Deschenes Group

My intent in this article is to share some of my personal experience that spans both the AI world and the role of the CIO in the hope that it will help you plan and execute a smoother AI-augmented future.

I have been leading IT for multiple companies in multiple industries for over 20 years. My DNA is in digital transformation and talent development. Between two VP-CIO positions, I worked as a delivery director in the AI software space, on the solution implementation side. I witnessed firsthand the challenges from the solution development perspective and from the customer side.

A few months before having that delivery director position, I was acting as master of ceremony for a digital transformation conference. In my opening speech, I made the point that CIOs did not need to have AI on their list of competencies but that those who did would differentiate themselves from the others. Looking forward, I would state that this is a prerequisite for long-term success!

In my role as delivery director, there were two main issues that we frequently faced with customers and that CIOs should likely have on their radar when considering AI solutions. The number one issue related to the data foundation is having to deal with a multiplicity of challenges in quality dimensions. Without good data, algorithms just won’t provide reliable value. The second most important issue we faced was related to the costs of building and implementing our AI solutions. Although there were great opportunities, the development proved challenging and took longer than planned, requiring highly talented and expensive resources. The hardware necessary to run the algorithms was very expensive, and the business expectations were often overinflated. Both sides of the economic ratio were going in the wrong direction! Each customer had its own challenges, but these came up consistently.

Now let me put on my IT leader hat. In my opinion, and by far, the most challenging task to prepare for a successful AI future relates to establishing a robust and comprehensive data foundation. To do so, tackling the mess that has built up over the years is an unavoidable and daunting task. By mess, I mean years of misaligned data history, from multiple sources, coming from multiple business acquisitions, in different technologies, format, locations, with bad and inherited habits—a mess! Therefore, organizationally, align, clean up, harmonize, and standardize! Easy!

The reality is that it is impractical and likely impossible to tackle it all at once. Therefore, consider prioritizing data sets based on use cases that align with your organisation’s strategy while leveraging quick wins (tactical and visible initiatives) to build momentum. I also strongly suggest you leverage cloud technologies to address some of your data problems. For example, use a cloud-based data lake architecture so you can manipulate mass data at will, benefit from advanced tools, and scale for speed and capacity when needed.

Then, once you have your data partly or fully under control, ensure long-term value with a governance body, a centralized or shared services master data management group, business data stewardship, and clearly defined data maintenance processes and KPIs. Ensure that as you iron out the issues, you implement the proper mechanisms and safeguards that will forever protect these newly straightened-up assets.

There is one more area to dig into before jumping to the next issue: missing data. To deliver on your target, you might be lacking some key data. You certainly can, and probably should, add new means to capture it. However, although progressing organically will be most relevant and reliable, it will also probably take a long time before you have all you need! Consider closing the gap by utilizing modern techniques such as machine learning, creating synthetic data, or utilizing complementary data sources from various public, federated groups, or commercial entities (such as Google, Microsoft, OpenAI, and many others). If you use these, however, ensure you truly understand their biases and intent, where they come from, and their various hidden aspects. That data will taint the outcome; it needs to be your reality! Alright, data, check!

The next challenge is with regard to talent. I currently work in the construction wholesale and retail industry, and although I have a remarkable team, realistically I will never be able to build, attract, and retain the competencies required to develop all of the AI algorithms and solutions we could need to succeed. So, if you are like me, in a traditional industry, attracting and funding a significant number of AI PhDs might be a challenge. Therefore, my recommendation is to invest in your data science team members so they increase their business knowledge and gain insight into your company, your data, and what is important to you and your customers. Then, complement your team with external partners—a multitude of them, but be selective and intentional. If diversity is a richness in itself, multiplicity of talent, views, and perspectives applies more than ever in the AI world.

As for costs and business cases, the situation has evolved significantly over the last little while. If you break down the pieces, you have options like never before, pushing the costs down. You have pre-built models and proven solutions; you have various sources of external talent; and you can leverage the cloud offerings with a multitude of twists and turns. It is no longer a question of whether or not there is an opportunity that is worth it. It is rather a question of which one gives you the best strategic impact and or the most return. I am not saying this is easy, but the opportunities are real! However, make sure you architect and monitor it correctly, or the costs will quickly escalate!

In the old days, I would have said and argued to take baby steps and crawl before you walk and run. Although I still believe in this to minimize risks, let me venture a question: how much time will it take you to be ready, and how much time do you still have in comparison to your competition? The gap between those who are already forging ahead with AI in a strategic way and those who are struggling to simply keep up with technologies is growing. Some are losing the race badly; some are not even realizing there is a race out there. You might not have the luxury of time anymore! So, make your time count. Prioritize and focus! Take your time learning, but also allow yourself to fail quickly so that your learning is accelerated! Focus on what will help you achieve your business strategy!

The bottom line is that you need to integrate a data science and AI chapter into your strategic plan. Where this used to be for the selected few industries, it is now mainstream and a key ingredient in the survival of the fittest race. You are now in hurry-up offense mode! What is your AI attack plan?