The Best Chat GPT Detector is Here (10 Testruns)

The Best Chat GPT Detector is Here (10 Testruns)

by The AI Advantage

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQFGenerative artificial intelligence
🎙 The AI Advantage 👥 480K 📅 February 1, 2023 ⏱ 38 min 👁 10K 📄 expert opinion 🧭 2026-09-08
Available in: English (current) Français

Keywords

AI Text ClassifierChatGPTAI detectionOpenAIpractical test

Summary

The video presents a live, hands-on test of OpenAI’s AI Text Classifier, released in January 2023. The host, Igor Pogany, explains the tool’s purpose, its limitations (minimum length, English-only, not foolproof), and then conducts a series of tests. He feeds the classifier various texts: a basic ChatGPT-generated essay, a text in the style of a human (John Oliver), a text in a style not typical of AI, a handwritten email, and a text in the style of Borat. He also tests the impact of rephrasing and using a paraphrasing tool (Quillbot). The results are mixed: the classifier correctly identifies some AI-generated texts as ‘possibly AI generated’ but fails on others, especially when the text is rephrased or stylized. The host concludes that the tool is not a definitive solution and can be fooled by determined users, but it can catch lazy copy-pasters. He emphasizes the need to adapt to a new world where AI-generated text is prevalent.

157 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides practical, empirical value by testing the AI Text Classifier in real-world scenarios. The host’s argumentation is based on direct observation and transparent methodology, though not statistically rigorous. He clearly acknowledges the tool’s limitations and the ease with which it can be bypassed, which adds credibility. The live interaction with the audience and the polls make the testing process engaging and transparent. However, the conclusions are based on a small, non-representative sample, and the host’s interpretations are sometimes speculative (e.g., his question about probability interpretation). Overall, the value lies in the practical demonstration and the honest assessment of the tool’s capabilities.

Scientific Rigor, Source Quality, Title Accuracy

The video cites the official OpenAI AI Text Classifier page and the OpenAI blog as primary sources, which are appropriate. The host also references his own previous videos on GPT-2 detectors, providing context. The title accurately reflects the content. The video is not a scientific study but a practical review, and the host does not claim otherwise. The methodology is informal, but the observations are presented transparently. The video’s strength is its practical relevance, not its scientific rigor. The host’s personal opinions are clearly distinguished from factual observations.

206 words

Title / Content Match

The title accurately reflects the content: the video tests the OpenAI AI Text Classifier, which is presented as the best detector at the time.

Quality & Reliability

6/10

The video is a practical, hands-on test of OpenAI's AI Text Classifier, conducted by a content creator with no formal research background. The methodology is informal and not scientifically rigorous, but the observations are transparent and the limitations of the tool are clearly acknowledged. The score reflects the practical value and honesty of the testing, tempered by the lack of statistical rigor.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

  • OpenAI Blog — The official announcement of the AI Text Classifier, which the video references and tests.

Contribution & Novelties

The video provides a practical, real-world evaluation of OpenAI’s AI Text Classifier shortly after its release, offering insights into its strengths and weaknesses in various scenarios. It demonstrates that the tool can be fooled by stylistic variations and paraphrasing, which is valuable for educators and content creators. The live testing with audience polls adds an interactive element that is uncommon in such reviews.

Pour aller plus loin :

  • AI Text Classifier — The official tool page, providing documentation and limitations.
  • OpenAI Blog — The official announcement and details of the classifier.
  • GPT-2 Output Detector — A predecessor detector, mentioned in the video as outdated.
  • QuillBot — A paraphrasing tool used in the video to test the classifier’s robustness.
  • AI detection in education — A general overview of AI detection methods and challenges.

132 words

Radar Profile

The radar profile shows a balanced but moderate performance across all dimensions. The video scores highest on quantity of information and practical value, but lower on technical depth and formal rigor, reflecting its informal, hands-on approach.

Reliability 6/10

💬 Sur les 0 commentaires analysés, aucune tendance n'est disponible.