At a hackathon in 2023, I worked alongside a rising college Junior majoring in Computer Science who was worried that he should switch majors because AI would eliminate software development jobs. At the time, I brushed off his concerns, as I expected most of the usefulness to be in written language, not code. However, an August 2025 study of on how AI is replacing entry-level software jobs has a lot of folks in Silicon Valley spooked – and new tools like Base44 and Lovable promise end-to-end AI-generated coding solutions, no coding skills required.
It’s easy to see the rise of coding tools over the last few years and draw a line to a future where all coding is automated. But technology adoption is not linear, and I think the future is more nuanced- and we’ve seen this before. I want to spend a bit of time looking at two technologies which embody the promises -and far more mundane changes- of how promising technologies integrate into our lives: synthetic food, and 3-d printing.
Synthetic food or meal-in-a-pill imagery has been a staple of futurism since the advent of agricultural science and modern nutrition, since at least 1893. As nutritional science and food engineering has improved, it’s possible to live on Soylent or a carefully-tuned ketogenic meal plan- but most people don’t choose to do so, and instead reach for the same relatively reliable set of staples. Food engineering is a huge industry- but it’s largely in the background, disguised as something familiar and comforting.
3-D printing is a more recent disruptive technology, which in the late 2000s and early 2010s promised to replace all manufacturing with on-demand components and 3-D print everything in our lives from clothing to houses, leading to huge valuations for 3D-printing companies. Mixing the two technologies we’re exploring, there even were folks who expected that we would soon be 3-D printing food in our kitchens, delivering just-in-time tailored nutrition in the most scientific way possible. Everything was about to change.
Ten years after we were promised a 3-D printing revolution, it remains a specialized field: replacing some manufacturing processes, used by specialized professionals and avid hobbyists, but not migrating into most homes or businesses in a visible way.
In both these cases, I’d argue that the barrier to adoption isn’t a failure of the technology -we have proven that we can print houses and live off Soylent- but rather a cultural decision not to do so, based on our values. I suspect that there are three main values driving this:
- Risk aversion: Changing to a new style of eating or living takes effort, with an uncertain payout. Even if the new technology might be expected to yield a slightly better life, the risk of a less-beneficial outcome might outweigh the benefits. Sure, eating Soylent might be equivalent to your current nutrition, but why change to something equivalent if there short-term ick risk and the long-term health uncertainties?
- Switching costs / Status Quo bias: There are real economic costs to changing our behaviors, whether that’s in building new expertise or just changing our grocery list. 3D printing remains a specialist tool because of the tech effort required to set up, debug, and maintain a printing pipeline. Where an upstream vendor (e.g. car or plane manufacturer) sees the benefit of changing to 3-D printing without us needing to do the effort, we will be happy to let them make that switch, so long as it doesn’t take effort from us.
- Trust: There are a lot of scammers and partly-functional products on the internet, yet we turn to trusted brands for many of our purchases. Structures like ISO/UL certification, indemnification, feedback networks, and guarantees/warantees underscore the importance of that trust. A small number of engineers, hackers, and DIYers are willing to debug failing tech themselves or write a patch- for everyone else, it’s immensely important to be able to call up someone for tech support or ask for a refund when things don’t work out.
In an AI Maximalist’s vision of the future, there is no more need for software engineers: we simply ask AI agents to create custom software for us as-needed to perform tasks, the software lives for as long as it’s needed, and then the software is torn down. I’d argue that this is remarkably similar to the initial future imagined by 3D printers: print the custom part you need at the moment, print it again as needed, and chip it up to reuse the material when you’re done.
Because of individual and institutional risk aversion, I would argue that we’re unlikely to see widespread adoption of one-off bespoke applications created by AI coding tools (like Base44 or Lovable): while technophiles might find it appealing to have exactly the feature set they want, most people will want to have the guarantees of robustness, security, and on-call support which come with a fully supported company or ecosystem.
The catch here is that we likely will buy services which are generated by AI agents, but only if they’re backed up by the same guarantees offered by a conventional company: in the same way we don’t care whether a vendor uses 3D printed parts as long as they guarantee its performance with their reputation, we similarly won’t care whether the software is generated by human hands or AI as long as we have similar guarantees about its robustness.
This means that companies and startups which can use AI to augment, extend, or build features while still benefiting from customer trust and existing brand relationships can overcome the three hurdles above to deploying new software products, regardless of how they structure their teams of developers and AI agents.
This gets back to a core maxim of product management: deliver a great user experience first, and worry about how it’s implemented later. Whether it’s coded by humans, AI, or monkeys throwing darts, if you can build a great user experience and robust company brand around it, it might succeed.