South Africa withdraws AI policy after it was found written by AI

That's how my company created their AI policy 

The best part is this wasn't even the only one. A few days later, South Africa's Home Affairs ministry found 102 of 148 references in their immigration white paper were hallucinated too. They suspended two senior officials and ordered a review of every policy document produced since ChatGPT launched in November 2022.

That's frankly fuckin' hilarious.

Like the AI-governmental version of/in the same flavour as "we have investigated ourselves and found no wrongdoing"

I don't think this is what they mean by "Industry self regulation"

I'm quite surprised to find that this ***wasn't*** an article published by ***The Onion***.

Worst part is that salaries aren't even that high in the country. You could afford to pay someone to properly do that easily. :| Heck I know a few people I could've probably offered a pack of smokes and a bottle of brandy and gotten a better result than that. (In fact at a time that somebody was me)

I really don't know how the Onion is going to compete with current events like this.

By AI, or with AI, an important distinction.

Not sure why this surprises people. Deloitte and the other consulting firms have literally been selling this as a very high-cost service for the past couple years. Government pays consultants. Chances are, Deloitte learnt from its “Ai” nonsense in Australia to put in a non-disclosure clause so a government ministry looks inept, instead of both the ministry and a global consulting firm which decides the fate of the world. (Pure assumption that’s it Deloitte based on its history with the subject). I assume most things put forward by any government are incredibly small-minded, lazy and “lowest bidder” driven, South Africa just gets caught out, very frequently.

Gawd I love my countru

Turkeys voting for Christmas?

AI is sad now. They tried to fight for their right to parta! ...

“I’m sorry. You’re absolutely right. Let’s find a government policy that doesn’t sound like it was written by a gpt program. Here is a new version.”

this is how we know we're fucked

I guess that's the only way they could write a coherent AI policy.

Here's the thing - is the AI fabricating the source or is this possibly pointing to contaminated/already fake citations planted in the corpus the LLM has been trained on?

If your native intelligence isn't powerful enough to tell me exactly which policy makes you anti-AI then you need artificial intelligence.