OpenAI and Google have introduced exciting new features just in time for Halloween. OpenAI’s ChatGPT Advanced Voice, powered by GPT-4o, is now available to all ChatGPT Plus subscribers. This voice bot can engage in human-like conversations, picking up on non-verbal cues and responding with empathy and emotion. While it cannot sing or access the internet, it can laugh, whisper, and remember previous conversations. Users can choose from nine different voices to interact with.

Google, on the other hand, has launched NotebookLM, a tool that can create engaging and realistic podcasts. Users can upload up to 50 sources, including PDFs and links, and generate an audio overview. The podcast features two hosts who engage in light banter while exploring themes in the material. The result is an entertaining and insightful podcast experience.

These AI features are not limited to Halloween activities. ChatGPT Advanced Voice can be used for role-playing, language learning, and more. It supports 50 languages and can follow custom instructions for interaction. NotebookLM, on the other hand, can analyze various documents and create engaging podcasts on a wide range of topics.

While OpenAI and Google have not disclosed specific future plans, they are continuously working on improving their AI models. OpenAI’s upcoming models, o1-preview and o1-mini, are more advanced than GPT-4o and can self-reflect and correct themselves. Google is also rumored to be releasing a product called Jarvis, which can complete agent tasks in the browser.