I’m interested in AI, neural networks and machine learning. Pictured above is a book I haven’t finished reading yet.
And I think neural networks are amazing.
But what I want to share today is my feeling that the current state of AI will only ever be able to solve problems created by humans.
Whereas humans solve problems.
The difference is a yawning chasm.
A simple example was provided by a friend of mine who works with metal.
He had to do a quick fix to a piece of metalwork, but complained to his boss that completion of the required health and safety assessment would take longer than the job itself. So his boss said don’t worry, we’ll get AI to write it, which it did with great proficiency.
So the problem – writing the assessment – was solved.
But the purpose of writing the assessment – to consider and inform the participants of the dangers of that job – was completely circumvented.
This was a great example of AI completing a task that was intended to show that the worker had sufficient experience and common sense to complete a potentially dangerous job with injuring himself or others.
But in the process, all it achieved was ticking a compliance box, because the worker didn’t read the assessment (nor need to in this case).
A bureaucratic task completed by a machine, filed by a human and effectively ignored…