All the lies about ChatGPT
Last month everybody on the internet had to comment about these AI-generated pictures. I don't claim to be an expert, but as a photographer I think that I know a thing or two about these "AI-artworks". They were easy to spot. All of them looked kinda similar. And the best ones were meh.
Now, the next fad has arrived. Over the last weekend the internet exploded over: ChatGPT. This acronym stands for "Chat Generative Pre-trained Transformer". It is a chatbot developed by OpenAI, a research laboratory based in San Francisco. Hundreds of gullible "influencers" jumped on that bandwagon and in unison blew the same horn: "ChatGPT is so frightening good at everything". Some even claimed "it has written my game 100 times better than what I could do."
Let's just have a look at an actual recent conversation with ChatGPT:
I'm at a loss of words about such stupidity. A paperclip has more intelligence than that.
Programmers that use ChatGPT be warned! Would you trust a hobo with your car keys? Any hobo can give better answers to the above questions. And you trust it with your software? "Well, it's better at software" you may say. But let's have a look at a software related real example:
As convincing as this answer sounds - it's utter bullshit. The worst plain old search engine - even google (well, you should better use more privacy oriented search engine) will give you a better answer.
So, what does it tell about a self-declared software specialist that publicly declares on youtube "ChatGPT has improved my software"? That software must have been in a really terrible shape before. I would advise any serious programmers to avoid such anti-experts. Unsubscribe if you have.
There is only one thing that ChatGPT handles well - the authors even say it - language. It can read and "get" what you say and it can write in nearly perfect language. (Yes, it can talk in multiple languages if you convince it) But it does not "understand" the meaning of what you write to it. It is simply a better Eliza. That's all.