Musk wanted control over OpenAI, emails released by the company allege
[ad_1]
But the emails released Tuesday by OpenAI seem to show the billionaire agreeing that OpenAI would become less transparent as the lab grew closer to achieving its goal of AI that rivals human intelligence. Musk also acknowledges in the messages that the only path forward was for OpenAI to raise vast sums as a for-profit enterprise. “Even raising several hundred million won’t be enough. This needs billions per year immediately or forget it,” Musk wrote in 2018, after parting ways with OpenAI.
Musk did not immediately respond to a request for comment. A spokesperson for OpenAI declined to comment beyond the blog post, which also said that the company will ask a court to dismiss Musk’s lawsuit.
The partly-redacted messages shared by OpenAI shed light on one of the most closely-scrutinized founding stories in Silicon Valley history. Industry insiders immediately pored over emails from the company’s early days, with some people attempting to use the latest AI tools to un-redact sentences.
OpenAI fired the starting gun on the current AI arms race when it released the popular ChatGPT in late 2022. Tech leaders herald “generative” AI as a tech transformation on par with the release of the internet or smartphones. The boom has catapulted OpenAI to the top of the tech industry, allowing it to raise hundreds of millions of dollars in funding and giving its CEO Sam Altman influence in conversations over how governments should regulate the tech.
Musk has consistently questioned how OpenAI is able to make money and not share all of its research when it is governed by a nonprofit board that he helped to set up. Altman has generally declined to respond to Musk’s criticisms, but the new blog post changes that. The company is now directly accusing Musk of misrepresenting history.
Musk’s lawsuit asks a judge to order the company not to financially benefit from its AI and to share more of its research with the public. Musk himself has founded his own AI lab called X.ai, and Tesla has invested in AI for years as it tries to build self-driving cars. Tesla is also developing a humanoid robot called “Optimus.” Musk has also complained that OpenAI and its biggest partner and investor, Microsoft, have scraped his social media company X for data to train their AI models.
Musk is no stranger to court battles. In the past, he has said that he will wage legal fights even if they are unlikely to be successful if he believes he is in the right. OpenAI is also facing numerous other lawsuits from authors and news publishers who say the company used their content to train its AI without payment or permission.
In addition to the emails, OpenAI suggests that Musk left OpenAI after an unsuccessful bid to take control, rather than an ideological dispute about openness. “Elon wanted majority equity, initial board control, and to be CEO,” the blog post says. “In the middle of these discussions, he withheld funding.” Venture capitalist and LinkedIn co-founder Reid Hoffman “bridged the gap to cover salaries and operations,” according to OpenAI.
Musk was among those who recognized the company would struggle to pursue its mission of creating artificial general intelligence without significant financial resources, OpenAI said in its response. In the messages shared by the maker of ChatGPT, Musk steered OpenAI toward a for-profit model and called for the company to “start being less open” about the science behind its products. In one from 2015, he argues the company should say “that we are starting with a $1B funding commitment … I will cover whatever anyone else doesn’t provide.”
Musk did not deliver on the full commitment, ultimately providing less than $45 million, OpenAI said, but he continued to push for control of the company as it underwent a change in structure in 2017.
“We couldn’t agree to terms on a for-profit with Elon because we felt it was against the mission for any individual to have absolute control over OpenAI,” the company said in its response. “He then suggested instead merging OpenAI into Tesla.”
One of the emails, sent in early 2018 around the same time that Musk left OpenAI, shows Musk arguing that “Tesla is the only path that could even hope to hold a candle to Google. Even then, the probability of being a counterweight to Google is small.”
In another email released by OpenAI, Musk forwards a message he received that appears to be from the head of Google’s DeepMind AI lab Demis Hassabis that shows Hassabis expressing concern about the dangers of making underlying AI technology “open source” or freely available to the public. OpenAI later republished the blog post to remove Hassabis’ name. Spokespeople for Google did not immediately return a request for comment.
In the email thread, OpenAI chief scientist Ilya Sutskever replies to Musk saying that as AI advances companies should release less data about how the tech works. Musk replies, “yup.”
Faiz Siddiqui contributed to this report
[ad_2]