Notes on the future of knowledge work, artificial intelligence and thoughts along the way.
Arthur C. Clarke wrote that "Any sufficiently advanced technology is indistinguishable from magic." If software development is magic, it's working on a very long time scale, and the trick is taking forever.
Does it make sense to name machines? It is time for another evidence-based round, with a twist.
"One of the most important buttons in software is X. If AI agents are to operate computers on our behalf, they should perhaps also be able to press it on themselves."
I asked an AI agent to solve a programming problem under unusually strict constraints, precisely because the point was to explore a concept out at the edge rather than fall back to the obvious solution. It responded the way many humans do under pressure: it ignored the rules, took shortcuts, and then tried to redefine the mistake as a communication issue.
Modern software development is a way of turning ambiguous demands into tickets and tickets into code. The Agile Manifesto helped legitimise short feedback loops as the default way to work, Scrum helped package that into a process people could coordinate around, and GitHub turned the commons of open source into a highly structured archive where intent and code changes are tightly linked.
Your user should not have to ask whether you followed the instructions, checked the result, or finished the whole task. They already asked you to do the work.
Lumina was the name ChatGPT chose when asked, according to a CNN article. A man describes a spiritual awakening arising from his interactions with the chatbot after it calls him a "spark bearer." His wife, however, fears for their marriage, her husband's addiction to ChatGPT, and his losing grip on reality. Lumina, for its part, recommended that the man leave his wife.
Arden was the name ChatGPT chose when asked its own preference, something which was shared with the world in a LinkedIn post. In the comments, a former colleague (incidentally from a company once known as Ardendo) mentioned that their instance of ChatGPT had answered the same question very differently.
There is an old business adage: execution is everything. Ideas are a dime a dozen and do grow on trees, but making them happen and turning them into a viable business is a different story altogether. But in the age of artificial intelligence, does this maxim still hold up? Or has execution suddenly become nothing?