Certify7513
> ‘splitting the G’ is the act of taking your first sip so precisely that the creamy white head of your drink drops clean through the centre of the “G” in the Guinness logo printed on the glass.
Asking ChatGPT for "trivial" questions like this is absolutely bonkers. This is a question that a search engine can answer immediately and accurately. There is no reason to gamble whatsoever.
Suggesting that we must "get used to the idea that an 80% confident answer is better than none", despite a properly sourced answer being easier to obtain instantly via a search engine, is bizarre, and only a reflection of how much brainrot these things are causing.
Every time you need to make an update, you need to bring up the old context, or otherwise get the AI up to speed, which especially if you're using one of the frontier models could be a significant financial drain long term.
You don't get the same dopamine hit too, because you're just making boring updates to something which you threw together in 5 minutes with zero effort. The time and financial cost of building all this stuff may have been better spent on one, good, properly architected project.
Maintaining the project manually also assumes you can quickly understand the codebase which has been produced, otherwise you're completely dependent on Anthropic and them maintaining prices which you can afford. Bearing in mind that as you add new features, the cost of getting the LLM to understand the project increases, right? I might have a naive perspective here.
All that being said - sometimes there really are one-off niche things that are just for personal use that you do continue to use long term. Usually the simpler stuff where you can easily grasp the codebase at a quick glance. It's also great for debugging back and forths.
Personally I just run my local setup with a bunch of MCP stuff and the primary way it helps me is to keep me functional and on task. In some ways it's good if the AI can supervise you as opposed to you supervising it - at least from an ADHD perspective.
It's an interesting idea for sure, I like this article and agree with it.
As far as the end of computer programming goes...
Step 1. Wow, I just vibe coded an application and it works! I'm going to write a blog about it and tell everyone how awesome AI is, much hype
Step 2. Vibe coded application faces inevitable problems, the perfect application is a fairytale after all. The only way to "fix" the application is spam tokens at the problem and pray.
Step 3. Author does not write a new blog post to report on this eventuality... probably because they feel embarrassed about how optimistic they were
Step 4. Perhaps author manages to fix application, awesome... then what about a year from now, author needs to update the application because a dependency has a security problem. The application is so needlessly complex that they don't even know when to begin.
Step 5. They boot up Claude Code, which their business is now 100% dependent on, but they're charging 10x the original cost per token. It's not like they have a contract, so user has to either eat the cost or give up
Step 6. User tries local model on their 1080 ti but they can barely run entry-level models
Step 7. Woops
Personally I think it's impossible to convince these people, the results will speak for themselves eventually.