Raw LLM Responses

I am a retired programmer, but I still keep my hand in writing my own software. There are some things I use a lot, so when I got this new PC a couple of weeks ago, instead of installing my applications, I rewrite them, sometimes adding features or maybe just adjust the interface to suit how I now work. So, I loaded Visual Studio and for the first time I had a Copilot Agent watching over me. I had gotten as far as the application name and the TODO list comments which are really just for me to remember which parts I have done and still to do. This was all that the AI needed, it popped up with ideas, so out of interest I let it change the bare Windows application just to see what it does. within about 1/2 hour I had the first of my applications (in this case Image/Video library with some editing functionality) and I did not write a single line of code. I did push it because it was not good at safety and security, I did get it to put some in, that was not great, it assumes it is going to get good data so it does nothing by default to test the data and take action for bad data. But still, imagine new programmers leaving UNI this year, they will make use of AI to write their software because it is easy and very quick to do, they will not know to add security and stability. I should imagine many contracts will be done using this technique, application and factory software out there and no one, including the programmer, knows how it works. In time there will be less and less experienced programmers out there to put it right, so the world will in time be running off weak, insecure software that no one is able to repair. It is quick and will be used, less of these sorts of developers will be needed because of the speed advantage. God help the world when everything it relies on is based on systems that no one alive understands. There is not even a certainty that future AI can maintain that code.