AIs must screened for the tasks and jobs as critically and carefully as humans applicants.
AIs must screened for the tasks and jobs as critically and carefully as humans applicants. None of the AIs is certified at grade school level. All of them make mistakes with arithmetic and calculating with numbers. They do not understand units and dimension, scale and numeric values. They have no button “always check your work before you speak”.
If an AI applies for work in medical records, make them take the same courses and be more critically graded. Then audit and grade them severely.
I tested the most heavily marketed ones. I give them all a failing grade for anything that affects money, life, property. I know why they fail and nothing they are doing is changing that. Can you get any human to do a good job ? Usually they know from grade school how to cite sources, reason, have social values and moral upbringing. Know how to listen and remember instructions and not make mistakes if asked to do something a hundred times.
Treat them like humans. We do not let humans get away with sloppy work, too casual attitude when working where life and health are at stake. Most everyone working in health has high barriers for entry and must not only be able to do tasks, but to know why they do it.
Zero mistakes is the goal and the cheapo OpenAI, Grok, Gemini and Copilot things are not anywhere near the minimum. (Said so you can hear my strong disgust with the lot of them).
Can computers help? Of course. If you can trust their companies and business processes to have your best interests at heart. To listen and prove their systems work in ways that any human would have to be tested. They do not, so do not trust or buy them for real jobs.
Richard Collins, The Internet Foundation
