When my wife was doing her undergrad her friends would meet up in the cafeteria regularly for ice cream and probably other food. They would always sit at a round table. If someone walked up to the table asking if they could sit there, the response was the same regardless of the number of people already there:
“There’s always room for one more person at a round table!”
I love this. There is of course a point at which it becomes a bit less practical, but I appreciate the mentality that there’s always room, that a person doesn’t have to leave for a person to join. There’s always room!
The tech table
As a millennial, I grew up in the prolific world of technology. My dad was an engineer, and so I’ve always had a computer in my house. I’ve been tinkering with them since I was a child. This exposed me to the introduction of the GUI-based operating systems, the Internet, broadband, smart phones, and so forth.
There was a mentality that crept in with this which I wouldn’t recognize until many years later: I assumed the progression of technology was linear and exclusive. That is, when DVDs came out I assumed this meant the end of VHS, when BluRay came out that was the end of DVDs, and when streaming came out that was the end of BluRay. To some extent there’s truth to this, but watching how Gen Z and Gen Alpha interact with technology has challenged my thinking.
Generations Z and Alpha grew up with this sort of technology already in place. The Internet existed, streaming existed, smart phones existed, within their conscious memories. To a degree, some of the proliferation has plateaued. Streaming is by no means new at this point, but nothing has really come out to challenge it. Instead, they did something unexpected: they decided vinyl and CD’s are amazing! Consider these interesting statistics:
- Vinyl sales have been increasing year over year for 19 years, and holds at ~9% market share.
- Physical books maintained a ~72% market share, with audio books (not e-books) overtaking as second at a 12% market share
- For video entertainment, streaming dominated at ~92% market share, but physical media (specifically 4k) grew last year (up 12%).
Preference plays a huge role in how people select technology. Vinyl, for example, is something a person can physically own. It’s satisfying to hold, comes in big, cool looking covers, is fun to put the needle on and flip, and has a distinctive analog sound. There’s an experience to vinyl that digital can’t entirely replace, especially streaming. Many people have both a subscription to a streaming service and own vinyl records.
Technology really does act more like a round table, or a buffet to select from, than a rotating door of progress. For the folks that build technology, we often like to fancy that their tech is better in every way, which is varying degrees of true. But more often trade-offs are made in progress, and these trade-offs leave room for preference. Recently, I’ve been thinking about the latest guest to the table: AI.
AI joins the table
AI joined the table much like the Internet did. It rolled up with sunglasses and smirking dimples of pure confidence and a silver tongue, eager to please and revolutionize every industry that is willing to talk to it. But it’s important to realize that AI is just another technology at the table. It will change things, but it’s not going to replace everything.
I’ll use software engineering as an example of this, in part because I know it so well after decades of doing it, and because it’s been one of the most impacted industries by AI. I was somewhat of an early adopter in utilizing AI within my coding practices. I started with Copilot which was like autocomplete on steroids. I later switched to Cursor and then Claude Code, which introduced the age of agentic coding. As such, I haven’t hand coded hardly anything in over a year.
This begs an uncomfortable question: Why haven’t I lost my job? Can’t product managers just assemble an army of coding agents and build out their hopes and dreams in a couple hours? The nuanced answer is “no”. I think it’s important to view technology through the lens of tooling.
Technological tools
Tools make things possible and faster. A contractor can build a house faster with power tools than manual tools, a software engineer can produce better code faster with IDEs and Continuous Integration (CI), and recorded music makes it possible to hear your favorite band without going to a venue. This view is not novel and is called the “Instrumental View of Technology”. The idea is that technology is a neutral thing that relies on humans to bring value, because we define value. One might think putting up a building is always valuable, but abandoned cities remind us this isn’t true.
AI is one such tool. What challenges this perception is how effectively AI simulates human language. We’re wired to view human language as the most vivid marker of human intelligence and reasoning. So when a dog pushes buttons that say words, a bird repeats phrases it heard, or a gorilla uses sign language, this strikes as something deeply profound. How much more when you have a machine that can speak in full paragraphs and even write books!
But even AI and the most impressive of agents fail to comprehend and create value in and of itself. I’m going to walk up to this philosophical line and stop for now with the intention of writing more on this later. But suffice to say that my perspective is that AI still relies fundamentally on people to create anything that is truly of value — whether that’s code, software, jokes, poetry, images, or videos. Even if it makes something amazing that feels to disprove this, think about the role you played as a person in both prompting the AI and identifying the result as valuable.
Back to software engineering
Let’s apply this again to software engineering. While I haven’t hand written code in months, I remain the central figure determining what the AI is to create, whether what it created is valuable, secure, and well architected, and connect it to our product objectives. It’s pretty cool when we can connect background agents to our product and project management tools to expedite processes, but I remain the fundamental engineer. AI hasn’t eliminated software engineering; it’s commoditized writing code — and coding is only one part of being a software engineer.
There has been impact in the industry, especially with Junior developers. Now, I have thoughts here. AI, in this moment, is extraordinarily powerful in the hands of those of us who have been engineering for years and decades. The business, product, project, and architectural perspectives we’ve honed are expedited by the agentic tools. But engineers are still important, and therefore there still needs to be a next generation of software engineers. My personal take is not that junior engineering is “dead” (I really don’t like that hyperbolic, clickbait term being thrown around), but that it’s going to change. The way I became a software engineer isn’t going to be the same path the next generation will. There will be trade-offs. They’ll be able to produce things much faster than I did for a long time, but they’ll also run into the issues of poorly engineered work and the trials of selling products faster, too. AI can write code, but it’s also the infinitely patient coding teacher I never had. I think the path for future junior devs will be AI mastery and building domain knowledge (e.g. architecture, security, and product) as a priority.
Conclusion?
AI has joined the table. Some might not like that but it has. It hasn’t, and I don’t think will, kick out everyone else at the table. It’s going to change things, but there will still be technologies and ways of doing things that will remain valuable. We’ll continue to find the fundamental limitations and trade-offs of LLMs that will better define its place at the table, and we may even see an uptick in things that were suppressed as we learned this.
As I conclude this, I want to leave an intentional question mark. I think the field of AI should humble even the savviest of nerds. It’s too easy to conflate what AI is actually capable of with our hopes and dreams of what we want it to be capable of. My main point of evidence in my reflections here are the patterns we see in history as other, even radical, technologies have entered the world. People are extraordinarily pragmatic, and will at least subconsciously identify the limitations in things and their impact. Even now we’re seeing that social media can’t wholly deliver on the promise of human relationship. It makes an effective tool but a miserable, and even damaging, substitute. We will see what AI ultimately does, but I think history can give us reasonable methods of prediction.
