Figuratively Speaking

Speaking effectively to non-technical people can be a challenge for technical folks, but it’s an essential task for all but the most mundane (read: least-effective) of roles. One mechanism that I’ve found helpful is the use of metaphor. I’m a huge fan of trying to describe complex topics by mapping them to more broadly understood concepts. Being able to come up with such mappings fluently is a powerful skill. There may be many ways to develop it, but I suspect one is cultivating a wide set of interests.

While I was writing Tuesday’s post, it occurred to me that today’s Generative AI tools are to software what today’s 3D printers are to physical objects. On one hand, it’s incredible to be able to provide a specification and have it manifested in near real-time. Printers can make a variety of solids: toys, some kinds of replacement parts, that sort of thing. GenAI can create chunks of useful code, quick user interfaces, and basic apps, like my Pinochle scoresheet. But there are limits. Can either of these tools produce high tolerance, precision parts / highly secure, performant code? Can they build complex solutions like electronics / web browsers?

I could be wrong, but just like we’re a long ways from 3D printing an iPhone, we seem a ways away from vibe coding Microsoft Word or an entire government system of record.