There are throughlines to the technologies I have gravitated to over the course of my career. It isn’t a question of fashion, it is a question of value, flexibility, or durability. Systems worth building that are capable of absorbing change are going to take considerably more thought than the digital ephemera that passes as much of software development.

It isn’t a question of over-engineering, but it is a question of engineering. This is a term we barely deserve the privilege of using in our industry given the rigidity and fragility of so much of our work product. The reality is, it takes time to meet the needs, both current and future, of the stakeholders funding the activity. We have methodologies that attempt to proceduralize the activities to accomplish this, but those too often become certification-focused self-licking ice cream cones.

We now find ourselves on the precipice of a new reality, one that threatens to toss out decades of experience Although the usurpers have chosen to train off of this experience, generally without compensation or the simple respectful nod of attribution. as no longer necessary. We are told by people who are funded by people who strive to eliminate the limitations of human frailty that we are no longer required. Anyone with even just a casual awareness of the realities of large language models knows how ludicrous this claim remains. But in the absence of a viable tech journalism cadre, we are simply fed press releases under the guise of “CEO Said a Thing Journalism” and left to make up our own minds.

And, so we must.

However, anyone who still says that there is no “there there” in the stochastic machinations of token-generating systems is also not paying attention or thinking enough. The genie is out of the bottle, and like toothpaste, it isn’t going to go back in. Yes, these technologies are over-hyped. Yes, there is a bubble. Yes, it will burst. Yes, there will be significant damage. In the aftermath of whatever this ends up looking like, however, these technologies will still be there and will still be useful under appropriate guardrails.

The negative externalities will be addressed. Power efficiency will improve even as demand soars. This will come from algorithmic, hardware, and, perhaps most inexplicably, substrate Cortical Labs is but one example of an electrical/biological hybrid with the goal of reducing the power consumption of learning systems by orders of magnitude. advancements.

As we hit the scale-driven limits of autoregressive token processors The Scaling Age is Over, we will discover the benefits of new, multi-modal systems (images, actions, concepts). It will involve the combined benefits of deductive, inductive, and abductive reasoning. Until then, we need humans in the loop.

More pressingly, we need a pipeline of up-and-coming humans in the loop. These younger generations need to still develop and absorb the concepts of business value, architecture, security, encryption, user experience, engineering, and more. API design is not less useful in the face of AI, it is more useful than ever. And the agency of agentic systems must be constrained by context-aware, Zero Trust architectures with attenuated authorization grants providing the least privilege needed to take an action.

We as a species were never meant to countenance the pace of change that is unlikely to slow any time soon. But we are not without our own agency. We are not subject to the whims of techno-utopians who undervalue humanity. We can proceed at our own pace, adopt what makes sense, and push back on what doesn’t. We cannot ignore the underpriced concerns about the social and cultural impacts of a myopic quest of pure productivity. We can benefit from the advancements, flawed and incomplete as they are, to do more with less. But we are also encouraged to contemplate that “with less” does not mean none.

Luddites are often derided by people with incomplete grasps of history as being slow-to-adopt or somehow anti-progress. On the contrary, they knew exactly what would happen and why. They weren’t against industrialization, they generally accepted it as a necessity to lower the costs of goods and services and broaden markets beyond the wealthy. Their larger concern was the resultant loss of the artisanal. Not every garment needed this attention to process and detail, but what a shame to lose the skills entirely.

The same goes for our brains. Humans that wield agentic tools will be able to accomplish wildly aspirational activities. But, if you fail to maintain your brain’s ability to read, write, and think critically, we face an even bigger challenge as a species.

My goal here is to provide 100% human-generated content in bite-sized pieces. I have nothing to sell you but ideas Most especially ideas that absorb other ideas gracefully. More on that soon.. I’m neither a shiny-object true believer nor a misguided perpetual naysayer. I am someone who believes in using my brain and my background to make decisions and evaluate the world around me.

What a shame it would be to lose those skills entirely.