IN THIS ISSUE

• Do AI Content Detection Tools Work?

• Dive Deeper: Resources to Refer To

• How To Detect AI-Written Content Without Using Tools?

TOP PICKS

Do AI Content Detection Tools Work?

No.

Not even the paid ones.

There is no such technology yet that can reliably differentiate between AI-written and human-written content.

1. These tools have a high rate of false positives. Meaning, they claim that yes, the content has been written by AI when it is not.

2. They are likely to generate a false positive result for those who speak English as their second language. Or, who may be writing in a language that is not their native tongue.

3. You can easily change the results and get the content detector to say ‘Not written by AI’ from ‘Yes, written by AI’. All you need to do is modify your prompts for your AI writing tool. [See #5 in Dive Deeper: Resources to Refer To section below]

OpenAI , the creator of ChatGPT, sunset it’s own content detection tool in mid-2023 citing lower accuracy

Why This Matters

Because of a lack of awareness, acceptance, and to some extent, fear of AI, many people believe that AI-written content can be detected.

While that may be true, there is a lack of reliability. AI detection tools differ wildly in their results and have a higher rate of false positives.

This can have a seriously negative impact on those whose future depends on their written submissions. Such as university students, content writers, and freelancers.

Your work may get flagged as AI-written for using Grammarly. It's a tool that many writers, this one included, use to check for grammar and spelling errors.

In general, AI detectors work by spotting the hallmarks of AI penmanship, such as perfect grammar and punctuation. In fact, it seems one of the easiest ways to get your work flagged is to use Grammarly, a tool that checks for spelling and grammatical errors. It even suggests ways to rewrite sentences using, you guessed it, artificial intelligence. Adding insult to injury, Gizmodo spoke to writers who said they were fired by platforms that required them to use Grammarly. [Source: Gizmodo]

There are instances where students are facing disciplinary action for using AI tools to submit their thesis, reports, and project work.

What Next?

AI content detection tools, if you continue to use them, are best used as indicators. But, do not trust them and use them to determine the veracity of the content.

Also,

If a tool states that the content is 50% AI-written, it means there is a 50% probability that AI has written that content. And not that 50% of the content is AI-generated.

Dive Deeper: Resources to Refer To

1. Ethan Mollick, Associate Professor at The Wharton School

Can you detect AI writing?

Read This Post

Read This Post

2. Turnitin’s AI Detector: Higher-Than-Expected False Positives, Inside Higher Ed Article

3. Do teachers spot AI? Evaluating the detectability of AI-generated texts among student essays, Elsevier

4. 5 Online Tools Reviewed to Determine if AI Content Can Be Detected?, ZDNet

5. Want Your AI Writing Tool To Sound More Human? Use These 2 Words in Your Prompts.

How To Detect AI-Written Content Without Using Tools?

Erik Ofgang published a wonderful article in Tech & Learning last month. He lists down 10 ways in which you can determine if the content has been written using AI.

I am listing down the 10 ways but I encourage you to read the article to understand the 'why' behind each.

1. The Submission is Too Long

2. The Answer Misses The Mark While Also Being Too Long

3. AI Writing is Emotionless Even When Describing Emotions

4. Cliché Overuse

5. The Assignment Is Submitted Early

6. The Setting Is Out of Time

7. Excessive Use of Lists and Bullet Points  

8. It’s Mistake-Free 

9. The Writing Doesn’t Match The Student’s Other Work  

10. Something Is Just . . . Off 

Read Here