Gregory's ServerMicrosoft paid money for this. A lot of money. And they gave it to us for free.
 16 comments
I'm looking at a demo of this paper right now, which is kind of interesting - https://arxiv.org/pdf/2005.11401.pdf - but... it relies, the same way most AI models do, on a tectonic amount of human curation effort that's gone on behind the scenes to make it work. I mean, it's nice I guess, and there's some nice features in a low-K-threshold, high-quality-training-data situation, but it sure looks like this will all fall apart if you point it at large, unvetted or adversarial data sets.  @mhoye And I bet this works for other basic-ass stuff too. Anything where the conspiracy content outweighs the debunking content is going to have the possibility of the conspiracy stuff seeming more plausible simply because that's the only content written in response to a dumb-ass question very few people ask.   @mhoye they're working off a wonky base. I don't know who uses bing typically, but I typed 'does...' and here's the auto suggest options... @toychicken @mhoye "Does the dog die" is a question that people ask about movies. Too much "Old Yeller" trauma. So that one makes sense.  @mhoye "does Australia exists?" no, according to #microsoft #ai in #bing   @mhoye *cringes in works for a Microsoft-dependent public service* :blobgrimace: @mhoye Sadly, it seems this entertaining fumble does no longer exist. Can't reproduce. |
@mhoye
"Well yah, you're the pilot and prompter now that you mention it but we provide a lot of value, look at how believable and plausible we sound. It's almost like it's true." - Microsoft probably