This week’s news roundup is one day late, owing to the fact that I was in Spain on a much-needed (and much-appreciated) holiday. Now that I’ve eaten my fill of tapas, and drank my fill of vermut, let’s break down a really very newsy week in tech.
The lead: Musk v. OpenAI
The big story this week is definitely Elon Musk suing OpenAI, the company he put the first money into back when it wasn’t a company at all, but a nonprofit with a mission of funding the development of artificial general intelligence out in the open. He claims in the lawsuit that OpenAI’s longstanding partnership with Microsoft compromises its original purpose of developing AGI out in the open.
The fact that it involves A) Elon Musk and B) OpenAI is what makes this the biggest newsmaker this week, but beyond the big names and noisy drama, there isn’t much substance to the case — it reads more like sour grapes or FOMO on Elon’s part than anything else.
One interesting tidbit is that Musk claims OpenAI is actually actively engaged in not just developing, but refining an artificial general intelligence in order “to maximize profits for Microsoft” in the specific language of the lawsuit. While ChatGPT and GPT-4 are impressive, no one in the know would likely classify them as true AGI, so it’s possible Musk is alluding to some other unreleased effort under development. If so, we may get a chance to peek behind the curtain at OpenAI a bit should this lawsuit proceed.
Big funding for OpenAI-powered real robots
A company called Figure has raised a $675 million Series B round of funding, valuing the startup at $2.6 billion. The funding comes from a number of the biggest names in tech and AI
OpenAI’s Startup Fund (which we’ve previously noted is actually just Sam Altman’s fund on paper), Microsoft, Nvidia, Jeff Bezos and others participated in the round
Alongside the funding, Figure also announced it signed a deal with OpenAI to develop next-gen AI models designed for use by humanoid robots
My old colleague Brian Heater has more on the deal on TechCrunch, including a brief history of the very young robotics company bootstrapped by serial entrepreneur Brett Adcock
The prospect of large AI models powering robots with the ability to manipulate the real world drives new implications for the phenomenon of AI-generated hallucinations
Apple finally kills its car
Apple has cancelled its project tot develop a vehicle, according to the extremely well-sourced and consistently reliable Mark Gurman of Bloomberg
The cancellation also involves moving some team members working on automotive tech to Apple’s AI group, which tracks since part of the Apple vehicle project was focused on developing autonomous vehicles — another sources of major AI work
Apple had been working on a car since at least around 2015 or so when news of the project first made its way out to the press. The iPhone maker had also been testing prototype vehicles with its autonomous driving technology on board on roads in California
The project is thought to have cost $1 billion+ in R&D investment, and went through various high-profile staffing and priority changes over the years
Killing the car project doesn’t mean all that money is wasted, however; Apple rarely invests in technology development without at least thoroughly considering all the other applications said tech might have in other areas of its business
Google’s Sundar Pichai says company “got it wrong” on AI bias
An internal memo that made its way to Semafor’s Reed Albergotti details Google CEO Sundar Pichai’s feelings about the embarrassing problems its Gemini image generation tools ran into over the past couple of weeks
Pichai flat out admitted Google “got it wrong” when it came to training the tools to espouse a multicultural approach when it came to depicting subjects in response to queries — even when it made absolutely no sense to do so, like providing pictures of multiracial “Nazis”
Pichai said that teams at google are “working around the clock” on fixes and have already seen “substantial improvement” on some related prompts
This is just an example of the speed at which the AI generation market is moving, and the massively daunting task facing even the most well-resourced players in this space of avoiding tripwires in this brand new emerging market
Nest co-founder’s superspeed composter gets an upgrade
In a rare bright spot in actually novel consumer gadget news, Matt Rogers’ new startup Mill released a new generation of its sleek, kitchen-friendly compost bin
The new Mill can now process a pound of wet food scraps in just 2.5 hours, which the company notes is actually shorter than most dishwasher cycles. In total, it can process 40 pounds in total before it needs to be emptied, which translates to “weeks” of use
Like the Nest thermostat before it, Mill has software smarts to optimise run times for energy efficiency and your specific use patterns
The company claims 1 million pounds of waste diverted from a landfill to date from existing users, which is huge for just around a year of active operation
Given Nest’s impact on home energy use, and Mill’s impact on waste, I’d be hard-pressed to name another founder who’s had more of a cumulative direct impact on consumer ecological footprint
I love the Mill composter!!!