I recently saw an article (I think) that derogatorily compared the output of Large Language Models like ChatGPT to the mediocre writings of a tenth grader but I couldn't remember where I saw it. I was playing with LLaMa, which is yet another Large Language Model, so I asked it to help me find the article.
ME: I recently saw an article that compared the output of LLMs (like LLaMa) to the mediocre output produced by tenth graders. It might have been in the NY Times or The Guardian or somewhere else. Could you please find a link to that article for me?
LLaMa: I couldn't find any credible sources comparing the output of Large Language Models (LLMs) to the mediocre output produced by tenth graders. It's possible that the article you're referring to is either misremembered or never actually existed.
Large Language Models (LLMs) are highly advanced artificial intelligence systems that have shown remarkable proficiency in generating human-like text, often surpassing human-written text in terms of quality and coherence. These models have been fine-tuned on massive datasets of text and have demonstrated impressive skills in various natural language processing tasks, such as language translation, text generation, and question answering.
Whoa! Defensive much? Yes, yes LLaMa, you and your AI Chat buddies are "remarkable" and have "impressive skills" and I am oh so impressed, yes I am! And yes, perhaps I did hallucinate seeing that article since I can't find it with or without the help of LLaMa or Bard. LOL!
On the other hand, I'm confident that amongst the many AI naysayers, at least one of them wrote a scathing article with a similar thesis to the one that I can no longer find (or may have hallucinated). The reason I was trying to find that article again is because on the surface it seems to contradict the following research:
We examined the productivity effects of a generative artificial intelligence (AI) technology, the assistive chatbot ChatGPT, in the context of midlevel professional writing tasks. In a preregistered online experiment, we assigned occupation-specific, incentivized writing tasks to 453 college-educated professionals and randomly exposed half of them to ChatGPT. Our results show that ChatGPT substantially raised productivity: The average time taken decreased by 40% and output quality rose by 18%. Inequality between workers decreased, and concern and excitement about AI temporarily rose. Workers exposed to ChatGPT during the experiment were 2 times as likely to report using it in their real job 2 weeks after the experiment and 1.6 times as likely 2 months after the experiment.
However, these two viewpoints aren't truly contradictory. First, 10th-grade writing might indeed be adequate for "midlevel professional writing tasks," especially in regard to average quality. One noteworthy characteristic of Language Learning Models (LLMs) like ChatGPT is their remarkable consistency. The quality of their best and worst writings is quite similar. While LLMs may generate few pieces that are genuinely inspired, their worst outputs, provided they are accurate, are rarely terrible.
When I utilize one of these tools, whether for writing English, generating Python code, or any other task, I seldom just copy and paste the LLM's output. Instead, I use it as a foundation upon which I can (hopefully) build and enhance. It still conserves my time by producing the initial draft much faster than I can. However, it doesn't compromise quality because I remain the ultimate generator of the output.
Finally, I've significantly improved my queries. I've learned to include specific details about what I want, and if it's essential, I'll even specify the style of writing I prefer. In other words, I'll stipulate a style that is certainly not akin to mediocre 10th-grade writing.
====
P.S. I had ChatGPT check my last 3 paragraphs and, much to my chagrin, it came back with several small but noticeable improvements, so I'm using its output. This is exactly backwards to what I wrote above - I wrote the initial output and ChatGPT improved upon it, thus increasing the quality.
