The capabilities of large-scale pre-trained AI models have recently skyrocketed, as demonstrated by large-scale vision-language models like CLIP or ChatGPT. These typical generalist models can perform reasonably well in tasks covering a large variety of fields, which has paved the way for their widespread adoption by the public. Pretrained large-scale AI models need to ‘forget’ specific information for privacy and computational efficiency, but no methods exist for doing so in black-box vision-language models, where internal details are inaccessible. Now, researchers addressed this issue through a strategy based on latent context sharing, successfully getting an image classifier to forget multiple classes it was trained on. Is AI development becoming dangerous for humanity ?
