Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
You need to get Jacob as a guest again. Preferably with Mike Elgin.
1 Like
Mike Elgin or Mike Elgan?
1 Like
If you don’t know how to code, how do you know the code produced by AI is correct? Doesn’t include bugs or security holes?
A lot of human programmers have been caught using example code, marked as example code with tags not to be used for production purposes, but they did anyway… And paid the price.
AI is being taught on the same example code, plus projects which have incorporated the buggy example code.
Somebody somewhere still needs to double check the AI produced code, just like we have to double and triple check the code produced by humans.
2 Likes
He sounds a lot like Mr Elgin. He was a good guest
3 Likes
One of the core issues with AI programming is that most of the cost associated with a service is its maintenance and day to day operations.
Providing support, updates and bug fixes.
AI will help someone who is not a coder write the first version of an app. It will do nothing for them when they want to investigate a bug.
1 Like
Well, it might swap one difficult bug for a half-dozen newer ones? 
2 Likes
I have never seen an example of AI being used to investigate a bug. The person who is using AI will need to understand the code in order to debug it an potentially ask AI questions.
What I have seen so far is someone who is not an expert asks AI to,produce some code and then use it as is without understanding exactly what is going on.
Of course a developer that uses AI as a glorified autocomplete they will be able to debug but in this case this is not code written by AI.
This is how I use copilot as well.
Security companies have been using AI to study code for a long time, looking for weaknesses and, on the other side, recognising malware through its actions.
1 Like
It would be interesting to see what techniques are used for this.