In today’s digital and dynamic world, it’s impossible to discuss business strategy without mentioning AI. AI can automate simple, repetitive tasks, analyze sales patterns, predict supply needs, identify opportunities for growth, and provide dynamic pricing suggestions based on current demand and supply. AI is no longer just a “nice to have”—it’s fast becoming a competitive necessity. 

However, it can take time to fully embed AI into a business’s practices and strategy. It’s not enough to simply invest in the tools; there must be a clear understanding of how AI is integrated, used, and will evolve for an organization to succeed. An AI maturity model is a framework that helps organizations assess their current AI capabilities and plan how they will advance their journey to maximize the use of AI to boost efficiency, foster innovation, and make data-driven decisions.1 Different models outline anywhere from 4 to 7 stages; this version presents a streamlined four-stage interpretation:

1. Awareness/Experimentation: At this stage, companies are aware that AI may help improve the business, but there is no formal adoption. Employees may be using AI to assist with some tasks on an ad-hoc basis, discovering its potential benefits and limitations.2
2. Active/Operational: Organizations have now started implementing AI into their day-to-day tasks and across teams. For the most part, AI is helping to simplify and automate processes and generate reports through descriptive analytics that support decision-making.3
3. Expansion/Mature: At this stage, organizations have developed an AI strategy and are embedding it across teams. Teams are using AI for more complex tasks, and the organization may have begun developing custom AI tools in-house. Predictive analytics may be a component of this stage, with companies using AI to support future decisions.3
4. Leading/Transformational: AI is a prominent part of a business’s strategy for continuous improvement, woven into how the organization runs. Employees are comfortable using AI and it is part of the organizational culture. AI is being used to drive innovation and provide the organization with a competitive advantage over competitors.2

While the underlying logic of AI maturity models is the same, terminology varies across models. For example, Microsoft uses “Foundational – Approaching – Aspirational – Mature.”4 The key takeaway is that maturity is less about labels, and more about organizations assessing their capabilities and readiness. Maturity models are important as they provide direction for where and how companies need to collect data, what technology they need to invest in, required change management, and how AI feeds into their broader strategy. These models help companies identify where they are, how they can improve, and create a strategic roadmap to evolve to the final stage.5 

Artificial Intelligence: A branch of computer science in which computers are trained on vast amounts of data to mimic human comprehension, problem-solving, and decision-making. AI allows computers to perform tasks that historically only humans could do.7 

Descriptive Analytics: The collection and analysis of historical data to identify patterns and provide reports or visualizations that employees can use to understand what has already happened. Descriptive analytics is usually used in the first couple of stages of maturity, focusing on the past and “what” rather than the future and the “why.”8

Predictive Analytics: The use of statistics and modeling technology to predict future outcomes. Predictive analytics uses historical data to identify patterns and evaluate whether those trends will continue. Predictive analytics is often used to make future decisions in the later stages of AI maturity.9 

Change Management: A structured approach to communicating and implementing change, which can include the rollout of new technology or processes. It is important for organizations to effectively leverage change management as they move through the stages of AI maturity, ensuring that employees are properly trained to use the technology and reinforcing its position as part of the culture.10

Data Governance: Policies, procedures, and guidelines organizations develop to ensure the ethical, accurate, and secure collection, analysis, and use of data. Data governance is an important component of a comprehensive AI strategy.11

In 1936, British mathematician Alan Turing described a theoretical computing machine that could perform any logical process, laying the foundation for the idea that intelligence could be simulated by machines. Alan Turing described a computing machine that had limitless memory, could scan symbols, and depending on the symbol it saw, follow a set of instructions. The thought experiment showed that it was possible to make intelligence computable, suggesting that a machine could, in theory, perform any task that could be defined as a logical process. However, at this time, computers were not widely available or sophisticated enough, so the Turing machine remained an abstract concept.12

Over two decades later, American computer scientist John McCarthy coined the term “artificial intelligence” in 1955, defining it as “the science and engineering of making intelligent machines,” with the hope of invigorating top minds to focus on it as a distinct and formal area of study. McCarthy believed that any aspect of learning or intelligence could be simulated by a computer, similar to Turing, but with advances in technology, AI now seemed like a closer possibility. At first, researchers focused on building programs to solve mathematical problems, prove theorems, and play (and win!) games like chess. However, without sufficient access to real-world data, the applications of AI stalled for a number of years during a time that later became known as the “AI winter.”13

Throughout the 1970s and 1980s, the first expert systems—computer programs that use AI to solve problems in specialized fields—were developed as computing power improved and industries like healthcare began digitizing records. MYCIN, one of the earliest expert systems, was developed to treat blood infections. MYCIN could analyze symptoms and medical test results to diagnose patients.14 

In 2007, American business author Thomas H. Davenport published his book Competing on Analytics, urging organizations to use data analytics to assess performance and find opportunities for improvement. At the time, mostly technology companies were using this kind of analysis, but Davenport argued that business analytics could provide all organizations with a sustainable competitive advantage and framed it as a maturity journey: the more it is embedded enterprise-wide, the more mature the organization. He proposed a five-stage analytical competition pyramid:15

In 2008, Gartner, a research and advisory firm, introduced its Data Analytics Maturity Model to help organizations assess and improve their analytics capabilities. Although not AI-specific at the time, it laid the groundwork for later AI maturity models, outlining types of analytics from descriptive to prescriptive. The model included four types: descriptive analytics, diagnostic analytics, predictive analytics, and prescriptive analytics. The first two stages focused on past insights, whereas the latter two were future-focused, making predictions to help businesses make decisions. Not long after, as access to data increased and AI continued to become more sophisticated, leading consulting companies like McKinsey, Deloitte, PricewaterhouseCoopers (PwC), and IBM developed their own maturity frameworks. AI was becoming more powerful, requiring companies to develop structured ways to navigate complexity and embed it into their strategy.15

Today, AI maturity models and strategies often incorporate data governance frameworks and establish ethical guidelines for using AI, ensuring that companies not only scale their technical capabilities, but also do so in a way that aligns the use of AI with broader social and organizational values.

Alan Turing

A British mathematician and logician who is considered the “founding father” of AI for developing the Turing machine in 1936. Turing also designed the Automatic Computing Engine (ACE) in 1945 when he worked for the National Physical Laboratory; this became one of the most influential digital computer designs. Unfortunately, Turing’s ideas were beyond his time, and the lack of technological advancements meant that many of his designs were not fully realized until much later.16 

John McCarthy

An American computer scientist who coined the term “artificial intelligence” in 1955, with the belief that human intelligence could be mimicked by a computer. McCarthy also developed the computer programming language LISP in 1958, which could process symbols and lists, making it ideal for early AI research. McCarthy continued his research into developing computer languages and also supported the development of Elephant 2000, a programming language that could process speech acts with the goal of making computers better at understanding human inputs and communication.17 

Thomas Davenport

A renowned American author and consultant on topics such as business analytics, information and data management, and enterprise systems. His book, Competing on Analytics, was foundational in bringing data analytics beyond tech companies and provided one of the first maturity models, although it was focused on analytics more broadly than just AI. In 2007, Davenport was named one of the most influential people in the information technology industry by the Ziff Davis magazine. 18

The adoption of AI maturity models profoundly shapes how organizations manage their AI journeys, influencing strategy, risk management, and innovation. By providing a clear framework to assess capabilities and guide growth, these models help organizations unlock AI’s full potential while mitigating pitfalls.

Enhancing strategic decision-making

By developing an AI maturity model, organizations can ensure that they are integrating technology and processes strategically rather than as a reactive response to advancements. Instead of viewing AI as an independent element of evolution, maturity models make it possible to integrate it into long-term strategic planning.

If you were setting out on a road trip without a clear plan, without even knowing where you currently are and where you want to go, it’s very likely you’d hit a lot of roadblocks (pun intended) and have an inefficient journey. Similarly, adopting AI without a clear understanding of your organization’s current status, capacity, and goals makes it difficult to most effectively leverage AI to improve processes. AI maturity models can help companies avoid failure.

Becoming proactive 

Developing an AI maturity model is not just about knowing where an organization stands today, but also where it would like to go in terms of its AI journey. As AI is constantly evolving, there will be future advancements that can help businesses innovate and transform. Having a maturity model allows companies to stay up-to-date with trends and work towards continuous improvement. Instead of scrambling when a new tool is introduced, a company will have an understanding of the role AI plays in evolving and reshaping its competitive landscape.2

An assessment tool can help businesses adapt to changing technological landscapes and ensure long-term AI sustainability, making organizations more future-proof. As AI tools can also predict future trends and risks, the more mature an organization is in its journey, the better it can tackle disruptions in the market. 

Establishing strong data governance frameworks

Adopting AI comes with significant risks. There are data privacy and security risks, and when AI is integrated without a clear understanding of its function, the lack of transparency undermines trust. By developing an AI maturity model, organizations can ensure that any security or ethical risks are addressed before technology is adopted or scaled, and can develop data governance frameworks.2 

When AI chatbots like ChatGPT and Copilot first emerged, most companies did not have processes in place for how they should be leveraged in day-to-day work, meaning that many employees used them without the knowledge of their organization. A McKinsey report showed that three times more employees were using generative AI for a third or more of their work than their leaders believe, indicating a lack of awareness among leadership about how these tools were being used. 19

As many of these platforms are open source, unmonitored use can lead to data breaches. For example, Samsung discovered that some engineers had uploaded sensitive code to ChatGPT, which led to a subsequent employee ban from using such tools.20 This halts the potential for efficiency and innovation, but if Samsung had had a clear plan in place for integrating AI into the organization and guidelines for how to use it responsibly, it could have avoided this breach. 

AI maturity models are powerful tools for guiding organizational AI journeys, but they also spark debate and challenges. Understanding these controversies reveals the complexities behind measuring readiness, adapting frameworks, and ensuring fair assessments.

Measuring maturity versus impact

While AI maturity models can provide organizations with a better sense of their journey with AI, some critics argue they can focus too much on internal capabilities, such as data infrastructure, talent capabilities and governance, instead of on how AI is creating business value. An organization can score well from a readiness perspective if it has centralized processes, standard guidelines, and the right infrastructure, but unless that translates to impact for their consumers, we may mistake preparedness for success.21

Organizations can incorporate outcome-based metrics, such as cost savings, increased retention, or improvements in predictive models, to ensure that maturity is measured by impact. This was a key message of Thomas Davenport’s book, Competing on Analytics, where he emphasized that businesses must have a clear understanding of their performance drivers and performance metrics, in order for analytics to become a differentiating value-add.22 

Frameworks in a rapidly evolving field

While frameworks like AI maturity models attempt to provide a standardized roadmap that organizations can use to assess their readiness and performance, this can mean that they are too static to keep up with a rapidly changing landscape. This is especially true considering the rapid pace of technological shifts and advancements in AI, where a company can quickly fall from being in a “mature” stage to an “experimentation”stage. Additionally, a generic framework can be challenging to integrate meaningfully into an organization’s unique fabric.

Instead of a one-size-fits-all static approach, organizations should tailor the model to their specific needs and circumstances, ensuring that they first have a clear understanding of their own goals for their AI journey, and should frequently reassess and adjust based on strategic direction and external shifts.

Subjectivity in assessing maturity

Although it’s important for organizations to tailor AI maturity models to fit their culture and vision, it also means that it’s hard to establish a benchmark to compare against competitors. If one of the drivers for adopting a maturity model is to understand how AI can be integrated as a differentiator, then we need to be comparing apples with apples, not apples with oranges.

Moreover, AI maturity models often rely on self-assessments and employee surveys to understand current maturity. These are subjective tools for measurement, which can lead to an inaccurate assessment. For example, researchers found that a telecom company in Europe rated itself at the highest stage of AI maturity, yet it has struggled to achieve its strategic goals for AI. There can clearly be a disconnect between where an organization thinks it falls within the model and where they actually stand.23

Identifying areas for improvement for Booking.com

Booking.com, an online platform that allows people to book accommodations, flights, and car rentals, has hundreds of different machine learning systems running to personalize recommendations, translate content, provide dynamic pricing, and other applications. With so many machine learning systems in place, it’s difficult for the company to have a clear understanding of how AI is being used across the organization or to evaluate its performance. 

Booking.com developed a machine learning (ML) Quality Python package that generates a standardized report for each machine learning system and provides a color-coded quality score, ranging from red (gaps that need to be filled to reach the next maturity level) to green (no gaps identified). This helped the organization develop a quality and maturity AI framework that incorporated the hundreds of systems being used across independent production teams. The framework allowed Booking.com to roll out a two-year AI strategy that included conceptualization, policy implementation, and tooling. The plan also included training and behavioral modeling, as a successful roll-out requires a shift in culture and collaboration among stakeholders. 

One of the systems that was analyzed was the departure airport recommendations model designed to help consumers booking flights. The Python package identified that the recommendation often defaulted to the most popular airport in the dataset (e.g. for Toronto, it would recommend Toronto Pearson Airport, even if Billy Bishop Airport may have been more suitable for the consumers’ specific needs). Booking.com was able to identify the issues with the machine learning system, adapt it, and test it to ensure it provided more personalized recommendations. 

Booking.com’s development of an AI maturity model shows how taking a step back and looking at how AI is used comprehensively across the organization can help address challenges and implement improvements to bring it into a more mature stage.24 

Model for responsible AI maturity

As AI becomes more present in our day-to-day lives, used by businesses worldwide, it is becoming increasingly important to ensure AI is adopted responsibly to avoid ethical risks. However, the development of tools to assess trustworthiness and ethical use of AI can be time-consuming to implement. 

Researchers aimed to develop a more practical tool to support organizations and created the Industry-Wide Maturity Model for Responsible AI. The tool helps organizations determine which of the following four stages of Responsible Artificial Intelligence (RAI) they are in:

• Unaware: the organization is not aware of RAI, and it is not incorporated into its assessments
• Exploratory/reactive: the organization has a reactive response to external pressures regarding RAI
• Proactive: the organization recognizes the benefits of RAI and incorporates them into business processes
• Strategic: the organization has adopted RAI as a component of its strategic framework

Researchers piloted the tool on a company in Portugal that provides information technology (IT) solutions for the financial services sector. The company was in the process of piloting its first AI development projects, including the use of voice interactions and chatbots, and predicting which financial products salespeople should promote to clients. The tool demonstrated that the company achieved the exploratory level for nine requirements of responsible AI use, but was in the unaware phase for ten other requirements, as visualized below:

The tool allowed the company to identify which areas needed to be addressed to move the company up to the exploratory level of responsible AI use, such as ensuring that risk management is integrated into AI use, and that environmental wellbeing is considered for the products it sells.25

The Potential and Pitfalls of AI in Healthcare

AI has the potential to transform a number of industries, including healthcare. Integrating AI can reduce the workload for healthcare workers, improve patient outcomes, and increase accessibility to services. However, it also has the potential to perpetuate bias and pose risks to data privacy, which demonstrates why it’s important to embed AI in a strategic, well-thought-out manner. In this article, Sophie Cleff explores the potential and pitfalls of AI in healthcare and provides guidance on how AI can successfully be adopted. 

“It’s Complicated:” An Ode to Our Relationship with AI

Many of us have a complicated relationship with AI. On the one hand, it can help make our lives easier, provide us with access to information, or act as a personal assistant (Hey, Alexa!). However, we often fear what we don’t understand, and AI is very new. There are concerns that it will take over our jobs, or worse, the world. In this article, Charlotte Sparkes explores her own journey with AI, provides an overview of important terms, and explores some of the reasons why healthy skepticism towards AI can be beneficial.

1. Mekala, B. (2025, February 6). Understanding AI maturity levels: A roadmap for strategic AI adoption. United States Artificial Intelligence Institute. https://www.usaii.org/ai-insights/understanding-ai-maturity-levels-a-roadmap-for-strategic-ai-adoption
2. Damco Solutions. (2025, June 18). Understanding AI maturity models: Stages of AI maturity. Damco Group. https://www.damcogroup.com/blogs/understanding-ai-maturity-models#stages
3. Burnham, K. (2025, February 25). What's your company's AI maturity level? MIT Sloan School of Management. https://mitsloan.mit.edu/ideas-made-to-matter/whats-your-companys-ai-maturity-level
4. The Pecan Team. (2024, April 4). What is AI maturity, and how can you accelerate it? Pecan AI. https://www.pecan.ai/blog/what-is-ai-maturity-how-accelerate/
5. Aboze, B. J. (2023, May 23). Understanding the AI maturity model: Advancing your organization's AI capabilities. Deepchecks. https://www.deepchecks.com/understanding-the-ai-maturity-model-advancing-your-organizations-ai-capabilities/
6. Harroch, D. A., & Harroch, R. D. (2025, April 25). 15 quotes on the future of AI. TIME. https://time.com/partner-article/7279245/15-quotes-on-the-future-of-ai/
7. Stryker, C., & Kavlakoglu, E. (2024, August 9). What is artificial intelligence (AI)? IBM. https://www.ibm.com/think/topics/artificial-intelligence
8. The Investopedia Team. (2024, August 29). Descriptive analytics. Investopedia. https://www.investopedia.com/terms/d/descriptive-analytics.asp
9. Halton, C. (2025, March 3). Predictive analytics. Investopedia. https://www.investopedia.com/terms/p/predictive-analytics.asp
10. Iacoviello, A., Whiteford, E., & Downie, A. (2024, July 4). What is change management? IBM. https://www.ibm.com/think/topics/change-management
11. Databricks. (2025). Data governance: A comprehensive guide to the processes, policies and tech that organizations use to manage and get the most from their data. Databricks. https://www.databricks.com/discover/data-governance
12. Copeland, B. J. (2025, July 30). History of artificial intelligence. Encyclopaedia Britannica. https://www.britannica.com/science/history-of-artificial-intelligence
13. Coursera Staff. (2025, May 23). The history of AI: A timeline of artificial intelligence. Coursera. https://www.coursera.org/articles/history-of-ai
14. Copeland, B. J. (2025, July 30). MYCIN. Encyclopaedia Britannica. https://www.britannica.com/technology/MYCIN
15. Namkoong, M. (2019, February 14). Competing On Analytics. Medium. https://medium.com/@daygamedestruction/competing-on-analytics-f54c98c9b9f
16. Copeland, B. J. (2025, June 22). Alan Turing – Computer designer. Encyclopaedia Britannica. https://www.britannica.com/biography/Alan-Turing/Computer-designer
17. Copeland, B. J. (2025, June 22). John McCarthy. Encyclopaedia Britannica. https://www.britannica.com/biography/John-McCarthy
18. Davenport, T. H. (n.d.). Thomas H. Davenport. IBM Center for The Business of Government. https://www.businessofgovernment.org/bio/thomas-h-davenport
19. Mayer, H., Yee, L., Chui, M., & Roberts, R. (2025, January 28). Superagency in the workplace: Empowering people to unlock AI’s full potential. McKinsey & Company. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work
20. Gurman, M. (2023, May 2). Samsung bans generative AI tools for staff following data breach concerns. Yahoo Finance. https://finance.yahoo.com/news/samsung-bans-generative-ai-staff-004831399.html
21. Yates, V. (2024, April 18). Mastering the machine: AI maturity models for organizations. Advertising Week. https://advertisingweek.com/mastering-the-machine-ai-maturity-models-for-organizations/
22. Davenport, T. H., & Harris, J. G. (2021, January 21). Competing on analytics; The new science of winning. Principus. https://principus.si/2021/01/21/competing-on-analytics/
23. Elsevier. (2025). Maturity assessment. In ScienceDirect. https://www.sciencedirect.com/topics/computer-science/maturity-assessment
24. Castelli, A., Chouliaras, G. C., & Goldenberg, D. (2025, February 12). Maturity framework for enhancing machine learning quality (arXiv:2502.15758v1). arXiv. https://doi.org/10.48550/arXiv.2502.15758
25. Ferreira, R. M. F. D., Grilo, A., & Maia, M. (2025). Piloting a maturity model for responsible artificial intelligence: A Portuguese case study. Journal of Responsible Technology, 22, 100117. https://doi.org/10.1016/j.jrt.2025.100117