AI got something right

by Daniel Fink, MD, Chair, The Quiet Coalition

Photo credit: Pixabay

According to IBM, artificial intelligence (AI) is “technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity, and autonomy.” I have always been suspicious of this technology, because each AI company uses different algorithms and bases its AI on information gleaned from different sources, without disclosing either. Blind reliance on black box-generated information is probably unwise. How does one know the answer is correct?

As recent reports show, my skepticism may be well-founded. The Grok AI chatbot owned by the world’s richest man recently spewed out answers favoring Hitler. A federal judge fined two attorneys for using what appeared to be AI-generated legal filings which included non-existent case law that an AI algorithm generated. More worrisome than that is a study done by researchers at MIT showing that using AI actually led to worse performance on intellectual tasks and changes in the brain, found with special imaging studies.

AI may have gotten one thing right, though. When I did a recent Google search for the safe noise level for the public, AI generated this answer:

“AI Overview: For the general public, a safe noise level is generally considered to be 70 decibels (dB) or below over a 24-hour period. The U.S. Environmental Protection Agency (.gov) identifies this level as one that will prevent any measurable hearing loss over a lifetime. According to the World Health Organization (WHO), sounds below 70 dBA are unlikely to cause hearing damage.”

Ten years ago, when I decided to become a noise activist, most links cited the industrial-strength NIOSH 85 dBA recommended exposure limit for occupational noise, often without A-weighting*.  

I hope that accurate information about the safe noise exposure level will lead to more people advocating for quiet.

A quieter world will be a better and healthier world for all.

*A-weighting adjusts sound pressure measurements to approximate the frequencies in human speech. Because the decibel scale is logarithmic, there is no easy conversion formula to convert A-weighted measurements to unweighted measurements and vice versa. My observation is that unweighted measurements for most sounds except machinery noise are about 5-7 decibels lower than A-weighted measurements.