- Bay Area Times
- Posts
- OpenAI, Snap talk GPT4-V-powered smart glasses, as Google, Apple explore options
OpenAI, Snap talk GPT4-V-powered smart glasses, as Google, Apple explore options
Bay Area Times
December 15, 2023
Brought to you by:
Top stories today:
- OpenAI, Snap talk GPT4-V-powered smart glasses
- AI toys voiced by Grimes launched at $99
- OpenAI's Superalignment team releases ’weak-to-strong’
- Balaji launches VC fund focused on network states
- Amazon leads in conversion rate
0. Data and calendar
All values as of 6 AM ET / 3 AM PT, other than S&P500 close (4 PM ET / 1 PM PT).
All times are ET.
1. OpenAI, Snap talk GPT4-V-powered smart glasses, as Google, Apple explore options
OpenAI recently discussed with Snap embedding GPT4-V into its Spectacles smart glasses. OpenAI doesn't have an internal team but is exploring working with partners, such as Jony Ive.
Google is also exploring Google Glasses 2.0. It is also planning an AI assistant called Pixie for its Pixel devices.
Apple is working on multimodal AI models for the Vision Pro, which launches in early 2024. Apple reportedly paused work on its AR glasses to focus on shipping the Vision Pro, but they could resume AR work at any time.
2. AI toys voiced by Grimes launched at $99
Plush toy that talks in a kid’s voice.
Parental app included.
The company started after an X exchange in April
This video shows the toy in action
Our view: this is a smart application of screenless conversational AI, more useful than the $699 Humane Ai Pin, for example.
Defensibility is not yet clear, but could arise from strong branding and network effects.
Harmonic provides market intelligence on 20M+ startups with tools designed to create better deal flow:
Discover new startups in any sector, geography, or stage (including stealth).
Prioritize companies consolidating top talent, rapidly adding headcount, and quickly growing their digital presence.
Engage founders through direct outreach to create better deal flow.
*Sponsored.
4. OpenAI's Superalignment team releases ’weak-to-strong’ model; segment still headed by Ilya Sutskever
Goal is to have weaker models be able to control stronger models, so that in the future, weaker humans can control stronger AI superintelligence:
