Google Gemini is a big fat liar!
Over the last week or so I have been using Google Gemini for a few things. Yes, I'm using what passes for AI now... I've tried a few of them, but Gemini seems like the smarter one. Except it totally isn't. It might have all the knowledge in the world, but it's dumb as a doornail. Not a single creative 'thought' in that data center.
For the most part I use it to verify or maybe improve stuff I made myself. I'm also kind of forced into it since normal search is so terribly bad lately. I hate it and I like it! Mostly it's just frustrating.
I like it because I can blunder my way into answers much faster. Just type in some vague prompt, more-or-less describing what I want, and about half the time I get a decent enough answer that I can use in further research through traditional search.
Other times the chat is a bit longer and the stupid bot actually gives me not-so-stupid feedback and no further research is needed. The frustrating part in all of this is that these chats are often the most intelligent conversation I have with a 'stranger' all day, or week.
Yes, the world at large is mindblowingly stupid lately! Most of you are being out-intelligenced by a chatbot without a personality or indeed actual intelligence!
Anyway, I also like Gemini because I can write a piece of code, or describe what I want from the to-be-generated code, and it confidently recommends dumb shit I can then ignore or figure out on my own. The stupid bot has a tendency to overcomplicate code, or make up function names that do not exist. And more than once I outsmart the thing because it won't provide an objective suggestion for what code or function to create.
What I don't like about all of this is that it slaps functions together to get a workable result, completely ignoring ready-made functions or methods that do what I need in an easier or more efficient way. If I then question the slob-code it defends itself by saying "that's how modern developers do it," well if that's true, then modern developers suck!
The longer the chat goes on the more argumentative it becomes because I simply know better half the time. Even though I'm not at all a modernist expert on the code I write. I'm very much stuck in my old ways - Hence why I use Gemini in the first place. To learn some new stuff with code examples. This is especially noticeable when I doubt the efficiency of my code, if my logic is sound, or ask it for feedback to improve my own slob. The counter-slob the bot churns out is often unusable or a convoluted jumble of crap that doesn't work.
If it works though, it often adds unnecessary steps to my code, or it assumes dumb things from earlier chats that are not at all relevant for the current prompt or use-case. It always has an answer why its ways are better, and when I refuse the reasoning it suddenly agrees with my view like a petulant child. Most of the time I just keep arguing to confirm, reconfirm and check the suggestons I get. Until I'm reasonably sure what 'we' came up with is better than the code I started with. But geez, that's a tedious and weird way to work.
I don't like it, even thought it yielded a few OK results - So far.
I'm not sure if I'm being more efficient using AI to do code, but I suspect I might attain a superiority complex from its stupidity.