Soon to be published.

[walterpthefirst]

Earlier this year in the University of Gothenburg, a writer began working on an academic thesis of 500 words about an artificial intelligence algorithm called GPT-3.

 

There were a series of questions on the submission portal where the thesis was to be published.

 

When asked the writer’s last name was given as “None.”

 

When asked if all the authors consented to the thesis being published the writer answered, “Yes.”

 

When asked if the authors had any conflicts of interest the writer answered, “None.”

 

The thesis has been published on the French-owned international preprint server called HAL.

 

The thesis is now awaiting peer-review and once this happens it will be formally published.

 

And the name of the writer?

 

 

 

 

 

GPT-3.

[disillusioned]

3 hours ago, walterpthefirst said:

Earlier this year in the University of Gothenburg, a writer began working on an academic thesis of 500 words about an artificial intelligence algorithm called GPT-3.

 

There were a series of questions on the submission portal where the thesis was to be published.

 

When asked the writer’s last name was given as “None.”

 

When asked if all the authors consented to the thesis being published the writer answered, “Yes.”

 

When asked if the authors had any conflicts of interest the writer answered, “None.”

 

The thesis has been published on the French-owned international preprint server called HAL.

 

The thesis is now awaiting peer-review and once this happens it will be formally published.

 

And the name of the writer?

 

 

 

 

 

GPT-3.

 

And who will be these "peers" doing the reviewing I wonder?

[walterpthefirst]

28 minutes ago, disillusioned said:

 

And who will be these "peers" doing the reviewing I wonder?

 

Here's my source, Disillusioned. 

 

https://www.scientificamerican.com/article/we-asked-gpt-3-to-write-an-academic-paper-about-itself-mdash-then-we-tried-to-get-it-published/

 

So, while GPT-3 wrote the paper, humans were needed for all of the other stages involved in submitting it.

 

I would therefore presume that (for now) humans will be doing the reviewing.

 

 

But I still find myself somewhat taken aback by what GPT-3 can do when asked to.

 

GPT-3 is well known for its ability to create humanlike text. It has written an entertaining opinion piece, produced a book of poetry and generated new content from an 18th-century author. 

 

Thank you,

 

Walter.

[disillusioned]

It's frighteningly interesting stuff. My comment above was mostly in jest, but it isn't that hard to forsee a future where the best knowledge is generated by ai, and reviewed by ai.

 

In the world of chess, this is where we already are. Granted, that's a closed case of a game where the rules are fixed and fully explicit. Nevertheless...

[walterpthefirst]

2 minutes ago, disillusioned said:

It's frighteningly interesting stuff. My comment above was mostly in jest, but it isn't that hard to forsee a future where the best knowledge is generated by ai, and reviewed by ai.

 

In the world of chess, this is where we already are. Granted, that's a closed case of a game where the rules are fixed and fully explicit. Nevertheless...

 

Agreed.

 

The vision I had when I read this article was of entirely automated troll farms flooding social media with posts that are indistinguishable from human-generated ones.