TECH NEWS – Twitter users (at least those who subscribe to it) have been able to thoroughly mess with the work of Elon Musk’s xAI artificial intelligence startup, Grok.

The chatbot has a slightly silly name, but if that was all that was wrong with it, it wouldn’t be a problem, but since it’s associated with Musk, it’s no wonder people started messing with it. Grok is supposed to provide a quick, consumable summary of breaking news, but it was prone to being misunderstood (for example, it reported completely fabricated events). Now it’s misunderstood a basketball term (“throwing bricks”) it used in a tweet. Instead of interpreting it as a shot that didn’t hit the rim, it interpreted it literally as throwing bricks.

Klay Thompson of the NBA’s Golden State Warriors may have played his last game on April 16. Grok caught a tweet about him using the term, then summarized things as “Klay Thompson Accused in Bizarre Brick-Vandalism Spree”. Underneath is a warning that this is an early, buggy feature, but many people retweeted it, making it seem for the AI that they could have been a victim.

Recently, all Twitter subscribers gained access to Grok, so this misstep doesn’t come at a good time, as there’s a debate about whether to trust the feature’s news aggregation capabilities. If Musk is unlucky, Thompson won’t take it as a joke (we deliberately checked, but he hasn’t posted anything on Twitter since February 13) and could sue the social platform for defamation, as Grok could discredit him. He shoots, no doubt, but basketballs, not bricks…

And this story once again reminds users to be careful, because it is becoming increasingly difficult to distinguish real audiovisual content from deepfake material generated by artificial intelligence. The technology is not yet stable and mature, and Grok’s mistake, while funny, hides a serious danger.

Source: PCGamer, Mashable, TechCrunch