6 ms·
I asked bard the other day about the prognosis for a 80 year old with a liver infection. It responded super confidently about a study which suggested the five y
by VBprogrammer 3y ago
I asked bard the other day about the prognosis for a 80 year old with a liver infection. It responded super confidently about a study which suggested the five year survival rate was about 50%. I asked it where I could find the study and sure enough it gave me a link to a study. It almost had me until I clicked on the link and realised the study had nothing to do with a liver infection!
- CharlieDigital 3y agoWe found similar issues with asking Bard for images. Bard generated Wikimedia links which look legit, but it's simply following the pattern of those URLs. When we loaded the links, they were all broken.
- earthboundkid 3y agoAsk ChatGPT to draw with ASCII art. It's very funny.
- sebzim4500 3y agoThis was its attempt at drawing a guy with a balloon ^ /|\ | / \ O I guess technically I didn't specify that the balloon was lighter than air, or that the guy should be holding it. I say it passes on a technicality.
- tazjin 3y agoYou also technically didn't specify that the guy is not one of the monsters from Silent Hill. Agreed, it passes.
- earthboundkid 3y agoAsked Claude to draw a dog: ____ / \ | .'.' |__ \ (___)\\ '.___.' (___/ \_____ Anthropic is making the world safe from an AI that's such a good artist that we all join its cult.
- esalman 3y agoSame experience with ChatGPT tbh. I was researching a topic, asked for some references. It generated a few. The links were all valid and the author list was the same. Only ChatGPT modified the title of the papers to match my topic of research, so basically the actual references had nothing to do with it. tbh given how language models work this is not surprising.