Oh, Bard…
Daily Photo – Oh, Bard…
I love playing around with new AI stuff. Some of it is great (Midjourney/ChatGPT) some of it is terrible and sometimes it’s really hit and miss. Regardless of the result it’s always fun to experiment. Recently Google added image uploads to Bard, their AI chat, and you can have it do things like describe the contents of the image.
Now knowing Bard had a tendency, even with normal chat, to invent nonsense facts I was curious to see how it would do with some of my images. To be fair, it started off promising, but then… well… it all got a bit “imaginative”. So, here’s one example:
“The image you sent me shows an aerial view of a city with tall buildings and a yacht in the middle. The city is located on a peninsula, and the water on the left side of the image is a bay. The buildings in the city are mostly tall skyscrapers, and there are a few smaller buildings as well. The yacht in the middle of the image is white, and it has a blue stripe on the side. There are a few people on the yacht, and they are standing on the deck.
The sky in the image is blue, and there are a few clouds. The sun is shining, and the light is reflecting off of the water in the bay. The overall impression of the image is one of a modern city with a beautiful skyline.
Here are some additional details about the image:
The image was taken during the day, in the afternoon.
The weather is clear and sunny.
The water in the bay is a light blue color.
The sky is a clear blue color.
The city is surrounded by mountains.
I hope this description is helpful.”
So the question is… is Bard “imagining” what is outside the image or is it picking up this being Hong Kong and therefore it’s describing aspects of that location, even though they are not in the frame… hmm…
