The demo of Galactica generated fake research and filtered out searches for ‘racism’ and ‘AIDS.’
Researchers developed a rival adversarial AI to trick the fearsome KataGo model into losing games.
The Stable Diffusion Explorer shows how words like "assertive" and "gentle" are mapped to sexist stereotypes.
Essays written by AI language tools like OpenAI's Playground are often hard to tell apart from text written by humans.
Yann LeCun, machine learning pioneer and head of AI at Meta, lays out a vision for AIs that learn about the world more like humans in a new study.
“Loab” is creepypasta for the AI generation, and nobody understands where she came from.
The company warns users that BlenderBot 3 will make ‘untrue and offensive statements.’ It’s also just not very good.
AI observed videos of lava lamps and inflatable air dancers and identified dozens of physics variables that scientists don't yet understand.
Amateur urban planners are using DALL-E to visualize what cities might look like if they were built for pedestrians and cyclists, instead of cars.
With a short text description, the tool can create surreal sci-fi environments and stylistic fine art—for a price.
Deepfakes are no longer allowed on Google Colab.