I often write in analogies… how the new AI wave compares to the dotcom boom or the industrial revolution. There is something to be learned from every transformation, whether it is from the factory floor or the financial markets. When we remove the specifics of the technology advancements, it becomes an economic phenomenon or a supply chain problem.

Looking at the way software is written by AI, I cannot help but compare it to 3D-printing in the manufacturing domain. 3D-printing was supposed to revolutionise manufacturing, but as of today it is still a rounding error in the overall manufacturing sector. It is worth noting that the technology is improving at a fast pace, and the adoption is also increasing. Where it makes the biggest difference is prototyping. Watch the popular maker channels on YouTube: they use 3D-printing to prototype and validate designs before committing to more expensive materials and scaled-up production. The adoption in the final production process is also increasing, but not at the scale or pace to replace the current techniques or processes.

AI-assisted coding is also going through such a stage. Prototyping software is easy. It is instantaneous. Using popular vibe-coding platforms, one does not need extensive knowledge of software engineering or expensive tooling. It runs in the cloud, and most of the time auto-deploys. But it is still a prototype. It might break when you add different categories of users or scale to production workloads.

There are a few reasons that stand in the way of 3D-printing adoption. They include consistency in quality, overheads during post-processing, the economics of scaling, speed and throughput for scaled-up volumes, and limitations in the materials suitable for 3D-printing.

I believe there are similar limits in software. The quality of the software generated by AI varies widely based on the people who prompted it and the technical vocabulary they used. There are of course overheads after the code is generated, with increasing demand for people to clean up AI-generated code. And AI-generated code has certain recognisable patterns, even when AI is used to generate the designs, which limits the variety and creativity.

The technological advancements in 3D-printing and code-generation are real and useful. In the short term it makes sense to leverage these nascent technologies to build prototypes, familiarise ourselves with them, push boundaries and establish best practices. And in some cases, take them to production and scale up. 3D-printing promises to cut the waste of subtractive manufacturing through additive techniques, and even create tailored products on-demand. The promises of AI-assisted software engineering are quite similar. They might be on the same trajectory. But software trends do move faster than materials and manufacturing.