6 lessons from IT leaders on their AI adoption journeys

An impressive 92% of organizations plan to invest or have already invested in artificial intelligence (AI). With so many organizations on their AI adoption journeys at varying levels of

AI maturity, we can learn from those who led the charge on these initiatives. From identifying use cases and integrating AI into your architecture to getting your employees on board and measuring your success, the IT leaders at these organizations have been through it all - and they have wisdom to share. 

The progression of AI adoption

In the past year, we saw a surge in AI adoption across the globe. A 2024 survey found that 72% of organizations integrate AI into at least one business function - this is a huge leap from the 55% in 2023. Still, large companies are taking the lead on AI adoption. Half of organizations with more than 5,000 employees1 use AI. As for industries, the manufacturing, information, and - perhaps surprisingly - healthcare industries are the leaders in AI adoption, while finance, insurance, and real estate have lower adoption rates. 

With this widespread AI adoption, the reality is that not all projects are successful. In fact, 70% of CIOs reported a 90% failure rate for their custom-built AI applications. But, it's not all bad news. The Boston Consulting Group found that the companies that have adopted AI early claim 1.5x higher revenue growth than other companies. In addition, 74% of enterprises using generative AI (GenAI) are seeing a return on investment. The successful projects will help you stay competitive, bolster your revenue, and advance your AI maturity. 

To help you on your AI adoption journey, I spoke to three IT executives who are early adopters of AI to gain insight into their AI journeys. 

What IT leaders have learned on their journey to AI adoption

1. Start with the problem

The best way to incorporate AI capabilities into your organization is to start with a high-value problem you're trying to solve. Rick Rioboli, EVP and CTO at Comcast Connectivity and Platform says, "Forget about AI, what is your biggest problem?" Focus on problems that, when solved, will have a dramatic impact on business. There are a variety of GenAI use cases that organizations are already exploring that you could take inspiration from. Once you've identified your problem, start thinking about what data you'll need to feed your AI model to address this problem. 

2. Embrace experimentation 

Cynthia Stoddard, SVP and CIO at Adobe, encourages her employees to get creative. Stoddard says, "We've created an innovation hub that allows employees to understand what tools and Adobe products they can use to experiment with and solve real business problems." This not only empowers employees to try new technology and create new solutions but also aids in the cultural transformation that comes with such a dramatic organizational change. 

3. Use the right data

And make sure it's quality data. Generative AI models are trained on massive amounts of data from the public internet, but they don't have current data and wouldn't have been trained on your data. To get the most value out of AI, you need to be able to pass your proprietary data to the generative AI model, which is done through retrieval augmented generation (RAG). On top of having the right data, you need to make sure that it's quality data and will give you relevant, accurate answers. Matt Minetola, CIO at Elastic, says, "Having a solid data strategy is essential. Without unified and accessible data, even the most advanced generative AI initiatives will struggle to deliver real value." 

4. Quantify impact

Once you identify your ideal outcome and confirm you have the right data, you need to continuously quantify what success looks like - from your MVP to your ideal solution. Stoddard says by keeping an eye on performance, you're able to determine if projects need to be tuned or, in some cases, dropped because you're not getting the results you were expecting. And while monitoring business impact, you should be monitoring the health and performance of your AI systems by looking at user satisfaction with the accuracy of the outputs. 

5. Avoid AI sprawl and technical debt

Organisations may be tempted to use different point solutions for different problems to try to get applications stood up quickly. Minetola warns that "the businesses that solved in pockets are starting to see the long-term cost. If they've done five to six different solutions with five to six different vendors and have to glue that together, the cost of that will be huge." The technical debt - the implied cost of the future work required to revise a project because speed was valued over long-term usability - and the data silos will make future AI endeavors a challenge. Stoddard says that all AI initiatives go through an architecture review to ensure they will fit into existing infrastructure. 

6. Implement guardrails

Governance and risk management are essential parts of your AI journey and must be prioritized. Stoddard says for AI at Adobe, the team relies on governance and examination of potential risks to "make sure it's safe, we're using the right data, and we're doing the right things for our customers." Compliance is only going to become a bigger issue across markets as more laws around AI technologies are put in place. "You're going to have multiple compliance issues if you don't understand how the data for your AI was generated," adds Minetola. 

Future-proof your AI adoption strategy

The AI adoption journey is not a race; it's a marathon. Start with a strong data foundation and a solid use case to expand from there. If you haven't started with AI yet, you haven't missed the boat. There's still time to future-proof your organization and stay competitive. You have an excellent opportunity to create an AI program that is scalable and transparent and works for your needs. Check out this webinar in partnership with Fast Company for additional insights from these CxOs to help you along your AI adoption journey.