I debated whether I should join the fray of technologists talking about AI but, as an elder millennial, I think there’s an angle to this story that isn’t really being discussed — and it has a lot to do with money. Or rather, the tenuous balance between innovation and profit.
Right now, everyone’s talking about how AI is our future and will deeply transform the way we live and work — this has already begun and I don’t think it’s far fetched to believe that we still don’t quite yet know the ways in which this will fully change us until we are living in the change. However, it’s also true that all the large operators in this space are desperately seeking profitability more than they are the lofty ideal of artificial general intelligence. And the need for a return on investment, especially with valuations reaching the trillions for some, will ultimately create winners and losers.
I think back to the early days of the internet — at least as I knew it. When I was an adolescent, we got a personal computer in our home and a subscription to America Online would routinely tie up the telephone line (pre cell phones!). In that time is when I started learning how to code by seeing HTML side by side with the website I was creating. And it was incredibly accessible to get started; whether it was HyperMart, Geocities, or AngelFire, you could very easily spin up a website for free, no ads.
With age and hindsight being 20/20 and all, I now realize what was happening. All that was free to us wasn’t actually free, we were a generation being poked, prodded and tested. Giving it all away for free was incentivizing us to get hooked on the internet which, after the dotcom bust, certainly resulted in some learning for the winners — they had to somehow shift consumer behavior, or tap into it, to drive profitability. We were all hooked on search engines, so that became the next logical place to put ads and a whole economy was born. And then we adapted: we went from the internet not being a valid source for school papers to Wikipedia — the internet’s encyclopedia — becoming a highly authoritative source of truth.
Now thinking about where we are in 2026, we’ve again been given free tools to play with and ultimately get hooked on. Remember when everyone was posting photos of themselves in classic paintings? Congrats, you were helping train the LLMs we have today. And, to be clear, you don’t actually have to pay for any subscription to get fairly decent outputs from Claude so my prediction is get ready to wait through ads (unless you are on a paid subscription)!
Why am I writing all this now?
When you look around, it all can look and sound fairly bleak — but I think that’s always been the case, even in the early days of the internet which we look back on as the halcyon days. In those times, there were strangers talking to a minor like me in chat rooms telling me their age / sex / location which was likely completely fabricated. There were no terms of service to agree to, no privacy policy. Yes, you could be creating a website or blog about your favorite TV or, you could easily be creating disinformation. It was truly a wild west time: lots of innovation, limited regulation, and a healthy amount of fear that this could all go terribly wrong.
We are seemingly back in the wild west period where we have a handful or so of top companies driving the innovation, limited appetite to strongly regulate (to help drive the innovation vs stifle it) and fear that we are effectively ceding all our agency to robots (building the SkyNet that existed in the Terminator film franchise).
Much like the early days of the internet, I think there will be a lot of promise fulfilled but not exactly the way we imagined. I remember thinking the personal computer would let me watch my favorite TV shows on demand — which eventually happened, all for the cost of a streaming service or two (or 5…). I also remember video being so friggin janky in those early days and now we assume video content should be relatively easy to stream from devices that fit in our pockets. And there are so many ways in which the internet has democratized access in really positive ways: ensuring stories are told from frontlines around the world, providing more accessible solutions to the impaired, creating a career path for yours truly, and everything else in between.
I’d like to believe that, when the dust settles, “artificial intelligence” (the blanket term we apply to all kinds of compute-heavy disciplines from natural language processing to computer vision) will hopefully stop being over-used as a term and we can really begin to see this innovation as assistive technology that augments human potential versus destroying it. And, instead of AI taking jobs (a convenient scapegoat to blame when you’ve over-hired and underperformed as a management team), my hope is these tools will just make up the standard issue workplace technology in the same way a mobile device / “bring your own device” is now a standard in most workplaces with appropriate safeguards in place.
I say all this also hoping that the impact of this technology on communities and natural resources is mitigated. With the high cost of compute and my imaging that consumers would be reluctant to pay for multiple LLMs (unlike streaming services), I imagine model providers will be required to optimize their most frequently used models to reduce their overhead and/or ensure the cost passed along to the customer is reasonable for them to bear — as, after all, it’s all about the money!