Chapter 13: Writing with Artificial Intelligence

ChatGPT, while advanced and capable of generating coherent and contextually relevant text, still has several limitations that make it inferior to authentic human writing. Here are some key ways in which ChatGPT falls short:

1. Lack of True Understanding: ChatGPT doesn’t truly understand the text it generates. It lacks genuine comprehension and is simply predicting what words and phrases are likely to come next based on patterns in the data it was trained on. Humans possess deep understanding, emotions, and the ability to connect ideas in a nuanced and meaningful way.
2. No Personal Experience or Creativity: ChatGPT lacks personal experiences, emotions, and creativity that humans bring to their writing. It doesn’t have genuine thoughts, feelings, or unique perspectives to draw from.
3. Limited Contextual Awareness: While ChatGPT can maintain a short-term context in a conversation, it can struggle with maintaining a deep, nuanced understanding of long conversations or intricate topics. It may sometimes provide responses that sound relevant, but are contextually incorrect or inappropriate.
4. Inaccurate Information and Biases: ChatGPT can generate inaccurate or outdated information since its responses are based on the data it was trained on. Additionally, if the training data contains biases, the model might inadvertently generate biased or prejudiced content.
5. Unsatisfactory Flow and Coherence: Although ChatGPT can produce coherent sentences, it sometimes struggles with maintaining a natural flow of thought, resulting in responses that may seem stilted or disjointed compared to human writing.
6. Limited World Knowledge and Creativity: ChatGPT’s knowledge is limited to what it was trained on up until September, 2021. It lacks access to new information and developments that have occurred since then. Moreover, while it can mimic creativity to some extent, its creative output is often based on patterns in the training data, rather than genuine imagination.
7. Empathy and Emotional Depth: Humans can infuse their writing with genuine empathy, emotional depth, and understanding of the human experience. ChatGPT lacks true emotions and can’t empathize in the same way.
8. Subjective and Aesthetic Judgments: Humans can make subjective judgments about aesthetics, art, beauty, and emotions that go beyond the scope of ChatGPT’s capabilities.
9. Adaptation to Individual Styles: Humans can adapt their writing to various styles, tones, and preferences of the intended audience. ChatGPT’s style may sometimes be inconsistent or not align well with specific audiences.
10. Errors and Misinterpretations: While humans can often catch errors, ambiguities, or misunderstandings in their writing, ChatGPT can generate responses that are grammatically incorrect, factually wrong, or contextually inappropriate without realizing it.
11. Dynamic and Evolving Responses: Humans can engage in dynamic, evolving conversations, responding to changing contexts and adjusting their responses accordingly. ChatGPT may struggle to adapt to rapidly shifting conversational dynamics.

In summary, while ChatGPT is a powerful tool for generating text and assisting with various tasks, it lacks the depth, authenticity, understanding, and genuine creativity that humans bring to their writing. It’s important to recognize its limitations and use it as a complement to, rather than a replacement for, authentic human writing.

Everything above that line was written by ChatGPT in response to this prompt: “In what ways is ChatGPT inferior to authentic human writing?” The only edits above were to correct for punctuation errors (yes, ChatGPT messes up its punctuation sometimes).

So, if those are the problems, how can ChatGPT help social media managers produce quality content?

Brainstorming

There’s nothing I particularly disagree with up there and I doubt I would have thought of all those problems with ChatGPT on my own (though it was a bit repetitive at times).

That’s the first hint about writing with ChatGPT. It’s good at gathering a range of ideas that you might not have thought of all by yourself. ChatGPT is good at brainstorming.

If you’re a social media manager, the pressure to constantly be coming up with new content can be frustrating, especially as you get the feeling that you have already posted similar content in the past. ChatGPT can help you come up with some new content ideas.

Outlining

Similar to brainstorming, ChatGPT will suggest an order for how information should be provided. It might need to be corrected, but it’s a starting point for how to organize your longer content, such as a blog article.

Speed

Even the most prolific writers with the quickest fingers in the planet would be hard pressed to produce quality original content at faster than 50 words per minute. (Yes, folks can type faster, but as they do, their ability to focus on quality content goes down.)

On the other hand, ChatGPT can produce about 400 words of passable—though rarely great—content in a few seconds. But there’s the problem again: a large amount of quickly available text is only good if the text is good; it’s mostly in the range of “okay.”

What does that mean for social media managers? ChatGPT can increase the speed of producing draft paragraphs. As noted above, ChatGPT loses the plot after a while, so you can’t use it to produce a first draft of a long document, but you can use it to produce a few paragraph, perhaps one contained section, as seen above. Given the problems, you’ll need to re-write those paragraphs to inject emotion and examples. You’ll need to make sure there’s continuity from one section to the next. But you’ll have a bit of draft material to work with more quickly.

Availability

Your generative-AI writing assistant never sleeps and never takes a day off. We mortal humans, on the other hand, need sleep and days off. Without that, our writing quality gets worse and worse.

Appropriateness

As a follow-up to the previous paragraph, ChatGPT never shitposts or drunk posts. The program is less likely to make embarrassing typos from fatigue or intoxication. For example, how easily can you post about your organization’s pubic places, instead of its public places? Oops for you, but that’s a mistake ChatGPT is unlikely to make.

A mistake ChatGPT will sometimes make, though, is to produce content that is politically problematic, such as reinforcing negative stereotypes. That must be accounted for and corrected.

And, for the record, when you’re tired, it’s time to stop posting on behalf of your organization. Never use social media while intoxicated.

Enhancing Vocabulary

In the same way that ChatGPT can help with brainstorming, it can help with rephrasing and vocabulary. The program can be asked to explain something better or with new word choices. In this way, it’s like a thinking thesaurus.

Role Playing

If you’re wondering how your audience might respond to a post, try asking ChatGPT. Ask the program how a well-defined audience (give some demographic and other information about the audience) would respond to a tweet you’ve drafted. Try it out with a few different audience archetypes described. Let this help you anticipate how people might engage with your content. Revise to get the results you want and see what happens.

So much of what is described above boils down to this: ChatGPT will save you time and effort, but you must intervene to ensure you’re posting quality content. As the program wrote previously, “while ChatGPT is a powerful tool for generating text and assisting with various tasks, it lacks the depth, authenticity, understanding, and genuine creativity that humans bring to their writing. It’s important to recognize its limitations and use it as a complement to, rather than a replacement for, authentic human writing.”