Did Artificial Intelligence think and Write it?

Authored by Alefiya Neemuchwala Oct 13^th 2026

Over the last five years, Artificial Intelligence tools have become embedded onto daily use and boost quality and productivity. So now, what happens to be creative, original and critical thinking. Other think that AI is a tool that will improve humanity and provides a tool to all to compete with others. Yes, in a office, a busy manager, can now organize thoughts within seconds, prepare presentations on the GO and also do performance review. Similar trends are seen within scientists, budding writers and students. In fact, many managers, teachers and principal investigators want to use AI. While, use of AI as an assistant is acceptable but now the question comes- who came with the AI. Are their tools that can check if this was an original idea or AI-generated idea?

As AI is increasingly used as an assistant, it was often observed that AI-assisted manuscripts/ work would often drop or change the original idea of the humans. Are their tools now to check and evaluate if a given manuscript is AI-generated, AI-assisted and to what extent was the idea originated by AI or AI-polished human idea?

I researched some of the tools that exist and I am sharing them here. Some very popular tools include ProseLens, Pangram, EditLens and IdeaLens. IdeaLens is particularly interesting. A paper released October 5, 2026, proposes identifying whether the ideas and organization of a document originate from AI, even when a human writes the final sentences. It represents documents as structured outlines and examines those patterns. Its experimental results suggest that merely rewriting AI-generated ideas in different words may not eliminate detectable signals. ProseLens evaluates prose-level authorship signals. It is presented alongside IdeaLens, allowing researchers to examine differences between the provenance of ideas and of wording. EditLens works by not only looking only for fully AI-generated material, it estimates the extent of AI editing in a document. The developers published a study reporting an F1 score of 94.7% on a binary classification task and 90.4% on a three-category task. Pangram is a commercial detector with research focused on identifying AI-generated and mixed human–AI text. Its July 2026 technical report reported a 0.0041% false-positive rate on its benchmark, although real-world accuracy depends on the type of text and testing condition.

Other AI detectors that are notable include GPTZero, which claims to detect both AI-generated as well as AI-assisted writing. Other tools include CopyLeaks, OrginalityAI, turnitin, Zerogpt.

While these tools are great, they suffer a major limitation. The developers acknowledge that they did not independently evaluate if the documents that their studies investigated were developed by humans or machines. Nor did they create their own datasets which contained confirmed human developed manuscripts or machine generated, or where datasets originated from humans vs machines.

This is a very gray zone and there is no black or white. While the ethical discussions on the use of AI is ongoing, it is very true that AI usage is here to stay and those who dont use face more hardships? Will you use AI in your next publication? Care to comment on?