OpenAI AI Text Classifier
å€éšOpenAI's AI Text Classifier was a free beta tool launched in January 2023 to detect AI-generated text versus human-written content, providing likelihood categories such as 'very unlikely' to 'likely AI-generated.' It offered a simple paste-and-submit interface for quick feedback but was discontinued in July 2023 due to low accuracy, identifying only 26% of AI text correctly while flagging 9% of human text falsely. This early detector highlights ongoing challenges in AI content verification, making it relevant for understanding detection limitations though not for reliable current use.
説æ
OpenAI's AI Text Classifier was a free beta tool launched in January 2023 to detect AI-generated text versus human-written content, providing likelihood categories such as 'very unlikely' to 'likely AI-generated.' It offered a simple paste-and-submit interface for quick feedback but was discontinued in July 2023 due to low accuracy, identifying only 26% of AI text correctly while flagging 9% of human text falsely. This early detector highlights ongoing challenges in AI content verification, making it relevant for understanding detection limitations though not for reliable current use.
äž»ãªæ©èœ
- Classifying text as AI-generated or human-written
- Providing likelihood categories (very unlikely, unlikely, unclear, possibly, likely AI-generated)
- Analyzing English text with at least 1000 characters
äž»ãªçšé
- 1.Supplementary feedback on text authenticity
- 2.Quick checks for writers and casual users
- 3.Educational awareness of AI-generated content
OpenAI AI Text Classifier ã¯ããªãã«åã£ãŠããŸããïŒ
ããããã®çšé
- Casual users seeking free, quick supplementary checks on long English texts
åããŠããªãçšé
- Educators or academics needing reliable high-stakes detection
- Non-English speakers or users with short/code text
- Anyone requiring accurate primary AI text detection tool
éç«ã£ãç¹åŸŽ
- Free to use
- Simple web-based paste-and-submit interface
- Trained on diverse human and AI texts from multiple models
ã¬ãã¥ãŒ
0 ã€ã®ãã©ãããã©ãŒã ã«ããã 0 ä»¶ã®ã¬ãã¥ãŒ ã«åºã¥ã
ãŠãŒã¶ãŒãã£ãŒãããã¯ã®ãã€ã©ã€ã
æãé«ãè©äŸ¡ãããç¹
- Free and accessible with easy interface
- Useful for supplementary, non-critical feedback
- Trained on broad dataset including various AI models
ããããäžæº
- Low accuracy: 26% true positive rate on AI text
- 9% false positive rate on human text
- Unreliable for short texts, non-English, code, or edited content
- Discontinued July 20, 2023 due to poor performance