𝐂𝐡𝐚𝐭𝐆𝐏𝐓 : 𝐅𝐞𝐰 𝐎𝐛𝐬𝐞𝐫𝐯𝐚𝐭𝐢𝐨𝐧𝐬

𝑪𝒉𝒂𝒕𝑮𝑷𝑻 𝒉𝒂𝒔 𝒃𝒆𝒄𝒐𝒎𝒆 𝒔𝒐 𝒔𝒖𝒄𝒄𝒆𝒔𝒔𝒇𝒖𝒍 𝒕𝒉𝒂𝒕 𝒏𝒐𝒘 𝒂 𝒃𝒍𝒐𝒈 𝒑𝒐𝒔𝒕 𝒎𝒖𝒔𝒕 𝒔𝒕𝒂𝒓𝒕 𝒘𝒊𝒕𝒉 "𝑻𝒉𝒊𝒔 𝒑𝒐𝒔𝒕 𝒘𝒂𝒔 𝒏𝒐𝒕 𝒈𝒆𝒏𝒆𝒓𝒂𝒕𝒆𝒅 𝒘𝒊𝒕𝒉 𝑪𝒉𝒂𝒕𝑮𝑷𝑻". 𝒀𝒆𝒔, 𝒊𝒏𝒅𝒆𝒆𝒅. 𝑻𝒉𝒊𝒔 𝒑𝒐𝒔𝒕 𝒊𝒔 𝒏𝒐𝒕 𝒈𝒆𝒏𝒆𝒓𝒂𝒕𝒆𝒅 𝒘𝒊𝒕𝒉 𝑪𝒉𝒂𝒕𝑮𝑷𝑻.

There is no doubt that ChatGPT has made Artificial Intelligence accessible to all.  It has shown abilities to write a poem, construct a tag line, build code, compose essays, and even musical notes in particular styles. All these seem magical. However, the following are the critical points to be pondered:

-  Lack of Source Transparency:  ChatGPT is constructing the responses from the sources but the sources are not transparent. As the sources are not transparent credibility determination becomes risky. 

- Inaccuracy: It is producing wrong answers without any warning. I am not mentioning about the complex mathematical queries but the simple text retrieval. In a conversation when I asked to give me a list of Indians born on December 12, the first name was Mahatma Gandhi. Satyameva Jayate :)

- Inconsistent Features: Another's thing that I observed with ChatGPT is that the prompts for which it answered earlier have now become "No response" prompts. For example earlier (immediately after release) when I asked it to give a sequel plot for Anbae Sivam movie, it gave a wonderful plot but yesterday when I tried, it simply denied. This is a critical problem when the features are not consistent.

Don't get me wrong that I am saying ChatGPT is of no use. It has lots of use cases. However, for a tool with tall claims, inaccuracy, and inconsistent features are major show-stoppers in making it a part of your regular workflow. My concern as a teacher is that if young students assume AI is like a magic box, we may end up with a situation where no one knows which information is real and which is fabricated.

Let us hope that we do not reach that point.

#artificialintelligence #ai #chatgpt #chatgptissues #openai