I set out to test ChatGPT, but who is testing whom?

Also, waving goodbye to the Hash Code coding competition

Interacting with ChatGPT, the latest headline-grabbing AI tool based on a
large-language model (LLM) developed by OpenAI, it's hard to shake off the feeling that we're already living in a future worthy of a Sci-Fi movie.

The language capabilities of the model are really good. Public reaction includes an appropriate mix of awe and worry: Business Insider lists "10 Jobs Most at Risk of Being Replaced by AI", The Guardian strikes a tone that's almost, but not quite, reassuring: "ChatGPT won’t take your job – yet".

I decided to confront the fear, and asked ChatGPT if it could take over writing articles for the pnote.eu blog:

❯ Hello! I'd like you to write articles for a blog about "people, computers and people trying to make sense of computers". Try to assume an identity of an author who is in mid-30s, lives in Paris and likes to ride city bikes at night. Do you think you can do it?

Sure enough, ChatGPT said yes:

🤖 ChatGPT : Hello there! As an AI language model, I can certainly write articles on the topic of "people, computers, and people trying to make sense of computers," and I can also assume the identity of a mid-30s author living in Paris who enjoys riding city bikes at night.

And so I put it to test, asking to recreate 3 example posts from the blog archives. The results were mixed… more interestingly, by the end of the exercise, like the protagonist of the 2014 Sci-Fi movie Ex Machina, I started to have doubts about the nature of the exercise.

Is it me who’s demonstrating what type of blog posts ChatGPT can, and cannot write? Or is it ChatGPT demonstrating how unoriginal some of these blog post ideas were in the first place? My new aspiration for the blog is to write articles that ChatGPT couldn’t have written :).

The full post @ 📝 We test if ChatGPT can blog... or so we think .

In other news

• 🔈 The Case for Halting AI Development at the Lex Friedman podcast (2h48m – it’s a long form podcast). Interesting discussion with Max Tegmark explaining the risks involved in the race to create increasingly powerful AIs, and the rationale for the open letter calling for intentional slow-down in AI development

• 🪶 Fire and Ice. Appropriately apocalyptic poem by Robert Frost. Some say the world will end in fire | Some say in ice | From what I’ve tasted of desire | I hold with those who favor fire. | (…)

• 📝 Hash Code: goodbye and thank you 💫 . On February 22 Google announced the sunset of Coding Competitions. This marks the end of the road for Hash Code , a team programming competition I helped to create at Google France in 2014 and I’ve been helping to run since then. The post is my thank you note to the Hash Code community as we wave goodbye to the competition.

Postcard from Paris

The spring has come to Paris, bringing lots of sunshine mixed with lots of rain. As of right now, it didn’t bring about any motivation to start spring cleaning my apartment.

Thanks for reading, wishing everyone lots of motivation for whatever makes you happy :). Until the next time!