ChatGPT's writing is insipid and boring.

And yet. It may herald a new era of literate writing.

Every time Tyler Cowen compiles a bunch of links about ChatGPT, or generative AI more generally, the goosebumps come back. His latest offering is no different:

Can ChatGPT pass medical licensing exams?  Speculative GPT thoughts.  And “I taught ChatGPT a language.” And “The return of the Socratic method, at scale and on demand.”  Get ready people!  “Rewrite this article so it sounds like a human.”  And Chat GPT recommends the best books on AI.  Get with the program!  Yes, I mean you.  Only yesterday I found the beast very helpful in prepping questions for Glenn Loury (thanks Glenn, that was a great episode!).

I try not to traffic in clichés here, lest I be accused of having used ChatGPT to write my posts, but: buckle up. The future is going to be wild.

Speaking of cliché-ridden prose and ChatGPT, I ran a little experiment in prompt engineering. I wanted to see whether I could force ChatGPT to dispense with its bland and prosaic styling, in favor of a more robust, evocative, and sophisticated prose. I more or less failed. What follows is an examination of my prompts, ChatGPT’s output, and some comments about how a better writer would have phrased things differently.

In the block quote I give above there is a link to a Google document, titled Speculative GPT Thoughts. The author, John Horton, writes:

Writing as a signal of ability is mostly gone. Well-constructed, well-organized writing now means almost nothing about the “author.” It will really be about detecting novel ideas or really thoughtful, original analyses, which will be hard for most people to do. Writing length as a signal of effort is dead.

I’m not sure that I entirely agree, at least as of the middle of January 2023. I don’t necessarily agree with this because ChatGPT’s writing is, at best, average. And the average person’s writing sucks. (If you’re reading this post a few months’ hence, we may well be into the ChatGPT 4 era, in which case the above observation may be accurate.)

Here’s my first prompt: Write a 12 paragraph blog post about how generative ai can be used for idea generation.

Here’s ChatGPT’s output¹:

Generative AI is a powerful tool that can be used for idea generation in a variety of fields. In this blog post we will explore how generative AI can be used to generate new ideas in product design, advertising, and content creation.

One of the most obvious applicatios of generative AI in idea generation is in product design. By using generative AI to create new product designs, companies can save time and resources by automating the design process. For example, a company could use generative AI to create new designs for a line of clothing or to design a new product that has never been seen before.

Another application of generative AI in idea generation is in advertising. By using generative AI to create new ad concepts, advertising agencies can save time and resources while also creating more effective ads. For example, an advertising agency could use generative AI to create new ad concepts for a new product launch or to create new ads for an exisitng product.

Generative AI can also be used for content creation, such as music, stories, and poetry. By using generative AI to create new content, musicians, writers, and poets can save time and resources while also creating new and unique works of art. For example, a musician could use generative AI to create new songs or a writer could use generative AI to create a new novel.

One important thing to note, is that generative AI should not be seen as a replacement for human creativity but rather as a tool to augment it. Generative AI can generate new ideas and concepts that can be used as a starting point for further human development.

One example of this is the use of generative AI in the automotive industry. Companies such as Ford and Volkswagen are using generative AI to help design new cars. The AI is used to generate new concepts and ideas for car design, which are then refined and developed by human designers. This helps speed up the design process and allows for the creation of more innovative and unique cars.

In the field of architecture, generative AI can be used to create new building designs and layouts. By using generative AI to create new building designs, architects can save time and resources while also creating more efficient and sustainable buildings.

In the field of fashion, generative AI can be used to create new clothing designs and styles. By using generative AI to create new clothing designs, fashion designers can save time and resources while also creating more innovative and unique clothing.

In the field of medicine, generative AI can be used to create new drugs and treatments. By using generative AI to create new drugs and treatments, researchers can save time and resources while also creating more effective and efficient treatments.

In the field of art, generative AI can be used to create new artworks. By using generative AI to create new artworks, artists can save time and resources while also creating more innovative and unique art.

In the field of marketing, generative AI can be used to create new marketing campaigns and strategies. By using generative AI to create new marketing campaigns and strategies, marketers can save time and resources while also creating more effective and efficient marketing strategies.

In conclusion, generative AI can be a powerful tool for idea generation in a variety of fields. By using generative AI to create new ideas and concepts, companies, individuals and organizations can save time and resources while also creating more innovative and unique products, services and experiences. However, it is important to remember that generative AI should not be seen as a replacement for human creativity but rather as a tool to augment it.

This is bland and insipid, though serviceable, writing. It’s grammatically correct, and as flavorless as un-sauced pasta. Which isn’t necessarily a bad thing! Un-sauced pasta is better than rotten meat. But this writing suffers from at least four defects:

• It tells, rather than shows. (“In this blog post,” “In conclusion,” etc.)

• It unnecessarily repeats a lot of phrases. While it is true that Martin Luther King, Jr., achieved great effect with repetition in his “I have a dream” speech, ChatGPT is not Martin Luther King, Jr. Repetition works for preachers. It does not work for most other writing.

• It repeats structures: “In the field of fashion,” “In the field of medicine,” etc. This is boring.

• It uses the most common English words, such as “the”, “of,” etc. far too much. Take the second sentence as an easy example: “In this blog post we will explore how generative AI can be used to generate new ideas in product design, advertising, and content creation.” A more succinct and livelier phrasing would be: “Product design, advertising, and content creation will prove fertile ground for generative AI’s idea generation abilities.”

So let’s see if we can do better.

My second prompt: Rewrite this using less repetition. Vary sentence structure and length. Also reduce the frequency of the following words: the, of, and, a, to, in, is, you, that, it.

Its output:

Generative AI is a powerful technology that can be utilized for idea generation across a diverse range of fields. In this blog post, we will delve into how generative AI can be employed to produce new concepts in product design, advertising, and content creation.

One clear use of generative AI in idea generation is within product design. By utilizing AI to create novel designs, companies can automate the design process, saving time and resources. For instance, a company could use AI to generate new designs for a clothing line or to design a product never before seen.

Advertising agencies can also benefit from generative AI in idea generation. By using AI to generate new ad concepts, they can create more effective ads while also saving time and resources. For example, an agency could use AI to develop new ad concepts for a product launch or to generate new ads for an existing product.

Generative AI can also be employed in content creation, such as music, stories, and poetry. Musicians, writers, and poets can use AI to create new and unique works of art, saving time and resources in the process. For instance, a musician could use AI to compose new songs or a writer could use AI to write a novel.

It's important to note that generative AI should not be viewed as a substitute for human creativity but rather as a tool to enhance it. AI can generate ideas and concepts that can serve as a starting point for further human development.

The automotive industry is one example of this, companies such as Ford and Volkswagen are using AI to assist in the design of new cars. AI is used to produce concepts and ideas for car design, which are then refined and developed by human designers, speeding up the design process and allowing for the creation of more unique cars.

Architects can also benefit from generative AI by using it to create new building designs and layouts. AI can produce more efficient and sustainable buildings while saving time and resources.

Fashion designers can use generative AI to create new clothing designs and styles, saving time and resources while producing more unique clothing.

In the field of medicine, AI can be utilized to generate new drugs and treatments, leading to more effective and efficient treatments while saving time and resources.

Artists can also benefit from generative AI by using it to create new artworks, saving time and resources while producing more innovative and unique art.

Marketers can use generative AI to create new marketing campaigns and strategies, leading to more effective and efficient marketing strategies while saving time and resources.

In conclusion, generative AI can be an incredibly useful tool for idea generation across a wide range of fields. By using AI to generate new ideas and concepts, companies, individuals, and organizations can save time and resources while creating more innovative and unique products, services, and experiences. However, it's important to remember that generative AI should not be seen as a replacement for human creativity, but rather as a tool to augment it.

Let’s look at the second sentence again. Rather than write a more vigorous sentence, it creates an even more awkward one: “In this blog post, we will delve into how generative AI can be employed to produce new concepts in product design, advertising, and content creation.” This is average, and boring, writing!

My last prompt: Rewrite this. Make sure to eliminate the use of phrases such as “in this blog post” and “in conclusion.”

Its output:

Generative AI is a powerful technology that can be utilized to generate new ideas across various fields. We will explore how it can be employed to create new concepts in product design, advertising, and content creation.

One notable application of generative AI is in product design. Companies can automate the design process and save time and resources by using AI to generate new designs for a clothing line or to design a never-before-seen product.

Advertising agencies can benefit from generative AI as well. By using AI to generate new ad concepts, they can create more effective ads while saving time and resources. For instance, an agency could use AI to develop new ad concepts for a product launch or to generate new ads for an existing product.

Generative AI can also be employed in content creation such as music, stories, and poetry. Musicians, writers, and poets can use AI to create new and unique works of art, saving time and resources in the process.

It's important to note that generative AI should not be viewed as a replacement for human creativity but rather as a tool to enhance it. AI can generate ideas and concepts that can serve as a starting point for further human development.

The automotive industry is one example of this, companies such as Ford and Volkswagen are using AI to assist in the design of new cars. AI is used to produce concepts and ideas for car design, which are then refined and developed by human designers, speeding up the design process and allowing for the creation of more unique cars.

Architects can also benefit from generative AI by using it to create new building designs and layouts. AI can produce more efficient and sustainable buildings while saving time and resources.

Fashion designers can use generative AI to create new clothing designs and styles, saving time and resources while producing more unique clothing.

In the field of medicine, AI can be utilized to generate new drugs and treatments, leading to more effective and efficient treatments while saving time and resources.

Artists can also benefit from generative AI by using it to create new artworks, saving time and resources while producing more innovative and unique art.

Marketers can use generative AI to create new marketing campaigns and strategies, leading to more effective and efficient marketing strategies while saving time and resources.

In summary, generative AI can be a powerful tool for idea generation across various fields. By using AI to generate new ideas and concepts, companies, individuals, and organizations can save time and resources while creating more innovative and unique products, services, and experiences. However, it's important to remember that generative AI should be viewed as an augmenter rather than a replacement of human creativity.

This is certainly better than the first version. And yet. Looking at the last paragraph of this output, I don’t like “In summary,…” Show me, don’t tell me. I know that the last paragraph summarizes the previous paragraphs. I don’t need to be told that it’s a summary! The second sentence of the last paragraph has a strange parallelism to it, which, while not inaccurate, sounds more awkward than informative. Finally, the last sentence of the post is, while true, also somewhat of a non sequitur.

So what we’re left with is a tool, albeit a very impressive one, which spits out boring, insipid, and average prose. It will not replace talented writers, at least not yet. I like Ethan Mollick’s suggestions:

You can ask the AI to use specific styles for writing. You will get different results from asking for an academic essay versus a persuasive article versus a blog post versus a corporate memo. The results are often surprising. So you should consider trying several writing types when experimenting with prompts.

Another way to get interesting writing out of ChatGPT is by asking the AI to be someone else. You can have the AI play characters by prompting it to think of itself as a chef, or a novelist, or Plato. This can lead to delightful results. You can start with this list of hundreds of persona suggestions. I used a modified version of the storyteller prompt: “I want you to act as a storyteller. You will come up with entertaining stories that are engaging, imaginative and captivating for the audience. It can be fairy tales, educational stories or any other type of stories which has the potential to capture people’s attention and imagination. You should provide lots of detail and make the story memorable.”

ChatGPT is a tool. It is a very impressive tool. And, like any other tool, if you use it correctly, you can hew your own David.

Use it poorly, though, and you’ll just generate words that any random seventh grader could have written.

1. For this post, I am using the convention of formatting ChatGPT’s output as code, even though natural language text isn’t computer code. I just think that this makes it easier for the (human) reader to discern the difference between AI-generated text and human-generated text in this post.