'Black Swan' Author Doesn't Want You To Write Your Research Paper With ChatGPT: Here's Why

Nassim Nicholas Taleb, the acclaimed author behind "Black Swan," voiced reservations about OpenAI's chatGPT's suitability for scholarly endeavors, highlighting concerns over its unreliability and referencing errors. 

What Happened: Giving fuel to the discussions about the limitations and potential risks of relying on AI language models like chatGPT for scholarly writing, Taleb took to Twitter to raise concerns about the reliability of the chatbot for serious research. 

He highlighted a reference error made by chatGPT's paid version where Scottish geneticist DS Falconer was mistakenly called "Mary Falconer." 

See Also: Check Out These 3 Jaw-Dropping ChatGPT Plugins… Today!

Talen further shared a screenshot where he appeared to have used "a few plugins," which referred to DS Falconer as Micheal Falconer. 

Now thanks to a few plugins, DS Falconer, formerly Mary, became Michael. pic.twitter.com/vkFnd9KqQZ

— Nassim Nicholas Taleb (@nntaleb) July 4, 2023

In the comments sections, while some users suggested Taleb to use another plugin named ScholarAI but admitted about chatGPT not being 100% reliable, others said that instead of exploratory research, the chatbot could be leveraged to summarize things. 

Thanks!

— Nassim Nicholas Taleb (@nntaleb) July 4, 2023

I use it for the other side of the barbell.

Instead of for exploratory research, I use it to reduce time on specific documents, summarizing things I may/may not want to read.

I have a set of long prompts, curtailing its "creativity", and only using it as a read/skip filter.

— Salim 🦇🔊🦞🌱 (@SaintSal) July 4, 2023

Why It's Important: Last month, it was reported that two New York lawyers were slapped with a $5,000 fine after unwittingly falling victim to the creative legal research capabilities of OpenAI’s chatGPT. 

The lawyers reportedly incorporated fictitious case citations generated by OpenAI's chatGPT into their legal brief without knowing that the information was completely made up. 

Generative AI models such as OpenAI's chatGPT, Microsoft Corp's MSFT Bing AI and Alphabet Inc.'s GOOG GOOGL Google Bard are known for their tendency to produce fabricated information confidently, a phenomenon commonly referred to as "hallucinations."

In April this year, Google CEO Sundar Pichai also admitted that current AI technology generally struggles with “hallucination problems.” 

Check out more of Benzinga’s Consumer Tech coverage by following this link.

Read Next: ChatGPT Gives Bing Search Feature A Break: What Went Wrong?