amount of quality data, we can probably train an AI system to perform business activities, automate routine tasks, and streamline processes to increase productivity, reduce costs, and improve customer experience. Today's technology permits AI systems in business to perform specific business tasks by being trained with a mix of datasets, and these AI applications can be classified broadly as follows:

• Speech or voice recognition
• Image recognition or visual inspection
• Text recognition
• Speech to text and text to speech
• Machine translation
• Robotics
• Self-driving cars and others

Let's take the retail industry, for example, and see how AI is being applied in multiple areas to optimize business processes and enhance customer experience.

In the front-end, AI is helping retailers enhance customer experience and raise productivity either by deploying AI-powered assistants to help call center agents find accurate answers to customer queries quicker, mainly over the phone, or by deploying AI-enabled chatbots that attend to customer queries 24/7. These AI chatbots are trained using NLP or natural language processing giving the system the ability to understand text or spoken words in much the same way a human can. On the marketing front, AI is helping retailers personalize the customer experience by building better recommendation systems and curated content matching the different needs and tastes of their customer segments. Customer retention and loyalty are important for retailers, and AI is helping retailers increase loyalty and improve online advertising through AI systems that use a combination of speech, image, and text recognition applications to crawl and screen multiple online sources, including social media platforms, to curate a personalized customer experience.

In the back-end, AI-powered robots, and image and text recognition tools are being used in warehousing to streamline operations, speed up packaging, handling, and delivery. Demand forecasting is another area where AI is helping retailers predict demand with a high level of accuracy reducing inventory pileup and optimizing inventory management.

AI and Innovation

Innovation is an area where many organizations are looking for improvements. First, let's differentiate between the three different types of innovation: incremental, sustaining, and disruptive innovation as characterized by Clayton Christensen.

Christensen places innovation activities on a continuum that distinguish between the three types of innovation. Incremental innovation entails slight changes to processes or products and services. Sustaining innovation entails significant changes and upgrades to products and services, whereas disruptive innovation entails an entirely new or novel products or business models that could disrupt an entire industry.

AI can help organizations improve incremental and sustaining innovation with sustaining innovation being the sweet spot. Radical and disruptive innovations are considered novel or entirely new to the world and no historical data exists around these innovations, which makes AI ineffective for these types of innovation. Since AI relies on past data, it makes sense to deploy AI to fuel incremental and sustaining innovation given the availability of historical data from previous products or services.

When leaders ask me if AI can help them innovate better, I usually try to understand what type of innovation they are talking about. If they aim for sustaining innovation, then AI can probably help them reach their goals. But if they are looking for those radical innovative products or services that are new to the world, then AI is probably not the ideal solution. This all depends on the scope of the project, which leads us to the next topic of discussion and that is a leader's framework for AI application.

Leader's Framework for AI Application

I like to think of this framework as a guide to simplify the process of applying AI in an organization and accelerate adoption. This helps leaders have something to work with, ask the right questions, clarify uncertainties, spot potential challenges, and enable an AI transformation that derives greater business value for their organizations.

Now, this does not mean that there aren't good AI guides for leaders out there. There are pretty good and excellent guides available in the market, but I find these guides to be a little bit challenging in practice in the Middle Eastern market due to cultural nuances. Risk aversion and tech adoption are quite high in the region, and when I'm working with leaders in the region using off-the-shelf guides it seems that these cultural nuances are sometimes overlooked. So, I came up with a customized leader's framework for AI application in business that worked pretty well for me so far in this part of the world.

I like to call it the triple S (SSS) framework: Scope, Strategy, Structure. These represent the three areas that we usually focus our efforts on to address leaders' questions (identified in the AI kickstarter wheel), clarify uncertainties, and brainstorm relevant solutions for implementing effective AI projects in organizations.

Having such an AI framework or model on hand can assist leaders with thinking about potential challenges and remedies, consolidate fragmented AI deployment initiatives and data sources, define clear AI objectives and strategies, prove the business case, present project's benefit and value, and accelerate adoption.

Now, let's briefly discuss how to use this framework.

Scope

The scope is the area where we spend a significant amount of time identifying what do leaders want AI to do, what it is doable and not doable, what data sources are required, and what are the expected outcomes? Here's where technology meets business to come up with clear objectives around the project purpose and value in business and technology terms, which can be classified under one of the following AI broad themes (i.e. What do we want AI to do for users?)

• Speed up research and finding
• Expand knowledge and expertise
• Recommend with high level of accuracy
• Enhance communication and interaction
• Detect and mitigate risks
• Predict and preempt problems

Since today's AI systems are capable of either predicting or recommending outcomes to help users make more informed decisions, similar themes are grouped together to clarify the AI project scope and expected outcome.

Typically, through workshops or group discussions we can reach consensus and team alignment around the AI project scope and objectives. It is worth mentioning that people from diverse backgrounds and different business functions are encouraged to participate in these workshops in order to define a clear scope for the AI project. AI training sessions also play a key role in helping leaders understand AI and identify the scope of work.

Strategy

Defining a clear scope helps us formulate an AI strategy to address the how to achieve these objectives and build an AI-enabled company.

Typically, questions like 'What resources are needed? What infrastructure? What is the best course of action? How to close the gaps? How to acquire or improve capabilities?' are addressed.

Project cost and investment questions are also addressed. For example, we can identify and set priorities, and then decide to adopt off-the-shelf AI tools that are affordable and can save businesses a lot of money and time on training AI systems internally. This can be a good starting point for some organizations as they start with smaller projects and then move on to bigger ones at a later stage.

This is also the right time to align your AI strategy with your business strategy. In the sense that you don't want your AI strategy to misalign with your business strategy. If your AI strategy is at odds with your business strategy, then it's better to revise your strategy in order to avoid confusion and skepticism on behalf of naysayers.

Structure

Adopting a new technology usually involves change. Change impacts people, structure, and processes, and we know that people hate change. When discussing AI adoption with leaders, they immediately refer to the incidents and pains they previously had to go through with getting people to adapt to change. They want to know the impact, identify skepticism, and remedy any potential harm or disruption.

Typically, controversies about AI legal, ethical, and social risks also come up at this stage. The best course of action here is to integrate explainable AI techniques to mitigate any current or future regulation and infringement risks.

Depending on the scope of work, some AI projects could be transformative in nature meaning that they could change the course of a business and the business model altogether. In this scenario, huge re-structuring efforts, job changes, and infrastructure development are required with a fair amount of time spent on planning and execution. Usually, I do not advise novice organizations to start by adopting AI projects at scale because of the implications this approach has on these organizations.

An incremental approach to AI is more suitable for these types of organizations at this stage, as it entails minor re-structuring and change interventions. They can start with smaller AI projects, get the low hanging fruit first, and then scale when the team accumulates the necessary knowledge and experience. Also, AI training here can assist in getting people to speed up adoption, overcome resistance, and prepare for AI projects at scale later on.

Written by Elias Hayek

Management, Strategy, and Innovation Consultant

#innovation #artificialintelligence #ai #aileadership #aiinbusiness #machinelearning #deeplearning #nlp #naturallanguageprocessing #aiethics #data #dataanalytics #datascience #robotics #retail #technology #tech #digitaltransformation #chatgpt #googlebard #bard #chatbots #framework #aimodels #strategy #middleeast الذكاء_الاصطناعي#