OpenAI's GPT-4o is the latest technically impressive Torment Nexus

OpenAI's latest hot shit is GPT-4o, which in the demos on their blog at least, is basically Scarlett Johansson's character in the movie Her. The dorks at OpenAI thought that's something they should build, reinforcing the meme that nerds keep missing the point of dystopian fiction. Anyway, while the demo makes me feel uneasy, it is impressive at face value. GPT-4o's voice mode can respond to audio inputs "in as little as 232ms, with an average of 320ms, which is similar to human response time in a conversation". GPT-4o differs from GPT-4 as it is a "a single new model end-to-end across text, vision, and audio, meaning that all inputs and outputs are processed by the same neural network", rather than converting audio to text like GPT-3.5/4. GPT-4o's text and image features can be used now in ChatGPT (even in the free tier) and via the API, with the audio and video features coming to a "small group of trusted partners in the API" and ChatGPT Plus "in the coming weeks". OpenAI also announced a Mac app for ChatGPT. The full 26 minute video of OpenAI's Spring Update is on YouTube.