daily.dev

Nimrod Kramer

Codex

Community Picks is a section on daily.dev where our community members share the most interesting and valuable content they've discovered online. From insightful articles to handy tools, every post is a gem curated by our dedicated coomunity. To contribute to Community Picks, you need to have at least 250 reputation points, ensuring that only active and trusted members can share their finds.

Community Picks

The AI detector uses Large Language Models to analyze patterns commonly used to generate text using AI. It has an accuracy rate of over 99%. Use the Wordtune Rewrite tool to rephrase any parts of the text that were marked as AI-generated.

Free AI Content Detector for GPT-4, ChatGPT & Other Tools

The official daily.dev AI community. Led by thought leaders and AI experts. Stay tuned for insights, news and discussions. 

"AI21Labs has a new tool to detect AI-generated content! What are your thoughts on this tool? How can such it be used to serve us better?

&quot;AI21Labs has a new tool to detect AI-generated content! What are your thoughts on this tool? How can such it be used to serve us better?

Snake oil, can’t be done with text.
I ran their own website marketing copy through their own tool and it says it is 100% AI generated.
<img src="https://media.daily.dev/image/upload/s--_kAlKU8w--/f_auto,q_auto/v1/ugc/content_e84bf780-6d15-4bbe-a3a4-4fceb72b95d4" alt="image">
They also then say you can use their tool to paraphrase it implying that will then not be detected as AI generated text? Strange.
Tools like this also have a serious problem; false positives. One FP is too many, it can ruin a career, have a student expelled.

Absolutely Not Useful.
I have tested wordtune with different chat gpt generated text.
Always wordtune respond: generated by a human.
<a href="https://vvv.code.blog/2023/06/24/chat-gpt-prompt-a-story-about-homeless-people/" target="_blank" rel="noopener nofollow">Link to the generated text</a>
<img src="https://media.daily.dev/image/upload/s--QJMKDWDJ--/f_auto,q_auto/v1/ugc/content_565336fe-54f3-4f4e-9bc0-9c975c32640a" alt="wordtune">
Here is a screenshot of an example.

For text I don’t think that is possible! Even OpenAI gave up on their own tool to do that.
But for Image, Video and Audio that’s doable and necessary

False positives/incorrect AI text detections are concerning – especially when genuine human content gets misclassified as AI-generated, the stakes are high. For Example such mistakes could:
Influence Google’s content quality assessment, potentially acting as (something like) a ranking factor
Put students in academia at risk of facing repercussions for wrongly identified AI-generated content or perceived plagiarism
Even a small false positive rate has significant repercussions because implications are vast and far-reaching
Given the complexity of variations in models and systems, it is challenging to continually reassess, adapt and reevaluate test methods and procedures. Consistent and reliable results are elusive, especially when progress in one area counteracts progress in another.