An example of model collapse in an image-generating AI. (Credit: M. Boháček & H. Farid / arXiv)
Generative AI exploded in popularity when OpenAI released ChatGPT.
A paper published in Nature looked at what happens when AI is trained on “synthetic data,” or content created by an AI rather than humans.
Flaws in the synthetic data led to even more mistakes in the AI’s output, a result dubbed “model collapse” by the researchers.
Sign up for the Smarter Faster newsletter
A weekly newsletter featuring the biggest ideas from the smartest people
Generative AI has been around for decades, but the systems exploded into the public consciousness in 2022, when OpenAI released ChatGPT, an AI chatbot that could produce remarkably human-like text.
The AI gained this ability by analyzing a lotof text created by people, mostly pulled from the internet. To put it simply, from this data, it learned to predict what word was most likely to come next in a sequence based …
