This is by no means an AI specific problem, but it is something that I personally noticed recently. My feed are filled with all kinds of βTips & Tricksβ:
- The new prompt just dropped β¦ from <INSERT FAMOUS GUY NAME>
- This new skill makes your LLM smart by 100x
- β¦ you get the idea
The reason it stands out to me is because how I actually use Codex day-to-day probably looks more like:
- β<test name>β failed, fix it
- β<command>β I need to run this command, figure it out
- Implement A, consider B, C, D.
Yeah, nothing special, no secret sauce and most of the time it just works. Perhaps those skills does help somewhat, but also in this age of information overloading, maybe just β¦ chill?