But does crazy speedup of tasks actually yield more productivity? Or does it yield greater numbers of pointless tasks?
Anecdotally, I've seen management using AI to produce more and more flashy looking dashboards and interactive displays of various metrics. Will this help our bottom line?
I’ve watched my work go absolutely ballistic for AI-ing all the things.
Lots of code gets written, that much is certain. A lot of slopped out PowerPoint presentations which are just worse to read than human-written ones. We have _not_ magically accelerated delivery times and the product which is 99% vibe-coded perpetually has “oh that’s weird” bugs. An obscene amount of effort is being poured into internal tooling, 90% of which is reinventing existing tooling. Important to note that the internal tooling isn’t actually usable or helpful in building the product, but we are pouring eye watering amounts of cash into it.
If management spends their idle time building dashboards by themselves instead of taking up developer time asking for dashboards to be built and populated for them, there might be productivity to be had...
Absolutely not. A big chunk of my last job was figuring out ways to gently help managers understand that some dashboard or report they built was producing incorrect numbers because they didn’t get the data munging quite right, or fundamentally misunderstood what some field was measuring.
AI made it worse because LLMs will instantly give a confidently wrong answer where previously that moment of uncertainty might have led to a conversation with the data’s custodian.
Anecdotally, I've seen management using AI to produce more and more flashy looking dashboards and interactive displays of various metrics. Will this help our bottom line?