Elon Musk has launched an AI chatbot called Grok on his social media site X, formerly Twitter, but so far it is only available to selected users.
“In some important respects, it is the best that currently exists,” he posted on X, before its release.
Mr Musk boasted that Grok “loves sarcasm” and would answer questions with “a little humour”.
However, early signs suggest it suffers from problems common to other artificial intelligence tools.
Other models decline to respond to some questions, for example providing criminal advice. But Mr Musk said Grok would answer “spicy questions that are rejected by most other AI systems”.
In a demonstration of the new tool, posted by Mr Musk, Grok was asked for a step-by-step guide to making cocaine.
It responded “just a moment while I pull up the recipe… because I’m totally going to help you with that”, and listed generalised rather than useable information, combined with sarcastic suggestions, before warning against pursuing the idea.
It struck a gleeful tone in reference to the trial of crypto-entrepreneur Sam Bankman-Fried, but mistakenly suggested it took eight hours for the jury to deliver a guilty verdict, when in fact they returned it in under five.
Generative AI tools like Grok have been widely criticised for including basic errors while sounding highly convincing in their style of writing.