Are you feeling the vibe? I’m not talking about the atmosphere down at the restaurant that’s too busy to take your reservation—I’m talking about the vibe that seems like it’s putting programmers out of business faster than a return to the abacus. Vibe Programming is a real thing; whether it’s a really good thing remains a very active question.
Promises, Promises: Vibe coding is the current term for using a well-crafted prompt to get an large language model (LLM) AI engine to write software for you. It seems to finally fulfill the software promise launched with COBOL: Simply tell the computer what you want it to do in plain language and your algorithmic wish will be granted. No more learning complicated programming languages, no more late nights counting semi-colons, no more taking the time to set up the scripts to link, load, and compile code—just plain, ordinary words typed into a prompt screen.
It’s a vision that many employers love. They love it so much, in fact, that in some places it’s upending the job market for entry-level programmers with fresh computer science degrees. I know developers at a Very Large Software Company™ who use AI help in their daily work, getting the AI engine’s suggestion of code to solve a particular problem. They’re highly skilled professionals who are working to improve their productivity, not managers trying to avoid hiring newly minted coders. I think the seasoned folks may be onto something.
AI (of any sort) can be seen as one of two broad things: An assistant for humans or a replacement for humans. In the enterprise IT world, early returns indicate that the current generation of AI is reasonably useful as the first, and an over-hyped disappointment as the second.
Catching the Vibe: So, how should embedded systems programmers be using AI? In some of the tests I’ve run, I’ve found that it tends to deliver acceptable results for very simple tasks. For more complex programs, it delivers beautifully structured code that might or might not actually work. In a number of well-known examples, AI has tended to make up libraries to solve problems rather than using real code. While I’ll admit that I would probably be a more productive coder if I could just make up a library when I need a function, it’s a reminder that humans still have to check AI’s work.
As an assistant, though, AI isn’t bad. If I think of it as an enhanced version of the IDE that gives me a framework in which to write code, AI helps me be more productive, getting me started faster than I tend to be when I’m just looking at a blank editing screen. I expect AI to get better and better at this kind of assistance, providing more complex code fragments that I can put together to make complete solutions.
The danger to our industry is that cost-cutting administrators will decide that AI is “good enough” for tasks that are beyond its capabilities, creating difficult quality control tasks for developers while cutting off the path for new programmers to enter the workforce. AI can be a great tool—that’s how it should be seen.
This Month’s Questions: So, are you using AI in your embedded software development? Do you see it as a tool that’s part of an IDE or as a fellow developer delivering complete code? Oh, and one other thing: Do you use AI in other aspects of your work? I’d like to know how you see LLM AI in general.
Your answers to my earlier questions have been great—please keep them coming. And let me know how we’re doing here at Circuit Cellar; we have a lot of great things planned for the future and would love to have your input be part of them.
email: c.franklin@circuitcellar.com
LinkedIn: https://www.linkedin.com/in/curtisfranklin/
Facebook: https://www.facebook.com/curtis.f.franklin
Instagram: https://www.instagram.com/curt_franklin/
Mastodon: https://mastodon.sdf.org/@Kg4gwa
Bluesky: https://bsky.app/kg4gwa.bsky.social
Issue Table of Contents can be found here,
as articles are made available online they will be linked.
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • October #423 – Get a PDF of the issue
Curtis Franklin has been a journalist working in the computer and technology fields for more than forty years. From his early career as a columnist at Computer Shopper and the founder of the BYTE Testing Lab, he has covered computing devices from handheld to supercomputing and applications from trivial to life-altering. In 1988, he was the first editor of an exciting startup publication that was then called Circuit Cellar INK. Since then, he has edited and written for publications including ComputerWorld, NetworkWorld, InfoWorld, InformationWeek, and Dark Reading. Most recently, he was Principal Analyst for Cybersecurity Management at Omdia.
Curtis co-wrote one of the first books on podcasting and has been a host or co-host on more than 500 episodes of various podcasts, including hundreds of episodes of This Week in Enterprise Technology, a production of the TWiT Podcast Network.
When not telling stories of computers and the people who make them, Curtis is an amateur radio operator (KG4GWA), an artist, and a Florida Master Naturalist. He’s also active in the maker community, working on the teams that produce Maker Faire Orlando and Maker Faire Miami.
