I posted on this thread earlier too: Hi! I am the person that wrote this, and used GPT-3 and a few other writing tools to help me wordsmith it. All the points, however, is original work and not AI generated. I am not a native English speaker, so I have been using these tools to avoid awkward sentences/paragraphs. (Clearly this has not been the outcome I was hoping for)

As for the erroneous citation to Forbes: there were two links, one to Forbes and the other to Digiday to backup the next point in the outline I wrote. While transferring the content from Huggingface to the substack editor, I missed that in the proofreading.

So why delete my comment on the publication instead of addressing it? Then why delete my second comment? And why not make it clear the article is generated by ChatGPT, at least partially?

Also, this is a good example on why anonymous accounts writing content will be trusted less. You say you used ChatGPT for parts of it. Unclear on what “parts” mean and how much input you had, versus what the AI wrote.

It’s a reason for people to stop reading anonymous authors, or articles that don’t make it clear that it’s not an AI writing part of the article.

I’ll be honest: I feel duped reading a wall of text to realise it’s at least partially generated, and this whole article could have been the prompts you used to generate it.

I am not as comfortable having a back and forth on substack than I am on HN. I am not sure as to why people who use tools like [1], [2], [3], [4] have to declare as such that the parts of the content may have been edited with help from computer tools if the end goal of conveying the information has been satisfactorily achieved. This seems like a rehashing of the "AI generated art is not real art" debate [5] again.

[1] https://huggingface.co/EleutherAI/gpt-j-6B [2] https://huggingface.co/docs/transformers/model_doc/bloom [3] https://www.jasper.ai/ [4] https://openai.com/api/ [5] https://www.nytimes.com/2022/09/02/technology/ai-artificial-...

We're on HN now.

Why have you deleted every comment on your post mentioning GPT-3? There were at least 4 or 5 comments mentioning it and you've deleted every one.

I mentioned, I am not as comfortable having a back and forth on substack than I am on HN. That is why I deleted the comments on substack, as they were not relevant to LLMs (or "GPT-3" as you like to call them)

> as you like to call them

Just using the same language as the deleted comments.

Not sure about the others, but for [4], it's in their ToS:

https://openai.com/api/policies/sharing-publication/#content...

I really enjoyed this exchange.

- This looks machine generated

- Hi! I'm the person who wrote this by using a machine to generate 'the words' but the important parts are original

- Oops, didn't cite my/its work

Especially in the context of an industry of content creation falling over.

Thank you both!

> This all underscores how it’s becoming hard to trust anonymous authors even now, and how this will just get worse.

What is the next step then? A more formal reputation system? How long until reputation is just another traded commodity?

Thanks for the updates, can you add another update with the author's response? Whether you agree with the argument or not it's relevant they have shown face and have justified their actions.