Lesson# 2: Creativity and Perfect Prompting

 

As I’ve said, large language models (LLMs) are a type of artificial intelligence (AI) that can generate text, translate languages, write different kinds of creative content, and informatively answer questions. They are trained on a massive amount of data and can learn to perform a wide range of tasks through repetition. Part of the beauty of the design is that you can take examples of your writing style and train you LLM app to write like you, but with a professional editing process.

LLMs have the potential to be creative. They can generate text that is both original and coherent, and they can even write poems, stories, and computer code. Recently, there was a huge dustup about Meta’s Voice Box (VB) entrance to the marketplace. Through AI/VB you can now mimic someone’s voice. Ferris Bueller would have absolutely loved that. Recently, NPR’s David Brancaccio used an AI version of his own voice to do his telecast. I honestly couldn’t tell the difference. Scary! Think about the ramifications of that!

In last Sunday’s New York Times Magazine (7.16.23), a college professor wrote to “The Ethicist” column to ask the following question:

I am a writer and a college professor (as well as an administrator) . . . I love to write, but when I have to do administrative documents --- I struggle. Would it be wrong for me to use ChatGPT to do this task?

The Ethicist (Kwame Anthony Appiah) said in paraphrase:

Administrative documents are not thought of as original compositions, and so most are working templates, or administrators are given past documents to update. Basically, you are just adjusting the facts with new information. If you are a small college --- that may not be the case, but you shouldn’t worry about not reinventing the wheel, as what you would be doing with ChatGPT isn’t so different.

You must use proper vigilance with the facts as LLMs are still under development. It’s been said that LLMs can “hallucinate” some facts. Looking at the ethics of this, I don’t think you need to credit ChatGPT more than you would be obliged to say that you used last year’s document.

Finally, he said: Your reports are not being evaluated for their scholarly effort, but if you find that they are doing a good job, you might want to speak with your deans to suggest the idea to others.

It is important to note that LLMs are still under development, and they are not perfect. They can sometimes generate text that is inaccurate or even offensive. But overall, LLMs are a powerful tool that can be used for a variety of purposes. They have the potential to be creative, but should be used with caution --- fact-checking is imperative.

Here are some examples of how LLMs have been used for creative purposes:

• Generating text for video games and movies

• Writing poetry and short stories

• Creating new forms of art and music

• Translating languages

• Answering questions in an informative way

And now we can add the drudgery of pedantic administrative documents to the list.

Source: New York Times Magazine, (7.16.23), page 14. Kwame Anthony Appiah.

# # # #