Comment from Anonymous
Anonymous AnonymousSupportIndividual
Summary: The commenter argues that AI systems are prone to regurgitating misinformation and lack the human judgment necessary to distinguish between correct and incorrect data. They advocate for increased human staffing rather than relying on chatbots to handle complex tasks or ensure safety.
Have you ever looked at Wikipedia's page "List of common misconceptions?" It is far from all-encompassing, but it serves as an example of a variety of things that most people are wrong about. Most "AI" programs that are advertised as assistance for real-world tasks work by compiling large amounts of data together and generating an answer that *best resembles* the most common responses to the prompts it's given. Because of this, if there is bias in its training data, say, some information that most people are wrong about, or if the user is trying to do work that requires precision (such as using specific units of measurement), or if anything related to the user's task involves something that is the exception from a general rule, the AI program will not be able to determine the difference between information that is correct and useful, information that is irrelevant to its task, and information that is incorrect and harmful. It will simply regurgitate answers that *resemble* what has been put into it.
Further, these tools are being sold to a consumer base that is already engaged with a very dangerous trend: short-staffing. Over the past 20 or so years, it has become extremely common for businesses all industries to attempt to employ fewer people without any meaningful reduction in or increased efficiency of output. This habit is dangerous because it dramatically undermines the ability to detect and correct mistakes -- a protection that is already necessary for humans, but becomes even more relevant with the involvement of generative AI, which is not capable of recognizing mistakes in the way that a human can, or, more accurately, in the way that multiple humans covering for each other can. What businesses need, for both the ability to fulfill their obligations to consumers, and, more critically, for their ability to operate safely and prevent emergencies, is more human hands and eyes to divide the workload between, NOT a mindless chatbot that can only reference its training data, and not the world around it.