Google and OpenAI show off impressive AI voice assistants at rival events

AI-palooza… A day after OpenAI announced its new-and-improved ChatGPT model, Google yesterday hosted its AI-focused developer conference, which kicked off with an intensely high-octane concert to demo the company’s AI DJ (part of its text-to-music tool). Tech giants have been sprinting (much like this guy) to churn out AI innovations. Some highlights:

• OpenAI said GPT-4o is 2x as fast as GPT-4 and will soon be released for free. It features a sophisticated voice assistant that the company says can hold realistic convos and respond to what you show it through a camera (see the demos).

• Google previewed a similar voice assistant (dubbed Gemini Live) that stops talking when you interrupt it, remembers interactions, and analyzes your face and environment through the phone camera (demo hero). 

• Custom bots: OpenAI already lets paid subscribers create and browse customized chatbots in its GPT Store, and Google intro’d a similar option for Gemini subscribers to create personalized assistants (think: calculus tutor, sous chef, and “Yoga Bestie”). 

Siri has a sirius problem… The voice-chat capabilities previewed by OpenAI and Google seem to be leagues ahead of OG assistants like Apple Siri and Amazon’s Alexa, which’ve been known to say “sorry, I didn’t get that.” In addition to holding eerily realistic conversations, GPT-4o and Gemini Live can “see” through the camera (in a demo where OpenAI’s voice assistant is helping someone prep for an interview, the bot tells the user that they should fix their hair). But wait, there’s more:

• Not just talk: Google introduced AI trip-planning features, showing how Gemini can pull your flight and hotel bookings from Gmail and create customized itineraries (picture: suggesting highly rated restaurants and activities near your hotel). It said Gemini could also locate your shopping receipts in Gmail and fill out return forms.