manton
manton

I asked Meta.ai to imagine Austin in the 1920s, in watercolor. It has a lot of mistakes that a human artist would never make, but I wish Austin still looked like this. Bring back the trolleys!

|
Embed
Archimage
Archimage

@manton That’s the thing with imagination. It’s subject to interpretation. What you wanted was ‘your’ image of Austin in 1920. If I were an AI, I would have produced an image of Steve Austin ($6M Man) in his 20s. ;) Does that mean the AI has an imagination? No. But it does interpret based on its learning biases.

|
Embed
In reply to
manton
manton

@Archimage Ha! It is really fascinating to think about.

|
Embed
Archimage
Archimage

@manton If I make someone think, I have succeeded. ;)

|
Embed
Archimage
Archimage

@manton What happens if you ask an AI to generate a picture of Steve Austin? Whether it uses the character, or the wrestler tells you about the biases of the AI’s developers.

|
Embed
jonesbp
jonesbp

@manton A history project I worked on a few years ago has a lot of photos of Austin in the 1930’s… local-memory.org/athens-on...

|
Embed
renevanbelzen
renevanbelzen

@manton This AI generated image supposedly took as much energy as charging the depleted battery of your phone. We really need analog computing to make genAI greener.

|
Embed