ChatGPT Can Be Devastating If You're Already Feeling Imposter Syndrome
by Mike LevinWednesday, December 21, 2022
So many people are enamored with ChatGPT teaching them how to code. You’ll hear the term “game changer” over and over. The funny thing is that we as human beings have always had the autonomy to do research and figure out how to do a thing, exactly as many of us have with Google. Easy peasy web research was the game changer.
So it’s ironic that today a system that lacks autonomy is helping people find theirs. Keyword and phrase searches that return matching resources isn’t enough. It’s got to perform the research and cleverly rework stuff that it’s found into a canned answer. And it’s sad that people have not had a human presence in their lives that could teach them how to learn for themselves what ChatGPT serves up on a silver platter.
Tapping a teacher for an answer is something you do when you can’t figure things out for yourself. Copy/pasting code out of ChatGPT is only marginally better than copying from Stack Exchange or Stack Overflow. You can get the magic words and do a lot of hand waving to prove to people you’re smart like the AI/Teacher who answered your questions, but the deeper learnings of having struggled through it for yourself are denied.
There could be no better time for me to be relaunching project Pipulate. To Pipulate is putting it all together into a working and automatable system. ChatGPT isn’t going to be able to do that for you. If you’re hosting something for somebody else, host it somewhere off-site. But if you’re doing the work for just yourself, keep your code at home and don’t pay for cloud services until you really have to.
GitHub is the exception today, and I’m leaning heavily on it for keeping my repositories, but that won’t last long. Breaking my GitHub dependency is high on my list of what Pipulate should evolve into. Pipulate at home will best be served as something that can host LXD Linux containers, plus 2 levels of file backup—plus the original on mirrored drives, so your data in no less than 3 @home locations.
As you do stream of consciousness writing like this, stay aware of meandering too far away from the goal of the writing. Sometimes we use this sort of writing to direct ourselves. This is the assertion of autonomy. We are channeling from places unknown and are able to examine our awareness and interrupt our flow. I should challenge ChatGPT to my current challenges. That can be a great trick to spur me on, perchance improve the quality-level of my work incorporating in anything ChatGPT’s seen that I haven’t.
I could really run with that. If you’ve been feeling like an imposter for a long time anyway, and along comes ChatGPT to beat you at your game easily, then it might be be time to do some soul searching. Writing is a soft skill. Writing code out of context of it actually running is a soft skill. Making a system you can use every day is a hard skill.
How long before you can talk an AI through the Browser-based web scraps automation you plan to do? Won’t it be able to do anything you can do as a human with a mouse and keyboard? Of course it will, and plenty of user-simulating “bots” already exist today. It’s just not a convenient easily accessible capability today. It takes some coding expertise.
There’s a lot of time between now and when an AI will adjust your Linux daemon scheduled tasks and server configuration at home without overlord cloud assistance. But we’ll get there, and between here and that vision is Pipulate. Eventually, all the Python code in Jupyter to Linux daemon conversion should be conversationally based. You talk code into existence in Jupyter (without paying cloud services). There’s an unlikely but potentially popular and soul-cleansing path here. Articulate it and promote it.