The Turing Test for Writing: Can AI Mimic Human Creativity?

AI is churning out increasingly sophisticated content.

It's getting harder and harder to tell the difference between human and machine-written work.

From news articles to novels, AI-generated text is becoming more coherent, more engaging, and more, well...human.

Which begs the question:

Can AI truly mimic human creativity?

Alan Turing (the father of modern computing) proposed a test for machine intelligence back in 1950.

The idea was simple:

But what about creativity? Is there a Turing Test for that?

It's a tricky question, because creativity is a bit harder to define than intelligence.

It's not just about producing something that looks or sounds human-made.

It's about:

• Originality – coming up with ideas that are new and surprising.

• Emotional resonance – creating work that speaks to the human experience in a profound way.

• Inspiration – tapping into something deeper than just pattern recognition and data analysis.

Can an AI do that?

Can it write a novel that makes us laugh, cry, and see the world in a new way?

Can it compose a symphony that stirs our souls?

Can it paint a masterpiece that captures the beauty of the human spirit?

Well, that’s what I’m going to explore with you today!

We’re going to explore:

1. How hard it is today to identify AI-crafted content.

2. Why AI-generated content may very well be the future of how we all write.

3. Ways you can make AI writing be more passable as human.

Let’s get started.

“Did AI Write This?”

How hard is it to tell the difference today?

According to one study, linguistics experts struggle to differentiate AI-generated content from human writing.

When reviewing research abstracts, these experts could only identify AI-created content correctly 38.9% of the time.

This is with the models we have today—imagine how difficult it will be in 1, 2, 5 years.

Also, think about the shear volume of content being published to the internet every day—humans don’t spend as much time discerning where the content came from.

It’s easy to see how a good portion of AI writing could fly under the radar.

We are still in the early days of AI’s exponential takeoff.

It’s writing will get better as the models do, and as people get better at prompting the models for better output.

Speaking of prompting, let’s talk about that next.

Common Features of AI-Generated Text

Before we learn why some think AI-generated text sucks (hint: it’s their prompts lol), let’s cover some common tells that text is likely generated by AI:

1. Buzzwords

   When you see industry jargon and buzzwords frequently in the text, you can assume it is probably AI generated at this point.

2. Big Words

   Do a majority of people use big words in their everyday writing and speech?

   Probably not.

   So what smart intelligence could possibly write like this?

   Yup—AI.

3. Repetition

   AI models sometimes struggle with repetition, using the same words, phrases, or sentence structures multiple times within a piece of content.

   This makes the text feel monotonous and less engaging.

   Human writers tend to vary their language and sentence structure to keep the reader interested.

4. Lack of emotion

   AI will eventually be able to mimic this pretty well (and can to an extent already).

   But it can be hard for the models to capture the nuance of human emotion that we are used to feeling in human writing.

5. Perfect grammar

   I’m not saying you shouldn’t use perfect grammar now—it’s just another aspect of text to look at when deciding if the text is AI-generated.

There’s also AI detection tools that you can use:

1. GPTZero

2. Content at Scale AI Content Detector

3. Copyleaks

Okay, now let’s talk about why prompting—the reason some people think AI-generated text sucks.

Check Your Prompts

There’s a debate between whether or not AI-generated text is passable online to other humans.

Some say no.

Some say yes.

I say it depends on the prompting.

Why do I say that?

Because, after experimenting with different prompting techniques for Claude 3 for 7 days, it is clear to me that the more effort you put into prompting the model, the better the output will be.

Will we Even Need to Write in The Future?

The value of learning how to write cannot be overstated.

It is the practice of clarifying your thoughts—putting words to paper forces this clarity.

But…

What if we don’t need to write in the future?

According to a recent MIT Technology Review article, one of the biggest changes ushered in by AI language models like ChatGPT might be the shift in how we evaluate written text.

Students might not need to write everything from scratch anymore in the future.

The focus will be on coming up with original thoughts rather than just correctly formatted writing.

As AI researcher Sebastian Raschka says:

Essays and texts generated by AI will eventually start resembling each other as the AI system runs out of ideas, because it is constrained by its programming and the data in its training set.

So while AI might make the mechanics of writing easier, it can't replace the unique perspective and creativity that only a human mind can bring to the table.

The real skill of the future may not be writing itself.

Instead, it will be the ability to think critically, ask the right questions, and come up with truly original ideas that push the boundaries of what's possible.

Ethical Considerations of AI-Generated Content

As AI-generated text becomes more prevalent, it's crucial to consider the ethical implications and potential risks, such as:

• Bias and discrimination from AI models trained on biased data sets

• Plagiarism and intellectual property infringement concerns

• Potential for spreading misinformation if not properly fact-checked

• Displacement of jobs in fields like writing and journalism

A few examples of unethical content:

• Phishing emails

• Misinformation (2024 presidential campaign is going to be interesting to say the least lol)

Being mindful of these ethical considerations is crucial for maximizing the benefits of the technology while minimizing its risks.

Okay so now…

Assuming we have original, creative ideas.

And we have addressed any ethical concerns around the generated content.

How can we make AI-generated text sound more like a human?

Techniques to Sound More Human

While AI-generated text is getting better at mimicking human writing, there are still ways to make it sound more authentic:

1. Use personal anecdotes and experiences:

   AI models are trained on a vast corpus of text, but they don't have real-life experiences to draw from. Including personal stories can help make your writing feel more human.

2. Incorporate sensory details:

   Describe how things look, feel, sound, taste, and smell.

   These vivid details can bring your writing to life in a way that AI might struggle to replicate.

3. Use humor and wordplay:

   AI models can generate jokes and puns, but they often lack the nuance and timing that makes human humor so effective.

   A well-placed joke or clever turn of phrase can make your writing stand out.

4. Vary your sentence structure and length:

   AI-generated text can sometimes feel monotonous, with every sentence following a similar pattern.

   Mix things up with short, punchy sentences alongside longer, more complex ones.

5. Show vulnerability and emotion:

   AI models are getting better at understanding and expressing emotion, but they still can't match the depth and authenticity of human feelings.

   Don't be afraid to let your personality and emotions shine through in your writing.

Example Prompts

Here are a few examples that incorporate some of the aspects we discussed above:

1. All Aspects Incorporated

Ontological Framework:
Narrator traits: Introspective, empathetic, resilient
Values: Honesty, personal growth, family
Experiences: Grew up in a close-knit community, faced challenges that led to self-discovery
Tone: Nostalgic, heartfelt, with a touch of humor

Instructions:
Write a short story about a memorable experience from your childhood that taught you an important life lesson. Begin with a humorous anecdote that draws the reader in, then use vivid sensory details to describe the setting and characters. As the story progresses, vary your sentence structure and length to create a dynamic narrative flow. Show vulnerability by sharing your thoughts and emotions throughout the experience, and end with a reflection on how this lesson has shaped your perspective. Ensure that the narrator's traits, values, experiences, and tone align with the provided ontological framework.

2. Personal Anecdotes, Sensory Details, and Vulnerability

Ontological Framework:
Narrator traits: Nature-loving, intuitive, mindful
Values: Environmental conservation, inner peace, simplicity
Experiences: Grew up exploring the outdoors, finds solace in natural settings
Tone: Reflective, appreciative, serene

Instructions:
Describe a moment of profound connection with nature. Share a personal anecdote about a time when you felt deeply attuned to your surroundings, using vivid sensory details to paint a picture of the scene. Reflect on how this experience made you feel and why it holds significance for you. Ensure that the narrator's traits, values, experiences, and tone align with the provided ontological framework.

3. Wordplay & Sentence Structure

Ontological Framework:
Brand attributes: Innovative, user-friendly, efficient
Target audience: Tech-savvy homeowners, busy professionals, convenience seekers
Desired tone: Witty, informative, upbeat
Style guidelines: Use a mix of technical terms and layman's language, focus on benefits over features

Instructions:
Create a product description for a new, innovative household gadget. Use wordplay and puns to create a catchy and memorable title for the product. In the description, vary your sentence structure and length to highlight the product's key features and benefits in an engaging way. Ensure that the brand attributes, target audience, desired tone, and style guidelines align with the provided ontological framework.

If you’re interested in diving deeper into prompting with writing, you can:

• Read my article on prompting strategies for better writing with Claude 3

• Join my free Skool to learn with likeminded people

With that, let’s wrap up!

Closing Thoughts

So, we learned:

• AI-generated text is actually pretty difficult to identify for humans already.

• The quality of AI writing depends on how well you prompt it.

• Techniques to make AI writing sound more human.

But, does this answer our question?

I think so.

AI can mimic human creativity but it depends on:

• Original input from the human.

• Prompt refinement to steer the writing in a more human style.

If you are worried about AI replacing you and your creativity, don’t be.

Humans will still be able to inject their creativity into their works—it will just be done in a different ways than it is now.

AI-generated content is a new opportunity for us to express ourselves creatively through collaboration with AI—I would recommend you start learning to collaborate sooner rather than later.

That’s all for this week folks! ‘Til next time, much love and peace y’all 🤠