By now, if you work in tech, you have probably sat through at least one meeting where someone said the phrase “leveraging AI to drive operational efficiencies” with a completely straight face. Maybe you said it yourself. I have. And I mean it when I say it, which puts me in an interesting position.
Because, I both work with AI, and dislike all the things about AI that so many others do.
And I also think a lot of what is currently happening with AI is a bubble. A big, loud, well-funded, aggressively marketed bubble. And I think we all know it, even the people inflating it.
Here’s the thing though…I’ve thought a lot about what happens when it pops.
The Bubble Always Pops
Every transformative technology in modern history has gone through this cycle. The internet in the late 90’s. Social media in the early 2010’s. Crypto, NFTs, the metaverse…each one had its moment of “this changes EVERYTHING”, its period of insane investment and insaner expectations, and then its reckoning. The crash where the grift burned off, the grifters lost interest and what was left behind was the actual useful thing underneath all the noise.
The internet didn’t disappear when the dot-com bubble burst. It became something more honest. Something more useful. Something that actually earned its place in daily life instead of just promising to.
AI will do the same thing.
When the dust settles, when the “make me rich with faceless reels” and “show me how it would look if Rihanna and I had a baby” crowd moves on to the next shiny object, when corporations stop putting “AI-enabled” in their product descriptions the way they put “synergy” and “paradigm shift” in their mission statements in 2004…what will be left is a genuinely powerful tool that, in the right hands, does remarkable things.
Just not magic things.
Enter: The Ethical Hacker Model
I’ve been thinking about this through the lens of a term borrowed from cybersecurity: “ethical hacking.” The idea that the same skills and knowledge that could be used to break into systems can be redirected, intentionally and consciously, toward making those systems stronger.
Ethical AI use, to me, works the same way.
It starts with being honest about what AI actually is and isn’t. It isn’t creative. It doesn’t think. It doesn’t have taste or judgment or genuine understanding. What it has is an extraordinary capacity for pattern recognition, data organization, and the kind of relentless grunt work that used to eat hours of a skilled person’s day. It is very, very good at taking a mountain of information and turning it into something navigable.
That’s not nothing. That’s actually quite a lot.
But here’s where the ethical part comes in. You have to know what you’re handing it, know what you’re getting back, and be willing to do the work that AI genuinely cannot do. The judgment. The taste. The craft. The accountability. The part where a human being looks at the output and says, “No. That’s not quite right. Let me fix it.”
You also have to be willing to draw the line where it needs to be drawn. To not steal from others in your quest for efficiency…something that can be difficult with a system that itself is already built on so much that was stolen. You have to choose not to steal art from creators, to not burn resources to create stupid things, to use this remarkable tool to make things BETTER. To see the machine for a machine and the human for a human.
A hammer is a remarkable tool. But, it has not replaced the carpenter.
What This Looks Like In Practice
I use AI regularly, both professionally and in my own creative work. And I have very specific rules for myself around it.
In my professional life, AI helps me analyze patterns across large codebases, draft documentation that I then rewrite in my own voice, and surfaces options I might not have considered. What it does not do is make decisions. It does not write final code that ships without my team reviewing it with actual human eyes. It does not replace the experience of a senior developer who has seen this particular kind of problem fail in production many times before.
In my creative work, specifically in the Sister Legends project, I use AI to help me organize and collate research that would take days to sort manually. In the first season, I used it to generate images styled after 19th century natural history illustrations, specifically choosing long-dead artists as my stylistic reference points because I have serious ethical objections to AI trained on the work of living artists who are still trying to eat. But starting with Season 2, I’m going further. The images are now built from open source and Creative Commons photographs and artwork that I manipulate myself, with AI only filling in the gaps where real world images simply don’t exist for what I need. The voiceover is now my own voice. The sound design is built from publicly available and Creative Commons audio files. AI is still in there…but its footprint gets smaller every season, because my own skills are better at filling the space it used to occupy. That’s not a rejection of the tool. That’s exactly how a tool is supposed to work.
AI is in there. But it’s a tool. It has a job. It does that job and then it waits.
The Future I’m Actually Betting On
Here is my honest prediction. The bubble is going to burst, possibly in 2027. The AI companies cannot continue to throw money into a hole with no return on that investment. They are not currently making any sort of profit, and from some reports are spending the equivalent of $4 to get paid $1. Its simply not sustainable.
Add to this the issues with the environment, the political fallout around data centers and much more, and the current course of events cannot continue.
So the bubble is stretched to its seams and about to burst.
Five years from now, maybe less, AI in the workplace will be as unremarkable as Excel. Nobody today has a meeting about “leveraging Excel spreadsheet technology to optimize operational efficiencies.” Excel is just there. It’s part of the toolkit. You either know how to use it well or you don’t, and the people who know how to use it well have an edge over the people who don’t.
AI will be the same. The developers and designers and writers who figure out how to use it with intention, with judgment, with a clear sense of what they’re responsible for and what the tool is responsible for…those people will be more effective than those who don’t. And the people who tried to get rich off it without putting any real skill into the equation will have moved on to whatever comes next.
The bubble will burst. The hype will evaporate. And what will be left is a tool sitting in the toolbox next to VS Code and Photoshop and a terminal window, waiting to be used by someone who actually knows what they’re doing.
I’d rather be ready for that version of the future than still arguing about whether the bubble is real.