CIO PRIORITIES '25

Introduction

Analyst Perspective

Time to prioritize like it's 1999

Have you heard? The ’90s are cool again. Get cozy in your athleisure wear and turn up the Oasis music. For elder millennials like me, it's a welcome cultural moment. So why not look to a management trend from the same era for some inspiration on how to navigate modern-day challenges for CIOs? In January 1999, MIT Sloan Management Review published a research feature on the chief knowledge officer.

The role was distinct from that of chief information officer, occupied by the IT function, and specifically focused on creating systems of knowledge management. That included creating, protecting, and distributing knowledge across the organization (Earl and Scott, “What Is a Chief Knowledge Officer?”). Over time, the capability of knowledge management became diffused across the organization and became integrated with the responsibilities of other leaders. Today, it's not a common title to come across on LinkedIn (my search retrieved 14,000 results compared to 443,000 CIOs). But have we truly mastered knowledge capabilities in our organizations?

We speak of making data-driven decisions and focusing on pooling our data together, freeing it from different silos. Data is only the most basic foundation of knowledge. Analyzing it produces information, which implies record-keeping of what's occurred. Studying that information leads to knowledge. When you can gather and systemize that knowledge to effectively prepare for the future, you've developed wisdom.

As CIOs enter the age of AI, they must look at the value they can create in organizations that want to harness its capabilities. To effectively use it to augment and automate cognitive tasks, we need to build from the bottom up. We know the foundation of AI lies in data. Aggregating it and making it accessible will fuel pattern identification and surface information. That information will aid in decision-making and give meaning to our work. But we can't stop there. CIOs must now seek to identify and surface patterns in information to identify best practices and create learning systems that will produce wisdom.

Analyst Perspective

What does a chief knowledge officer do in 2025?

Knowledge management may have been distributed across the organization. But has it become siloed as a result? CIOs that have spent much effort digging data out of silos to effectively pursue digital transformation know well how that fragmentation slows and limits organizations. To continue our journey with AI, we need to make sure that knowledge isn't fragmented in the same way. But succeeding will challenge CIOs to broaden their purview beyond information and technology.

Success will depend on continuous adaptation of organizational structure, evolution of its people and their talents, and synthesis of feedback from the external environment. As we foster that virtuous cycle on our path to wisdom, we must also be aware of what threatens to destroy knowledge. From protection against data loss and corruption, to filtering out the signal from the noise in our information, to pushing past misinformation and stagnant culture – knowledge assurance involves people, process, and technology working together.

So embrace the reunion tours from the bands of your youth. Dig your old wardrobe out of storage and hope it still fits. And think about what a chief knowledge officer might look to accomplish in 2025. Much has changed over three decades. The past still holds seeds of wisdom, but we must also recognize the new context of the present day.

This time, it won't be enough to merely manage knowledge. We must cultivate it.

Brian Jackson

Principal Research Director,
Special Projects,
Info-Tech Research Group

Methodology

CIO Priorities 2025 is built on the insights of primary research including Info-Tech's Future of IT and IT Talent Trends surveys, its diagnostic benchmark reports, and in-depth interviews with IT executives. Our survey and diagnostic data provides answers to the questions of “what” organizations are going to do and “when.” The in-depth interview process adds the "why" and the "how" with deeper insights on execution.

The Future of IT 2025 Survey was conducted between May and June of 2024. The online survey received 970 responses from IT decision-makers across a broad range of industries and regions, with a focus on North America. Almost six out of 10 respondents hold director-level seniority or higher.

IT TALENT TRENDS SURVEY

Info-Tech’s IT Talent Trends 2025 Survey collected responses from April to May 2024 with Centiment. The online survey analyzed 461 responses from IT professionals with various specializations from organizations of different sizes and industries.

INFO-TECH DIAGNOSTICS

Diagnostic benchmark reports including CEO-CIO Alignment, CIO Business Vision, End User Satisfaction, IT Staffing Assessment, and the IT Management & Governance Diagnostic, reflect Info-Tech member results for the period of August 1, 2023 to July 31, 2024. Refer to source notes for sample size information on specific metrics.

PRIORITIES INTERVIEWS

In-depth interviews were conducted with IT leaders between August and October, 2024, to collect insights on priority-making and agenda-setting for 2025. In total 25 interviews and written submissions were completed to contribute to the CIO Priorities 2025 report. Unless otherwise noted, the "Opportunities," "Risks," and "Examples" sections of each priority are sourced from these interviews.

Moving from trends to priorities

Understand the CIO priorities by analyzing both how CIOs respond to trends in general and how specific CIOs responded in the context of their organization

CIO PRIORITIES 2025

01

Distribute Data & AI Access

02

Develop a Future-Proof Workforce

03

Extend Identity Assurance With Zero Trust Security

04

Proactively Mitigate the Risks of Emerging Technology

05

Build Exponential Product Teams to Realize AI Value

PRIORITY 01

DISTRIBUTE DATA & AI ACCESS

Broadly expand the availability of data and democratize AI capabilities.

• Data Governance
• AI Strategy
• Data Insights & Analytics

AI may be new, but the business demands it

Most organizations only started investing in AI in the past two years but must quickly move from experimentation to transformation.

AI is different from many technology investments made because it holds transformative potential. Rather than touching only on one part of an organization, or making a broader but incremental impact, AI's promise is both broad and significant, and moving from investing in AI to squeezing the best value out of it is an important journey.

It's one that many organizations will struggle with. In the same way that digital transformation efforts were slowed by technical debt and different data silos in traditional organizations, now AI efforts are hindered. IT leaders say they feel that their teams lack the data management skills needed to set the foundation for AI. They also worry about how to govern its use and recognize their data platform isn’t optimized for AI.

The more knowledge workers are provided with regarding AI tools, the more they can tailor it to their workflows and improve productivity. But before AI can be democratized, it must be governed. And before it's governed, it must be useful.

To avoid errors in generative AI output and provide useful context, large language models (LLMs) benefit from retrieval augmented generation (RAG) in which additional data is pre-loaded into the memory of the model. This is a common method to customize AI to an organization's context, although additional pre-training and fine turning are also options.

Any of these approaches requires good data quality. Feed a LLM a biased data set, get biased answers. Provide conflicting answers from many different data sets, get different answers. The bad news is that most organizations struggle with data quality.

Info-Tech's CIO Business Vision diagnostic benchmark for the past year shows that data quality suffers the largest gap between importance and stakeholder satisfaction out of all core services. The related analytical capability is the third-largest gap.

The business wants access to new AI tools that will help them work. Yet many organizations feel they aren't ready to fully realize the value of AI because of challenges with their data. So there's work to be done.

Top three challenges to adopting AI

1. Lack of AI or data management skills
2. AI governance
3. Data platform not optimized for AI

Future of IT Survey 2025, 2024; n=326

Opportunities

DEMOCRATIZE INNOVATION

Opportunity: Position AI as a strategic business enabler

Tactic: Enable every employee with AI

Using AI goes beyond typing a prompt into ChatGPT and pasting the output into another document. AI tools help knowledge workers think, organize vast amounts of information, review documents, and create high-quality deliverables faster. AI tooling can allow users to quickly create mini-apps to automate or augment tasks they do every day, without writing code. More formally, innovation can benefit from AI-augmented rapid prototyping and ideation. IT leaders are pursuing this vision through an enablement approach, working with business domain experts to understand when and where AI can provide value. With this approach, IT teaches what AI is capable of and how to use it, then steps back to let the business users execute.

DRIVE PRODUCTIVITY

Opportunity: Deliver many small efficiency gains that add up to significant savings

Tactic: Deliver AI solutions that can scale across the organization

Text-based agents (or chatbots) are a popular application for using LLMs because they can be applied to so many different types of task. A researcher may analyze survey results and identify themes or sentiment. A marketer may turn a long, dry, corporate positioning statement into an eye-catching social media post, and so on. CIOs successful with deploying AI to production are often identifying use cases that can be easily adapted across the organization. Paired with training, an AI application that broadly helps in many different areas can also help workers add up productivity gains.

AI-DRIVEN BUSINESS MODEL TRANSFORMATION

Opportunity: Reimagine the business for future success in a marketplace where AI drives competitive advantage

Tactic: Modernize infrastructure and foster a data-centric culture

As AI shifts from emerging technology into a transformative one that will touch on many different aspects of business, CIOs recognize the need to be ready. Even CIOs not investing in AI directly yet speak about modernizing their infrastructure to deliver more computer power, data storage, and data accessibility. Most are pursuing it through a hybrid cloud strategy that emphasizes growth in cloud computing while operating an on-premises data center for applications requiring data to be closer to the user. The focus is on moving to AI adoption and scaling. At the same time, CIOs are leading on data-centric culture efforts, positioning data as a valuable asset that's to be managed effectively. Leaders are building the infrastructure to support that culture through data lakes and warehouses. The vision is a self-service model where users can access data products from a centralized marketplace.

AMPLIFIERS

What organizational context makes this a higher priority?
A culture that values data and experimentation; established data governance; hybrid cloud infrastructure; Agile methodology; cross-functional collaboration

Risks

AI COULD CREATE NEW SILOS ON TOP OF OLD ONES

Risk: LLM fragmentation

Mitigation: Centralized AI governance and strategy

IT leaders can direct a centralized AI strategy that avoids further fragmentation of organization systems. Data silos that were created in legacy organizations deployed point solutions to manage data by function. Only in hindsight did it become apparent this would create different versions of the truth and cause problems with intra-departmental collaboration. Now organizations must avoid AI silos that create many different models that serve as reflections of only one part of the whole. Instead of allowing different departments to deploy point solutions independently, CIOs should apply a systems-thinking approach and provide a path to a holistic strategy that unites all stakeholders onto a single platform. We want to distribute the use of AI, but not its governance.

BEYOND COMPLIANCE: TRANSLATING RESPONSIBLE AI INTO ACTION

Risk: New AI regulations will catch the organization unprepared/AI-related security and privacy threats harm the organization

Mitigation: Execute on responsible AI

Working with AI models means there is risk of bias or errors in the outputs. Depending on how sensitive the use case, organizations will need to mitigate this risk through various means. Europe's AI Act specifies high-risk systems are ones that affect the health and safety of people. Similar legislation is in the works in Canada, and California has recently passed AI regulation. While it could take some time before enforcement begins, no organization wants to be the first to receive a fine as an example. Beyond compliance considerations, organizations can still be harmed financially or reputationally from negative AI outcomes. To avoid this, organizations will need to go beyond setting responsible AI principles and put them into practice. CIOs should build a holistic operational framework that governs data inputs and outputs, integrations with applications, and the LLM layer itself.

LEAKY DATA SYNDROME

Risk: Data perimeter becomes near impossible to maintain due to expanded attack surface and the proliferation of LLMs

Mitigation: Clear policies and robust data governance

The number of LLM-powered tools available on the web have skyrocketed over the past couple of years. Employees can access AI tools to help them with their work often for free, or with an affordable monthly payment. Beyond that, different business divisions will own contracts with vendors providing LLM-powered features. This creates risk of turning any organization into a leaky data sieve – through exposure of data to train publicly available models to bad actors looking to harvest corporate secrets from organization-trained LLMs that are external facing. CIOs are trying to steer their users away from the dark side of shadow AI through a combination of policy, training, and controls. Policy can set the standard for when and how data can be provided to LLMs, how training can educate and inform of the risks and guide toward trusted avenues, and how controls provide a hard defense when these methods fail. All three approaches will be required to guide users into the light.

DAMPENERS

What organizational context makes this a lower priority?
Fragmented data stores; poor data quality; poor data literacy; inadequate infrastructure; lack of integration; concern for compliance obligations; concern for security and privacy