Google’s Bard AI makes stuff up in presentation
Google’s new Bard AI was presented in a fairly hefty blog post yesterday and it contained some made up information… in this case that third item in the example is demonstrably false. This is from an animation on the blog post announcing it (about 1/3rd down the page).
Results were that Alphabet’s stock price declined from $107.64 on $Tuesday to 98.78 as of this writing (over a 100 billion USD market value loss)… which is odd because ChatGPT tends to do the same thing… make stuff up. We’re building a set of AIs that lie to us like 7yos who snuck a cookie and expecting perfection. A simple “cite your sources” would be useful.
I’d say if anything this demonstrates Google’s Bard is on the level of ChatGPT/Bing. Easily spotted AI with fake information. I mean, if I had a lot of money at this point I’d invest it in Alphabet because I’m pretty sure if they get this out there right it’ll be what the AIs write blog posts about.
[9to5Google]
