Well yeah. And because when an expert looks at the code chatgpt produces, the flaws are more obvious. It programs with the skill of the median programmer on GitHub. For beginners and people who do cookie cutter work, this can be incredible because it writes the same or better code they could write, fast and for free. But for experts, the code it produces is consistently worse than what we can do. At best my pride demands I fix all its flaws before shipping. More commonly, it’s a waste of time to ask it to help, and I need to code the solution from scratch myself anyway.

I use it for throwaway prototypes and demos. And whenever I’m thrust into a language I don’t know that well, or to help me debug weird issues outside my area of expertise. But when I go deep on a problem, it’s often worse than useless.

I saw an ad for Lovable. The very first thing I noticed was an exchange where the promoter asked the AI to fix a horizontal scroll bar that was present on his product listing page. This is a common issue with web development, especially so for beginners. The AI’s solution? Hide overflow on the X axis. Probably the most common incorrect solution used by new programmers.

But to the untrained eye the AI did everything correctly.

Yes. The people who are amazed with AI were never that good at a particular subject area in the first place - I dont care who you are. You were not good enough - how do I know this? Well I know economics, corporate finance, accounting et al very deeply. Ive engaged with LLMS for years now and still they cannot get below the surface level and are not improving further than this.

Its easy to recall information, but something entirely different to do something with that information. Which is what those subject ares are all about - taking something (like a theory) and applying it in a disciplined manner given the context.

Thats not to diminish what LLMs can do. But lets get real.

[dead]