PSA—AI is not a thing, it’s a tool

If you’re old enough to remember when telephones came out in the business world (you’re probably not reading this article on your phone or iPad), you’ll remember how they decided who got one. It was simple, the execs got them, and then whoever they thought was important and responsible enough to get one—or argued loud enough. Then over time, everyone realized it was important for everyone to have one and it became ubiquitous. Seems kinda silly, right? Well, what happened when computers came out, then email, then cell phones . . . whether it’s history repeating itself, rhyming, or we just didn't learn—why on earth are we doing the same thing with AI? When I ask around the industry if they’re using AI—the response is basically, "we’ve been experimenting with Copilot and giving it to some execs and managers"

***sigh***

Let me just set the record permanently bent and claim that AI is not a thing—it’s another tool in your growing war chest to get stuff done. In the collective rush in the fear to adopt technology, all we talk about are the safeguards, constraints, who has access to it, and governance. They treat AI as a scary standalone invention (ok, I’m envisioning prehistoric human philosophically arguing about who gets access to fire or the Bible, but I digress…), a magic box that promises transformation and will destroy the enterprise if everyone gets to use it and there isn’t a policy in place! 

***gasp***

This framing is flawed, and silly. AI is not a “thing” to be bolted onto operations or feared (ok, it is—read Asimov’s Three Laws of Robotics and how AI routinely circumvents the laws to dominate and enslave the human race), it’s simply a capability—a set of tools, models, and systems—that must be embedded into your workflows to deliver real value. When AI is treated as part of the workflow, not as a separate entity, it becomes a trusted resource like your smartphone, Google search, or photo magic eraser, and not a separate entity that must be contained and limited.

You see, there’s a simple myth perpetuating in the industry that AI is a standalone solution. The allure of AI as a standalone solution is understandable, it’s hyped as revolutionary, capable of automating tasks, generating content, and making decisions. But when organizations deploy AI in isolation—without integrating it into their existing processes—they often fail to generate the results they expect or constrain it with human foibles. I’m looking at any of you that have implemented AI underwriting and then overturn or stip the loan decision with human underwriters! Worse yet—those of you too afraid of harnessing the relentless velocity of terabyte-scale AI computation to transform data into real-time intelligence and decision-making precision (generated by AI because I don’t have the hyperbole vocabulary to make it sound cool) because you’re afraid of the ephemeral boogeyman that will exploit your company because teller exposed a security flaw in your system because they used AI—the shame!

There’s a recent MIT report making the digital news circuit that shows despite the $30–40 billion invested in generative AI, 95% of organizations saw no return on investment. The reason? AI tools weren’t embedded into day-to-day operations. They remained siloed, disconnected from the workflows that drive business outcomes. They were treated as a separate thing.

The companies and people that will see real impact from AI aren’t chasing headlines, cutting jobs, or repeating the now trope ‘AI won’t replace jobs, people who use AI will replace jobs’ or whatever. In other breaking news, a recent report found that water is wet. Instead—start rethinking how work gets done. Invest in curiosity and orchestration—connect AI to systems for text analytics, research, forecasting, and dare I say it—decision-making!  Said simply—start designing processes where AI augments human judgment and replaces legacy system and redundancies. What would it look like if your teams don’t ask, “what can we automate?” and instead ask, “what’s breaking right now, and could AI help? What can AI do for free that I’m currently paying for? What if I go to AI first in the process?” This mindset leads to new processes and developing insight—it’s about building systems that evolve and changing the way work gets done. Remember—we’re not limited by legacy debt anymore. Why are you choosing to be?

Take some lessons from our friends in software development, AI workflows have reshaped how teams operate. Developers now talk about AIOps, LLMOps, and AgentOps—operations functions powered by AI models.  These workflows aren’t static. They adapt in real time, ingest telemetry, recognize patterns, and guide decisions. In a Forbes article example from incident management, AI doesn’t just send alerts. It provides contextual guidance—flagging relevant runbooks, historical examples, and remediation options. This reduces cognitive load and speeds up resolution. AI becomes operational, not ornamental. 

For us to operationalize AI demands new thinking. Organizations must start embedding AI at key nodes and use it as skill augmentation. It’s not about replacing roles—it’s about enhancing them. Effective AI workflows help humans accomplish through its speed, scale, and consistency.  Collaborating often with AI builds the machine’s learning capabilities and increases the users speed to market.  

Viewing AI as a “thing” leads to several fallacies and failings, delaying the promised AI golden age.

Ok, let’s address some real fears of AI that I’ve been making light of. AI hallucinations and subsequent data quality—we’ve all heard about the lawyers that use AI to build their case only to find out AI fabricated the cases that it cites. Hallucinations occur when a model—especially generative ones like large language models—produces outputs that are confidently wrong, fabricated, or misleading, despite sounding plausible. These aren't lies necessarily as AI has no intent, but instead statistical guesses based on training data that may be incomplete, biased, or outdated. So, how do you limit it? 

Actually, the whole point of this article in the first place is establishing a ‘human-in-the-loop’ workflow where AI IS A TOOL, not the final output. Essentially, you’re having people review and verify data and output, especially for GenAI outputs. They’ll do this by recognizing signs of hallucinations like an overconfident tone, fake citations, and lack of supporting data—fact check the AI. There are other things to consider and address like cybersecurity, bias in the models, financial privacy, and data protection. Frankly, those are all topics better suited to an AI expert—but at least consider this: worry all you want about AI compromising cybersecurity in your organization while the bad actors don’t have any qualms or ethics policy about AI usage and they’re actively using it against you and your members, customers, and audiences. Bias in AI output comes from the models that train it—this isn’t garbage in, garbage out—this is systematic bias in a data set that AI quickly identified that many are trying to justify and ignore. Financial privacy and data protection? Can you protect or commit to either without AI while the aforementioned bad actors are using AI to its full potential to compromise your systems, people, and users?

AI is not (yet) a sentient force, a colleague, a competitor, or a replacement. It’s a tool—like a camera, a calculator, or a video editor. Its power lies not in what it is, but in how we use it. Just as Scotch whisky is defined by three ingredients and distinctly shaped by the distillery’s character, AI is shaped by the intent, data, and design of its users. When we combine the right mix of channel expansion, audience expansion, and message amplification, AI becomes a force multiplier—not a magic wand.  When we treat it like a “thing”—a standalone solution—we risk chasing rabbits, ignoring fat tails, and falling prey to black swans. But when we treat it as a tool, we stay grounded in strategy, focused on outcomes, and aligned with our audience’s needs. Don’t meander in the maze of mediocrity and justify all the reasons to limit it’s use. Instead, embrace that AI is here to help us think better, act faster, and serve deeper. But only if we remember: it’s a tool. Not a thing. Cheers!