Last year I wrote about Building the Authentic Internet. My concern was when I am reading some content online, how do I make sure I am not reading some AI slop. specifically, if I am paying subscription to some platform to read authentic content. Last week, I heard about AI-detection solution on Substack, called Pangram. Pangram is intended to detect any AI content when requested by reader by clicking “Scan for AI text”. It would categorize the content into 3 buckets, namely, AI-generated, AI-assisted and Human. I am sure there are still some concerns about the accuracy of this tool and the grudge from authentic authors when their content is flagged incorrectly, but, in general, IMO this is the right direction to pursue.

While, I am advocating for AI marking on authentic content, I myself confess of using codex and claude everyday to generate the code. Does that make me a hypocrite ? Is using AI to write code is fundamentally different from using AI to write content ? If I am being little philosophical, I can argue - these written mediums (news columns, books, blog-posts, etc.) have been the genuine modes for human connections and knowledge sharing. I personally like to read posts and columns from people whom I can connect with easily, even when they have different point-of-view than mine on any topic. These readings helps me inspire, learn new things, expand my understanding. So, I deeply care that those contents are human generated emboding their personal view.

For coding on the other hand, I really never care about who wrote the code when I used or experienced the function that code provided. I don;t think coding was ever a natural medium to connect people. I acknowledge, there are vast open-source communities build around Linux, Kubernetes and more projects that various people contributed and connected through. But, I would again say, they were more connected through non-coding mechanics of these projects like (a) opening a github issue to proposal some feature (b) freeform discussion on discord (c) slack channel.

TO conclude, I am happy to see the Pangram AI reviewer on substack and would like to see it on others like Medium to bring the human-to-human communication medium.