OpenAI's real target isn't search engines; it's personal assistants
Today, OpenAI announced GPT-4o, a faster, freer model that "can reason across audio, vision, and text in real time.” In other words, you can talk to it and show it stuff and it will respond in kind. Supposedly!
While some text and image capabilities are launching today, the rest will be “rolled out iteratively.” Its demos are cool — it can describe what’s happening in a room, from what a person is wearing to the lighting or help you order a replacement iPhone — but limited.
There had been lots of speculation as to what exactly OpenAI was going to debut this week, including that the ChatGPT maker was going to launch a rival Google search engine. But this appears to be a play for a Google Assistant killer rather than a Google Search killer. (The Information had previously reported that OpenAI was developing a voice assistant.)
It’s what Meta’s Ray-Ban smart glasses and Rabbit r1 (which may basically just be an Android app) are going for but haven’t quite reached. As we’ve reported, voice assistants have seemingly stagnated as Big Tech companies try to transition them from natural language processing to generative AI capabilities.
It’s tempting to think this could be the holy grail when it comes to human interactions with computers — something that’s easy and intuitive. But that’s only if actually it works. We’ll see.
There had been lots of speculation as to what exactly OpenAI was going to debut this week, including that the ChatGPT maker was going to launch a rival Google search engine. But this appears to be a play for a Google Assistant killer rather than a Google Search killer. (The Information had previously reported that OpenAI was developing a voice assistant.)
It’s what Meta’s Ray-Ban smart glasses and Rabbit r1 (which may basically just be an Android app) are going for but haven’t quite reached. As we’ve reported, voice assistants have seemingly stagnated as Big Tech companies try to transition them from natural language processing to generative AI capabilities.
It’s tempting to think this could be the holy grail when it comes to human interactions with computers — something that’s easy and intuitive. But that’s only if actually it works. We’ll see.
