Hi ,
If you had to choose between believing a human or AI, who would you believe?
Research shows it's probably the AI.
Several years back (pre-ChatGPT), while rushing to meet a deadline, I made the mistake of blindly copying Python code from Stack Overflow because I trusted the (unknown) person who posted it on the site. The result was hours of work in restoring my coding environment to a usable form after it crashed.
Recent research has shown software developers behave similarly when given coding advice by Generative AI - blindly downloading Python packages that don't exist and opening the door to cybersecurity threats.
This creates a problem when using AI as a
coding assistant while assuming (your own) human oversight as your last line of defence.
Here's the thing...
Generative AI is fantastic and has the potential to save hours of precious time. But automating tasks previously
performed by humans also has the potential to de-skill humans in performing those tasks - making it harder to spot errors when they occur.
Any muscle will atrophy when not used, and there is a fine line that exists between AI assistance and AI reliance.
By all
means, use AI to help make you better at what you do.
But if doing what you do requires you to retain a certain set of skills to effectively oversee the AI, then it may be beneficial to put the AI aside from time to time to avoid losing those skills altogether.
As
the old saying goes, "move it, or lose it".
Talk again soon,
Dr Genevieve Hayes.
