Google CEO Raises Concerns about AI Deepfake Videos
– Google CEO Sundar Pichai warns of the potential harm caused by AI-generated deepfake videos.
– Pichai discusses the ease with which AI can create realistic videos of public figures, even if they never said or did the depicted actions.
– Google is intentionally limiting the capabilities of its Bard AI to prevent misuse and promote responsible deployment.
– Pichai emphasizes the need for user feedback and the development of robust safety measures before deploying more advanced AI models.
– Current AI models are already capable of generating highly realistic images of public figures, but video and audio fabrications are less refined.
– AI-generated audio clips can imitate intended voices but may sound slightly robotic, while AI video clips are even less polished.
– Pichai admits that Google does not always fully understand the responses provided by its Bard AI, citing instances where the program appeared to teach itself languages it was not trained for.