OpenAI

Here’s what Eric Schmidt, who was Google’s CEO from 2004 to 2011 and then chairman until 2015, had to say recently during a talk at Stanford:

Google decided that work-life balance and going home early and working from home was more important than winning. And the reason startups work is because people work like hell.

Update, August 14th: It appears that Schmidt didn’t intend for his comments to make headlines! Stanford has taken the video of his talk down, so here’s a clip that is still online:

For more of the backstory and ambition behind Worldcoin, the eyeball-scanning-for-cryptocurrency startup that Sam Altman thinks could one day save us all from an AI-controlled world, check out this piece from Bloomberg’s Ashley Vance:

Blania and Altman, who formally outlined their Worldcoin master plan a year ago, have since received feedback that might be generously described as mixed. On one hand, they’ve already persuaded more than 6 million people to go before an Orb and sign up for a World ID, and the sign-up rate has been surging this year. The total value of the digital currency (WLD) is more than $550 million. At a factory in Germany, Orbs are heading toward mass production and will soon be dispersed around the globe in a bid to push these numbers even higher.

The startup has appointed Zico Kolter, a professor and the director of the Machine Learning Department at Carnegie Mellon University, to its board of directors. Kolter will also join the Board’s Safety and Security Committee alongside a group of other board members and CEO Sam Altman.

Kolter has a background in developing safety methods for large language models and formerly worked as the Chief Data Scientist at C3.ai.

OpenAI is leading a $60 million funding round for Opal, the same company behind the high-end Tadpole webcam, according to a report from The Information.

While Opal will reportedly continue to sell its webcams, The Information reports that it’s also working on AI-enabled devices that people can use as “creative tools.”

The newspaper will file an amended complaint by August 12. If the Times wins its suit, adding those works means those two companies are on the hook for a minimum of $7.5 billion in statutory damages alone.

John Schulman is leaving to work on alignment at Anthropic, OpenAI’s chief rival. In a reply post on X, CEO Sam Altman thanked Schulman and said he “laid out a significant fraction of what became OpenAI’s initial strategy.”

In his new job, Schulman will work closely with Jan Leike, another senior leader who recently left OpenAI for Anthropic due to concerns that safety had taken a backseat to business priorities.

CNBC spotted the update this week in Microsoft’s risk factors with the SEC. These are managed by lawyers to help shield companies from shareholders lawsuits and generally pretty conservative. Still, the change feels like a sign of how OpenAI and its largest investor are drifting apart.

Relatedly, I couldn’t help but notice the number of times Microsoft execs mentioned OpenAI during their earnings call this week: zero.

I was once told that the only reason people are wowed by AI products like ChatGPT is because it mimics us, and we’re obsessed with ourselves. I think about that a lot, especially as I watched this video of ChatGPT Advanced Voice Mode catching its non-existent breath as it counts quickly to 50. So strange.

Sam Altman announced that OpenAI is collaborating with the US AI Safety Institute for early access to their next foundation model (but no release date was specified.)
He also emphasized OpenAI’s commitment to dedicating 20% of computing resources to safety, a promise originally made to the now-defunct Superalignment team.

Plus, he noted that OpenAI has removed non-disparagement clauses for employees and provisions allowing the cancellation of vested equity.

The trend of hallucinations showing up in public AI demos continues. As noted by a couple of reporters already, OpenAI’s demo of its new SearchGPT engine shows results that are mostly either wrong or not helpful.

From The Atlantic’s Matteo Wong:

In a prerecorded demonstration video accompanying the announcement, a mock user types music festivals in boone north carolina in august into the SearchGPT interface. The tool then pulls up a list of festivals that it states are taking place in Boone this August, the first being An Appalachian Summer Festival, which according to the tool is hosting a series of arts events from July 29 to August 16 of this year. Someone in Boone hoping to buy tickets to one of those concerts, however, would run into trouble. In fact, the festival started on June 29 and will have its final concert on July 27. Instead, July 29–August 16 are the dates for which the festival’s box office will be officially closed. (I confirmed these dates with the festival’s box office.)

According to a post from OpenAI CEO Sam Altman, ChatGPT Plus subscribers will get access to its new voice feature next week. The company initially demoed this feature back in May, you know, the one that sounded like Scarlett Johansson’s character from the movie Her? They yoinked that voice option though, sorry guys.

This will be interesting to test out, especially ahead of ChatGPT getting baked into Siri.

Open source developers dismissed OpenAI from their 2022 lawsuit alleging that it violated copyright law by reproducing their code without attribution.

As Bloomberg Law writes, the lawsuit will continue against GitHub and Microsoft (although without the Digital Millennium Copyright Act claims that the judge dismissed this month).

According to The Information, OpenAI is in discussion with Broadcom and other semiconductor designers about developing its own artificial intelligence chip to address shortages in its supply chain and reduce dependency on Nvidia. OpenAI has apparently also hired former Google chip staffers.

Bloomberg previously reported in January that OpenAI CEO Sam Altman was planning to raise billions of dollars to set up a network of chip factories.

OpenAI announced that it is teaming up with Los Alamos National Laboratory to explore how advanced AI models, such as GPT-4o, can safely aid in bioscientific research. I’m a bit disappointed because this was the plot of the science fiction horror book I always wanted to write.

The goal is to test how GPT-4o can help scientists perform tasks in a lab using vision and voice modalities.

A recent study found that if a coding problem put before ChatGPT (using GPT-3.5) existed on coding practice site LeetCode before its 2021 training data cutoff, it did a very good job generating functional solutions, writes IEEE Spectrum.

But when the problem was added after 2021, it sometimes didn’t even understand the questions and its success rate seemed to fall off a cliff, underscoring AI’s limitations without enough data.