Completely, 100% written by a human.
The jury’s still out on the real costs & returns of AI. But some of the verdicts have finally started to come back, and you know what gang? It’s not all roses.
We’ve heard companies cite cost-savings by AI as a driver for layoffs and restructuring. The implication is that efficiency happens when AI can be used to do more things, and more quickly, with fewer people at the helm.
(Friendly reminder that AI doesn’t displace people. People displace people.)
But more recently, we’ve started to see companies dial that back. Some have rehired people they’ve laid off OR they’ve come out publicly to grapple with one of AI’s many awkward, current realities: it costs us more $$$ to use it than we thought.
It also costs environmentally, but I digress.
The consumption challenges that major enterprises are facing may feel far off. After all, many of us are only just starting to wrap our heads around AI and how to responsibly bring it to our orgs.
But given how notoriously difficult it is to qualify whose AI implementations are actually going well/ driving growth / making an impact, this latest discourse begs an important, perhaps un-fun question.
How will we know that our AI efforts are actually working?
The most severe inefficiencies often feel productive.
This is not an AI-only problem. While technology has certainly brought innovation and efficiency, at times, it’s also done a bang up job of making us feel way more productive than we are.
Think every time you tried to use a new tool to manage your to-do list. Or your projects. Or your finances. Or your household.
Now, try to remember what it felt like to land on that new system. The curiosity of discovering a new tool that seemed to have every feature you could need. That wave of superhero-dom as you locked in and made it perfect. The initial victory those first few times you put that system to work.
Then, eventually, the disappointment when you reflected on how long it’s been since you last opened it…because one day, it just didn’t work for you.
You can see how this sense of false efficiency goes beyond AI. But when we’re talking about technology that has the power to generate massive amounts of content at rapid speed, that automates workflows in ways we’ve not seen before AND is capable of superhuman intelligence… the stakes get higher.
AI can accelerate efficiency and effort. It can just as easily amplify misdirection and dysfunction.
The major difference here is that either way, AI can still feel productive. That’s because it helps users do things they simply couldn’t do before – or at least not without spending significant time or capital to figure out.
AI removes a lot of friction from historically complex processes. That is both its biggest boon and trap.
What can be done?
Over the coming weeks, I plan to share specific ideas to help you determine how well your AI pilot is working. But until then, the short answer is to remember this.
While AI is one of the most disruptive technologies in our lifetime, AI on its own doesn’t create impact. It all comes down to how you use/configure it.
We’ve already touched on the productivity angle of this. But consider the organizations using AI right this moment to do powerful, pro-social or pro-climate work. Maybe their offerings have AI bundled in. Or maybe they’re using AI internally behind-the-scenes. to deliver on their mission in more efficient ways. Many of us likely aspire to this latter bucket. Not all of us are doing it well.
And on the flip side, consider the bad actors leveraging AI to wreak havoc on all fronts: scams, system hacks, and more. (If cybersecurity professionals were ever bored before, they’re certainly stimulated now).
My point being that learning-language models (LLMs), machine learning engines, enter-any-tech-here….none of these things have a conscience, aspirations, or morality. They’re just tools, things to be wielded by us. You know, the humans 👋
We still run this show. Even if it doesn’t feel that way.
So part of this answer is almost painfully conventional for my adventurists. Treat your AI pilot like any other successful implementation project. Set clear requirements for what you want to see AI help accomplish. Create a plan driven by your goals and considerate of your constraints. Establish a process for evaluating how well your org is hitting on those desired outcomes. Acknowledge the cultural elements at play with your staff, being extra thoughtful in how you introduce and ultimately roll out AI. AKA, all the usual suspects for a technology project done right.
But the other part of the answer is painfully dynamic for my stability lovers. Approach your pilot with the knowledge that the tech, regulatory & threat landscapes continue to rapidly evolve. The AI context I’m writing about today may look very different next month, next year, or even next week. That’s par for this wild course we’re all currently on.
And if you’re really just getting started with AI, consider hitting the ground running with a few discrete experiments first (vs trying to come up with a large scale tech overhaul). They may not sound needle-moving or particularly exciting! But isolated AI experiments:
- Are a great way to build familiarity with tech you haven’t fully used, if at all
- Make it easier to hone in on specific problems/use cases – a task that can be surprisingly hard even for effective teams
- Are ideal for mitigating risk, since you set your testing confines (🎉 *celebrates*)
- Leave space to learn/make mistakes in a way that tech builds normally don’t
- Still provide an opportunity to bring real time-savings and innovation
- Won’t cost you half-a-million bucks. Plain and simple
Wrapping Up…
Let’s start asking those fun, un-fun questions about our AI pilots – and see how they’re all really going.
p.s. Is your org looking for the best way to get started with AI? Check out my free e-guide – written for orgs who have every intention of moving forward, but no clue where to start.
Leave a Reply