How reliable is AI?
I was working on a lengthy essay involving astronomy and astrology. I thought perhaps the time was right to step up and align with modern working methods and technologies. So, I asked ChatGPT to assist me with data and numbers research. It is important to acknowledge that the assistance refers strictly to numbers and raw data, and no interpretation or possible points of view are involved.
When ChatGP provided me with the answer, I noticed something was off.
I asked ChatGP to verify the information again, and ChatGPT apologised and gave me a different data set.
I still felt something was off, and I started verifying the information.
More than half of it was erroneous, and others inexact.
Most people take the answers provided by AIs for granted. However, this is not the case.
On the one hand, AI is programmed. It can contain both errors and biased information or misinformation on purpose. On the other hand, AI scans the internet for information, and the internet is often a garbage dump. Nowadays, misinformation can have multiple sources. Just because a piece of information is confirmed from multiple sources doesn’t necessarily mean the information is correct.
Human intelligence is unreplaceable and non-reproducible. However, brute intelligence is not enough. Our mind requires perpetual education and challenges to develop.
Education is critical for the healthy development of the individual and society.
Yet, the quality of education has declined and continues to slump.
As with many other things, the educational system is broken.
Discerning between possible and improbable is vital. Instead of memorising data often presented as certitudes, we should teach ourselves and our children to think independently, apply critical thinking and develop various methods of analysing and evaluating information.
#attilakarpathy #chatgpt #AI
