AI enters the mainstream

AI is on the mind of many organizations, and for good reason. It can provide a lot of value in terms of cost efficiencies and innovation. Generative AI has certainly added visibility to AI work that has been going on for decades. In a 2024 TDWI survey, for instance, generative AI was ranked higher than machine learning in terms of an analytics priority — and it was in the top 4 priorities for analytics in organizations.

So, one question I had is, “Are organizations ready for AI?” I’ve seen in our research that many organizations we survey are still struggling to move past self-service analytics. I’ve seen that a top priority for data management is getting data ready for machine learning and AI. And yes, while organizations can make use of generative AI to help with content generation, summarizing and analyzing content, and some other simple use cases, the reality is that they are going to need to use their own data to really drive value in AI applications.

This means that they need to have a solid data foundation in place, not only for their structured data, but their unstructured data too. This also means that they will need to have the development and analytics skills in place to build applications. Data and AI governance will also be critical. We are starting to see projects fail already because of these factors.

The AI Readiness Assessment

I started creating assessments at TDWI ten years ago when I joined the company. The first one I created with one of our faculty, Krish Krishnan, about big data maturity (remember big data?). This past year, I created the AI Readiness Assessment.

The purpose of the AI Readiness Assessment is to evaluate an organization’s preparedness for adopting and leveraging AI, focusing on five key dimensions: organizational readiness, data readiness, skills/tools readiness, operational readiness, and governance readiness. You can go to the link provided here and take the assessment. There is also an accompanying guide. The assessment has about 70 questions. After you take the assessment, you’ll get an overall score as well as a score in each dimension, that will let you know (in a rough way), how ready your organization is for AI.

To date, the AI Readiness Assessment results indicate a median readiness score of 62/100 indicating that most organizations taking the assessment are in the “Standardizing” stage, where AI initiatives are underway, but significant gaps in operationalizing AI, developing data infrastructure, and AI governance remain. The assessment highlights that, while progress is being made, many organizations are not yet fully equipped to deploy AI effectively at scale, particularly in areas like data accessibility, AI skills development, and responsible AI governance.

A break-down of each dimension (each worth 20 points) reveals the following:

Organizational Readiness

The Organizational Readiness dimension measures how well leadership, strategy, culture, and funding align with AI goals. In the 2024 assessment, the average score for this dimension was 12.6 out of 20, placing organizations at the latter stages of the “Standardizing” phase. This score indicates that many organizations have started exploring AI and understand its potential value. Leadership is generally committed, with respondents scoring high on leadership understanding of AI’s impact. However, the assessment found that only 30% of organizations have an AI strategy in place and are executing on it, while the rest are either developing plans (37%) or have no plan at all (33%). This highlights a disconnect between leadership enthusiasm for AI and the resources and planning needed for widespread implementation. Additionally, fewer than half of respondents indicated that their leadership has committed the necessary funding to support AI initiatives. A cultural shift towards innovation and adaptability remains a challenge, with many organizations still working to develop a mindset conducive to AI-driven innovation.

Data Readiness

The Data Readiness dimension measures whether organizations have the data infrastructure and data accessibility needed to support AI. This dimension scored 12.2 out of 20, indicating that many organizations are making strides in building data platforms but still face challenges. While organizations are investing in cloud data warehouses and modern architectures to support AI, only 42% of respondents said they have systems in place to ensure data accessibility from diverse sources. Data silos remain a significant obstacle, with many organizations still struggling to integrate structured and unstructured data for AI use. While some organizations have begun building enriched datasets, most are still focused on structured data, which is sufficient for early AI adoption but insufficient for more advanced use cases. As AI models rely on diverse data types (e.g., text, image, and voice data), building a robust data infrastructure will be critical for advancing beyond predictive analytics.

Skills/Tools Readiness

In the Skills/Tools Readiness dimension, which assesses whether organizations have the technical skills and tools to support AI, the average score was below 11 out of 20, reflecting significant gaps. Many organizations are still working to develop core AI competencies. For example, only a third of respondents reported having data science skills in place, while another third are in the process of acquiring them. The remaining third do not yet have these skills. Similarly, data engineering and DataOps/MLOps skills, which are critical for building data pipelines and operationalizing AI models, are lacking in nearly half of the organizations surveyed. Some companies are relying on external vendors to fill these skill gaps, while others are developing internal teams. Moreover, less than half of respondents indicated that their organizations provide regular AI training for employees, underscoring the need for more comprehensive AI literacy efforts to build a workforce capable of supporting AI initiatives.

Operational Readiness

Operational Readiness focuses on the organization’s ability to move beyond AI model development and into production, where AI can deliver real business value. The score for this dimension was 11.9 out of 20, placing organizations in the early stages of operationalizing AI. While many organizations are starting to build AI models, they are not yet equipped to deploy and manage these models at scale. For example, MLOps capabilities—responsible for putting models into production and monitoring them—are only in place in about one-third of organizations. Additionally, many respondents indicated that they have not yet developed the processes needed to integrate AI into existing workflows or ensure that models can be updated, monitored for performance, and maintained over time. This indicates a significant area for improvement, as operational readiness is key to realizing the value of AI beyond proofs of concept.

Governance Readiness

The Governance Readiness dimension examines both data and AI governance. Data governance is consistently cited as a top priority in TDWI surveys, but only 20% of respondents in the AI readiness assessment reported having a solid data governance program in place. Many organizations are still in the process of building governance frameworks that address both structured and unstructured data, and over 23% still struggle with ungoverned data silos.

The emergence of AI governance adds a new layer of complexity. Less than 40% of respondents reported that they have clear policies for managing diverse data types and handling the other concerns associated with AI, such as bias and fairness. Moreover, AI model governance—tracking models, monitoring for freshness, and addressing AI ethics—remains an early-stage effort for most organizations. As AI technologies like generative AI become more prevalent, establishing strong governance frameworks will be crucial for managing risks and ensuring responsible AI use.

What are you seeing?

What are your thoughts about the results of this assessment? Are you seeing similar challenges in your organizations? What insights can you share?