AI Could Be the Next Tech Wave – But Only If We’re Careful

Before we can unlock AI’s full potential across enterprise and public sectors, there needs to be improvements in explainability, consistency, and control. During Enterprise Connect 2025 in Orlando, I had the privilege of dining with Terry Matthews and three others in what not surprisingly turned out to be a thought-provoking discussion about the dominance of the phrase “AI” not only at the show this year  but also in the marketplace generally. According to Mr. Matthews, founder of Mitel and a vast number of other successful ventures, we are “in the midst of a tech wave, and that wave is AI.” 

The first challenge, however, is that what the phrase “AI” means is incredibly broad and means different things to different people. In any case, AI is the buzzword of the moment, which is reason enough to be wary when the au courant buzzword is frequently and loosely bantered about.

Even in the case of those who think that they understand what “AI” means in a specific context, fewer still understand the subtleties of how it works in the products and systems they’re considering and securing, and how it performs the special magic that it does. Because talking about AI sounds trendy and “in the know” (and it is both), and because AI is an incredibly powerful tool, the phrase really requires context to make its use meaningful in any important and relevant way.

Relevance is another key consideration. In her wonderful book Weapons of Math Destruction, author and PhD mathematician Cathy O’Neil presents information on baseball as an “ideal home for mathematical modeling.” She goes on to reference the fact that part of what has made baseball so ripe for such modeling is its recognition of the fact that circumstances (read: algorithms) must be modified and adjusted warranted. So, it’s not sufficient to deploy an AI driven product; its output must be continually observed, managed and modified in response to circumstances to remain as beneficial as it can be.

Those who are selling AI-based solutions also have to remain current and engaged, not only to make the next sale, but to make sure that they understand how the sausage is being made in the first place, and whether it’s seasoned properly to yield the optimal results as systems and processes evolve.

The enterprise space isn’t the only one that’s relying on AI tools. Recognizing the challenges that AI systems could pose at all levels of government, during 2024, the Biden administration Executive Order 14110 was put in place to create guidelines and parameters “on AI use throughout the federal government.” While the 2024 executive order has largely been scrapped since the new administration took office, it had laid the groundwork for consideration of the impact of AI at the federal level. While the guardrails may largely be gone, the previous administration recognized the need for the management of AI tools. However, in 2025, each state has taken steps to not only craft, --but enact-- legislation regarding AI deployment use within its own jurisdiction both by public and private entities.

Very presciently, in 2019, an article appeared in the Columbia University Law Review that made this frightening claim regarding government use of AI systems: “...When challenged, many state governments have disclaimed any knowledge or ability to understand, explain or remedy problems created by AI systems that they have procured from third parties. The general position has been ‘we cannot be responsible for something we don’t understand.’  This means that algorithmic systems are contributing to the process of government decision-making without any mechanisms of accountability or liability.” 

Litigation over the use and misuse of AI has begun, with the now infamous cases of both Avianca Airlines and Michael Cohen, both of whom used Chat GPT to prepare legal documents. But the cases that will rock AI deployment in more significant and costly ways will be those that come up when humans make bad decisions based on the outputs from their AI tools. When these cases begin rolling through the court systems in the U.S., Canada, and throughout the world, defendants will be forced to reckon with the amount of clear understanding each possessed at the time of acquisition and throughout the term of its use.

In the wake of March madness, it is  important to note that while the ability of AI systems to crunch numbers is unsurpassed, it is` still often wrong. Look the 64 teams in the men’s and women’s brackets and how AI tools still couldn’t accurately predict outcomes based on vast quantities of player and team stats – and examples of that (lack of effectiveness) reminds us all why it remains critical for those deploying AI to remember that the technology does not have all the answers, regardless of how prominent a role it plays in creating outputs used to make important decisions. 

I’m not suggesting that AI isn’t a valuable tool. It certainly is. But what remains clear is that even if it is managed and massaged on a regular basis, AI outputs should remain only one of many tools in the toolbox.

Mr. Matthews's comments to me about the current tech wave of AI were, of course, pragmatic and forward-thinking as to the potential of AI.  He said that AI has the potential to trigger a technology wave for enterprises in an even more significant way than has the Internet — and we’re just at the very beginning! For that wave to be truly transformative and sustainable, key challenges must be addressed. Ethical consensus will be critical as AI moves toward general intelligence, ensuring these systems align with human values and do not cause harm. At the same time, advancements in privacy-preserving technologies, smaller personalized AI models, and secure communication protocols between AIs will enable more scalable and decentralized adoption, particularly in edge computing environments. More and more enterprises and government departments want to have data processing occur closer to the source, such as on devices or local networks, rather than centralized cloud servers.

Equally important is making AI more trustworthy and consistent. Today’s models are probabilistic and non-deterministic (sometimes even providing hallucinations) — which can make them unreliable for mission-critical applications. To unlock AI’s full potential across enterprise and public sectors, there needs to be improvements in explainability, consistency, and control. If these foundational issues are addressed, AI will not just scale — but it will grow globally in a healthy, trusted, and universally adopted way, impacting every industry more deeply than the Internet ever could. Only then will it be an example of the greatest tech wave we have yet to experience.

Originally published in No Jitter April 7, 2025