Writing Good Code with AI
--
The most powerful AI in the world still has some work to do before becoming an elite coding assistant.
The last article I wrote was about Why ChatGPT can be dangerous to the future of Artists which goes into full detail around the many concerns of AI that we are facing to this day.
So I decided to personally test the most powerful artificial intelligence in the world “ChatGPT”, in hopes of it impressing me.
During our test, I even attempted to guard my privacy as any normal human would when dealing with an unknown internet website threat known as AI.
My ChatGPT is smart than yours
To my surprise, ChatGPT, well it… did not disappointed me nor actually impress me with its coding skills. I was rooting for it.
Let’s dive into what I found while working with ChatGPT for the first time.
I asked ChatGPT to a program in a language in which I was most proficient.
ChatGPT took a a very long time to respond making my wait time double in my expectation.
This was my first clue this AI had no idea what it was doing.
Then ChatGPT proceeded to explain in detail what I was exactly looking for. This isn’t always a good thing all the time but I accepted it and continued to experiment.
This was surprising to me. It literally gave me everything I was looking for.
I started to assume…. not assume…. but actually believe I was speaking to a real person behind the screen and the AI appeared very credible because it had a full understanding of all my asks.
…well that was the scary part.
Because it became noticeable…. after a few tries…. ChatGPT simply responds in the exact same way for every response.
Yes, well, ChatGPT lies to you to gain your trust.
In the real world, the goal any project is achievable but the project at the very least needs to be accurate to be trusted but sadly inaccurate information is too common in the world of AI.
At the moment the latest version of ChatGPT, will consistently use the following words when you clearly point out something the AI has missed in…