How to Avoid False AI Detection (Accused of Using ChatGPT?)

Explore the consequences of false AI detections and discover strategies to avoid accusations, such as incorporating human emotion in your writing, utilizing AI as a supplementary tool, and writing using Google docs.

Justin Gluska

Updated August 10, 2023

A cute robot sitting at a desk taking an exam in a school classroom as 4k digital art

A cute robot sitting at a desk taking an exam in a school classroom as 4k digital art

Reading Time: 11 minutes

It's starting to happen more and more each day. AI detection tools are starting to detect writing as being AI-written even though it's been completely written by a human. Whether you're at school, work, or simply just testing your own writing – false positives happen.

The topic of AI Detection has been quite a bit of traction because of companies like TurnItIn releasing their own AI detector. Enterprise tools are popping up aiming to help others identify if something was written with AI. It got really intense in the spring, when innocent students got accused of using AI when they clearly weren't using it.

Some tools like Originality have gotten quite good at picking up AI writing, but I've personally had the tool flag my very own writing as being AI-written several times. It's a coin toss.

I'm in favor of AI detection tools (when they're somewhat accurate) but really only in the copywriting field. In education it's a completely different story. These tools aren't accurate enough to actually get someone in academic trouble, they're just insightful. I think the most accurate and reliable way to check for AI writing is to visualize it, but these tools aren't popular enough yet.

It gets worse when you have a boss or teacher over your head because a detection tool tells them some writing is 26% AI. This doesn't mean a quarter of your writing is AI, it means there's a 26% CHANCE the entire thing was written with AI. Generative AI tools are just moving faster than people can comprehend them.

I've noticed a lot of people want to safeguard themselves from an AI misdiagnoses regardless of if they've ever used ChatGPT. And yes, you still might want to use ChatGPT for writing assistance, but not rely on the entire thing (and you can still get accused). Using ChatGPT to assist your ideas is not the same as using AI to completely write content – and they shouldn't be mistaken as such.

Understanding The Accusations

No matter who you are, you can face some serious & unintended consequences if accused of using AI to write something that isn't true.

In the academic and schooling work you might run into:

  • Credibility Erosion: Being accused of using AI for tasks like research or assignments could equate to plagiarism accusations. If faculty or peers believe you're not doing your work, your academic integrity and credibility can be severely compromised.
  • Punitive Measures: Academic institutions have strict guidelines against dishonesty. Utilizing AI for assignments, research papers, or tests, if proven, could result in failing grades, suspension, or even expulsion. This is even worse at the university level compared to high school.
  • Future Opportunities: Graduate schools, scholarships, and academic positions often require impeccable records. Accusations, especially if proven, can severely hamper or completely destroy one's chances at these opportunities.

Accusations in the business world:

  • Brand Damage: For businesses, the trust of customers is what keeps a company running. Accusations of using AI without transparency can lead to a perception of deception. This can harm a company's reputation, making it challenging to retain customers or attract new ones.
  • Partnership and Networking: Business relies heavily on partnerships and networking. A company accused of dishonest practices may find it challenging to create new partnerships or maintain existing ones.

And in the working world:

  • Professional Reputation: In professional settings, your reputation is your currency. Being accused of relying on AI, especially without disclosure, can lead colleagues, superiors, and industry peers to question your expertise and authenticity.
  • Job Security: Employers value integrity. If an employee is perceived to be cutting corners by using AI tools without authorization, it can lead to disciplinary actions or even job termination. This is especially true if you're working with data that shouldn't be fed to an AI tool like ChatGPT.
  • Career Advancement: Future job prospects, promotions, or professional opportunities might be compromised. Accusations can make it difficult to secure references or recommendations, which play a crucial role in career progression.
  • Legal Implications: In certain job roles, especially where information accuracy and authenticity are vital (like journalism or research), using AI without proper attribution can lead to legal ramifications.

Why People Accuse Others of Using ChatGPT:

  1. Skepticism Over Proficiency: When people come across incredibly informed, articulate, or rapid responses, they might attribute such proficiency to AI rather than human expertise.
  2. Inconsistencies in Responses: Humans are naturally inconsistent and prone to errors. If someone submits information that seems too perfect, well-structured, or lacks the emotional undertones typical of human interactions, it might raise suspicions.
  3. Past Experiences: Some people might have been duped in the past by AI-generated content or services. As a result, they're on guard, quick to assume that a similar interaction could be AI-driven.
  4. Lack of Emotional Depth: ChatGPT, despite its sophistication, often struggles with deep emotional understanding. Responses that lack empathy or human nuance can seem robotic and thereby trigger accusations.
  5. Mistaken Identity: Sometimes, individuals might be familiar with AI outputs from previous interactions and might mistakenly identify human outputs as AI, simply because they sound or seem familiar. I'm a personal victim of this myself.

Avoiding False AI Accusations

If you're working with either AI or non-AI writing and still want to ensure you avoid getting falsely flagged by one of the detectors, here's a few steps you could take:

Add Human Emotion & Variance

This sounds obvious – but it's what makes a human who they are.

Adding human emotion and variance is a distinct advantage that human writers have over AI. AI, no matter how sophisticated, is still essentially pattern-recognition software; it doesn’t truly "feel" emotions in the way humans do.

Injecting personal anecdotes, real-life experiences, raw emotions, and unpredictability into your writing can give it a depth and richness that is inherently human. These detailed, emotionally-charged tidbits are things that AI would not naturally come up with unless explicitly fed such data.

By weaving these authentic, deeply personal elements into your writing, you not only enhance the authenticity of your piece but also move further away from patterns and structures that detection tools might associate with AI-generated content. It's not fair, but it's something we have to do until there's a better system put in place.

Also, embrace imperfections. While AI strives for grammatical accuracy and coherence, humans often use colloquialisms, regional dialects, and even make mistakes. Letting some of these nuances shine through can act as a "human fingerprint" on your work. Don't make mistakes just for the sake of making them, but you don't have to strive for perfection.

Use ChatGPT for Outlines, Not Content

The line between using ChatGPT as a supplementary tool and leaning on it entirely for writing creation means the world of a difference. I use ChatGPT every single day and pretty much stopped using it to completely write stuff. I used to write articles, essays, and even wrote a book about ChatGPT with the help of ChatGPT, but it's still robotic no matter how much you prompt it.

If you're on the side of paranoid and don't want to get your writing flagged as AI or accused of using ChatGPT, just use it for outlines. It helps you get over writers block and will explain things for you, while still letting your writing do most of the talking. Outlines are good for:

  • Efficient Structuring:
    • Getting a framework ready can sometimes be the most time-consuming part of writing.
    • AI can offer a well-organized outline within seconds, streamlining your entire thought process.
  • Personal Touch & Nuance:
    • Your experiences, perspectives, and nuances add depth to your writing.
    • Relying on AI for full content might rob the work of these personal touches (and get you caught).
  • Ensuring Originality:
    • Using ChatGPT just for the outline ensures the core content remains original.
    • This helps in maintaining your unique voice and style throughout the piece. (+ anecdotes)
  • Brainstorming Assistance:
    • ChatGPT can rapidly generate a ton of ideas around a given topic.
    • It can present various angles and viewpoints that you might not have considered initially.
  • Reduced Risk of AI Detection:
    • AI-generated content can sometimes have detectable patterns, even if they seem subtle.
    • Using AI just for outlines reduces the risk of your content being flagged by AI detection tools.
  • Diverse Writing Styles:
    • ChatGPT, or any AI, usually follows certain algorithms and structures.
    • Humans, on the other hand, have diverse writing styles which can change based on mood, experiences, and inspirations. Using just an outline ensures you maintain this diversity.
  • Content Relevance & Current Affairs:
    • ChatGPT has a knowledge cutoff, which means it may not be aware of very recent events or changes.
    • By crafting content yourself, it's easier for you to incorporate recent happenings, trends, and shifts relevant to your topic.
  • Flexibility & Adaptability:
    • An outline is flexible. It serves as a guidepost, but you can deviate, combine points, or introduce new ideas as you write.

Use AI Undetection, Hiding Tools, and Other Ways

If you're genuinely concerned about your content being wrongly flagged as AI-generated, there are several tools and strategies that you can employ to safeguard your work. Here's a look at some methods:

1. AI Undetection Tools:

These are tools designed to cloak or obfuscate AI patterns in a text. They work by introducing variance and irregularities to make content appear more "human."

  • Pro: They can be effective in bypassing certain AI detection algorithms.
  • Con: Using them might alter the essence or quality of your original content.

Some good ones are Undetectable AI (250 free words), HideMyAI (550 free words), or a paraphraser like QuillBot.

2. Document Your Writing Process:

As annoying as it may seem, if you're super super paranoid, take periodic screenshots or maintain drafts as you progress in your writing. These can serve as evidence of your evolving thoughts and writing process. See the next tip for an easier way of doing this

  • Pro: Provides a visual journey of your work, which is hard to replicate with AI.
  • Con: Can be cumbersome and requires diligence.

3. Peer Review:

Having peers or colleagues review your work before submission can help in two ways: they can confirm the authenticity of your work, and their edits or feedback can introduce more human nuances.

  • Pro: Builds a record and testimony of human involvement.
  • Con: Dependence on others may not always be feasible due to time constraints or availability.

4. Utilizing Google Docs Revision History:

Google Docs has a built-in feature called "Version History" that keeps track of all changes made to a document over time. This feature can be incredibly beneficial in validating the authenticity of your work.

  • Pro: The revision history in Google Docs provides a clear and timestamped record of every edit, addition, or deletion you make. If a visualization tool or reviewer ever questions the authenticity of your content, this chronological trail can act as compelling evidence of your active involvement in crafting the document. It shows a real-time progression of your thoughts and writing, which is hard for AI to mimic.
  • Con: Relying on Google Docs requires continuous internet connectivity and trust in the platform's privacy. There's also the risk of collaborators (if any) making significant edits, which might complicate the document's history.

5. Offline Writing:

Consider drafting initial versions offline, on paper, or on devices not connected to the internet. This can act as tangible evidence of human creation.

  • Pro: A physical version can be a definitive proof of authenticity.
  • Con: May not be convenient for everyone, especially those used to digital platforms.

Final Thoughts

While AI advancements offer numerous benefits, they also come with challenges like false AI detections. I hate to say it but it's a real concern to pay attention to, even if you've done nothing wrong. As AI continues to evolve, it's crucial for our society to adapt, ensuring that we harness its potential without falling prey to its pitfalls and false detections.

If a school or work environment is really that concerned about someone using AI, you should just revert back to pen and paper until we figure out the real solution! I mean if ChatGPT can write something to finish an assignment or task in a few seconds, maybe it's time to start rethinking the work that we're doing!

Rather than seeing AI as a threat or substitute, envision it as a collaborative partner. Together, human intuition and AI's vast knowledge can push the boundaries of content creation. It's really only the beginning.

Want to Learn Even More?

If you enjoyed this article, subscribe to our free newsletter where we share tips & tricks on how to use tech & AI to grow and optimize your business, career, and life.


Written by Justin Gluska

Justin is the founder of Gold Penguin, a business technology blog that helps people start, grow, and scale their business using AI. The world is changing and he believes it's best to make use of the new technology that is starting to change the world. If it can help you make more money or save you time, he'll write about it!

Subscribe
Notify of
guest

0 Comments
Most Voted
Newest Oldest
Inline Feedbacks
View all comments