In recent years, the rapid development of artificial intelligence technology has attracted widespread attention. However, this advancement has also brought a series of controversies and issues. Last week, billionaire Elon Musk initiated legal action against the U.S. artificial intelligence company OpenAI, accusing it of deviating from its original mission and using AI for profit rather than for the welfare of humanity. In further developments, OpenAI issued a founders’ joint letter this week and disclosed the email exchanges between Musk and the OpenAI founding team members over the past 8 years. Today, Musk posted on his own social media platform X, stating, “OpenAI has always lived in lies,” as a response to OpenAI’s recent email attacks.
The origin of this turmoil can be traced back to 2018. In 2015, Musk, with other Silicon Valley luminaries including Sam Altman, Peter Thiel, LinkedIn co-founder Reid Hoffman, and Y Combinator co-founder Jessica Livingston, co-founded OpenAI. Endorsed by these stars, OpenAI naturally attracted top talent in the field. OpenAI announced its ambitious goal at inception: to create a non-profit organization that focuses on developing artificial intelligence “in the most beneficial way for humanity,” with a joint pledge to provide at least $1 billion in funding for the company’s development. At the time, Musk believed that artificial intelligence posed the “biggest existential threat” to humanity.
However, over time, OpenAI has invested heavily in the field of artificial intelligence but has failed to produce breakthrough research results. Musk’s story with OpenAI has turned into a classic capitalist script: the founders of a company initially have a common goal, but later separate due to differences in vision, leading even to intense legal disputes. In early 2018, Musk felt that OpenAI was significantly behind Google in the AI race. To catch up with Google, Musk proposed a possible solution: to merge OpenAI into his electric vehicle company Tesla and lead it personally.
During the process of convincing OpenAI researchers to join Tesla, Musk had conflicts with other colleagues, which exacerbated the divergence between the two parties.
By February 2018, the OpenAI founding team, including Altman, did not accept Musk’s suggestions. Faced with this rejection, Musk then decided to step down from OpenAI’s board of directors. He publicly announced that because Tesla’s advances in AI for autonomous driving technology could potentially compete with OpenAI for talent, this created a conflict of interest, which is reasonable in business competition. Indeed, Tesla had already successfully poached one of OpenAI’s top talents, Andrej Karpathy, appointing him as the technical director for its self-driving project. However, many at OpenAI were skeptical about the real reasons for Musk’s departure. His resignation speech in the OpenAI office mainly emphasized this conflict of interest but did not win the support or conviction of the majority of the staff. Moreover, compared to the $1 billion in support Musk initially promised to OpenAI, so far, only about $100 million has been realized.
As financial support was interrupted, OpenAI faced the immense costs required to train large artificial intelligence models, plunging them into a predicament. In the autumn of 2018, the OpenAI team further realized the exorbitant costs necessary to become a leading AI company. The emergence of Google Brain’s “Transformer” model opened up new research areas and continually enhanced the capabilities of AI. However, it meant that there was a need to continuously provide large data sets for model training—a costly endeavor. Consequently, OpenAI chose an important turning point and decided to begin training the Transformer model.
On March 11, 2019, OpenAI announced plans to establish a for-profit entity aimed at raising sufficient funds to cover expensive computational costs. The company stated, “We want to increase our ability to raise funds while continuing to fulfill our mission. Based on our understanding, the existing legal framework cannot balance these two aspects.” The company also announced that it would limit the return on investment for investors, with any excess profits being returned to the original non-profit organization.
Insiders revealed that Altman took an unconventional approach by deciding not to hold any equity in the new for-profit entity. Altman publicly asserted that as an individual who is already very wealthy and has invested in many successful tech startups, he is not seeking additional wealth. He maintained that transforming the non-profit into a formal business company was to continue advancing the company’s business and not for personal gain. Such a decision would help him stay aligned with the company’s original mission and goals.
Within the following six months, OpenAI received $1 billion in funding from Microsoft, which provided capital support and vast technical and infrastructure expertise. Together, they developed an advanced supercomputer specifically for training large-scale models, leading to the creation of ChatGPT and the image generator DALL-E. Their latest language model, GPT-4, is leading the industry with a staggering 1 trillion parameters.
With the release of ChatGPT, OpenAI quickly became one of the most watched technology startups in the industry, prompting even Google to cautiously launch its competitive product, Bard. The release of ChatGPT rapidly gained attention, eliciting a strong response from Musk. Just one month after its release, Musk revoked OpenAI’s access to Twitter data, a privilege that had been guaranteed by a contract before his acquisition of Twitter.
On February 17, 2022, Musk stated on Twitter that OpenAI was originally an open-source and non-profit organization dedicated to countering Google. However, it has now transformed into a closed-source, for-profit company controlled by Microsoft. He wrote: “OpenAI was meant to be an open-source, non-profit platform to balance Google, but now it has become an entity pursuing profit maximization, tightly controlled by Microsoft.”
Lately, on March 15 Musk tweeted: “I still don’t understand how a non-profit organization to which I donated about $100 million transformed into a for-profit enterprise valued at $30 billion. If this is legal, why doesn’t everyone do it?” In the face of Musk’s series of questions, OpenAI chose not to respond publicly.
On March 25, Sam Altman responded to Musk’s critique in a podcast episode. While Altman expressed respect for Musk’s achievements, acknowledging him as a hero in the past, he also criticized Musk’s conduct on Twitter, suggesting it was disappointing.
Against this background, years of dissatisfaction and resentment towards OpenAI have emerged, and billionaire Musk has even filed a lawsuit against OpenAI. In the lawsuit filed with the San Francisco court, Musk claimed that OpenAI had deviated from its original mission of developing general artificial intelligence for the public and open-source, and that the collaboration with Microsoft harmed that commitment.
A 53-page document that circulated online and garnered widespread attention revealed that OpenAI plans to launch GPT-8 in 2027 to achieve true full-functioning Artificial General Intelligence (AGI). According to the document, since August 2022, OpenAI began building a massive, multimodal supermodel with 1.25 trillion parameters, as an initial stage for the planned GPT-5 project. The first phase of the project, named Q*, is expected to complete training by December 2023, reaching an IQ level of 48. Afterwards, OpenAI plans to launch Q*2024 (GPT-7) and Q*2025 (GPT-8) in 2026 and 2027 respectively, each reaching IQ levels of 96 and 145.
OpenAI once set a series of ambitious plans to advance the progress of AI technology. However, due to the high costs of research and development as well as the impact of litigation with Elon Musk, these plans were forced to be put on hold. Musk has consistently emphasized that he believes some of OpenAI’s practices have deviated significantly from the original philosophy with which he co-founded the company, candidly expressing feeling deceived.
Opening the dialogue on this issue, OpenAI stated: “Our mission is to ensure the development of Artificial General Intelligence (AGI) benefits all of humanity. This means we need to develop AGI that is safe and beneficial, and also ensure that the benefits are widely shared across society.” The company then addressed the lawsuit directly: “We will refute all accusations raised by Musk one by one.”
Musk’s dissatisfaction with the company isn’t without reason; he has strongly suggested that OpenAI introduce profit mechanisms for financing to complete its mission, and discussed such proposals through emails with other founders. In these emails, Musk even spoke of a potential merger with Tesla as a solution to OpenAI’s fundraising issues in the development of AI technology.
Especially, in an email from 2018, Musk referred to Tesla as the only company in the AI field capable of competing with Google, pointing out the hefty AI research and development costs posed financial challenges for OpenAI. “We may not want to admit it, but Tesla is probably the only existence that can contend with Google. Even so, the chances of success are slim,” he expressed in the email. In another email from the same year, Musk further clarified that OpenAI needed “billions” of dollars to successfully achieve its ambitious goals.
In an email from 2016, Ilya Sutskever, one of the founding members of OpenAI, mentioned that maintaining a certain level of secrecy was necessary for the impending AGI technology, and they would not fully open up every scientific principle of the technology.
In summary, OpenAI has shown a clear stance in responding to challenges and criticisms: sticking to its mission while also seeking necessary commercial avenues to promote its development. As for Musk’s lawsuit and related controversies, the company stated it would respond legally and maintain its development direction.
Within the philosophy of OpenAI, it is a firm belief that the benefits of artificial intelligence should be accessible to everyone, which does not require disclosing every technical detail. Elon Musk holds a supportive stance on this, replying with a simple “Yes” to the relevant emails.
However, at the end of the article, OpenAI sadly criticized Musk’s contradictory actions, re-emphasizing its future development orientation. OpenAI expressed disappointment with the once revered figure—a person who once inspired them to pursue higher goals and later told them they would only fail, and went on to establish a competing company by himself. When OpenAI began to fulfill its mission without Musk’s involvement, he chose to sue them. Despite a long road ahead, OpenAI remains focused on advancing its mission. As their tools gradually improve, they are pleased with the prospect of deploying these systems for everyone to use.
In response to Musk’s lawsuit, internal unity at OpenAI was stronger than ever, altering the previous public perception of internal chaos and leadership disagreements. The act of publishing private emails has led to criticism from online users and industry figures towards OpenAI. New York University Professor Gary Marcus mocked OpenAI’s statements and pointed out that OpenAI’s proclaimed adherence to its mission and pursuit of open-source technology is merely lip service. In reality, Microsoft now has exclusive ownership of OpenAI, reaping 49% of the profits. Marcus criticized that OpenAI is not helping humanity but is instead stealing from artists and writers to increase its own profits and emphasized that OpenAI has almost completely abandoned open source, becoming a very closed AI company.
Regardless of the discord between OpenAI and Musk, OpenAI claims that adhering to their founding charter is untrue. Musk stated in his latest social media tweet that if OpenAI renamed itself to ClosedAI, he would withdraw his lawsuit against them. He even added the label “ClosedAI” to a photo of Sam Altman wearing a visitor badge. Some netizens joked that OpenAI might as well change its name to OpenEmail, which Musk liked.
In this era filled with both challenges and opportunities, the true masters are not those who indulge in solitary refinement but those doers with a grand vision for the greater good. They do not shy away from difficulties, they are pragmatic, laying the foundations and bridges in the vast wilderness of technology, and together they compose the legendary tales of the tech world.
To help more developers gain valuable practical knowledge, the Didi Tech Community brings everyone the highly anticipated “Didi Technology Practices 2023 Annual Collection”. This compilation includes the Didi tech team’s exciting shares from the past year and provides deep insight into numerous technical topics—from the ingenious application of algorithms to the profound value of data mining, from the rigorous assurance of testing procedures to the construction of efficient and flexible front-end and backend technical support systems.
These experiences and knowledge are not only closely related to Didi’s business but also offer broad universality and learning value, suitable for all kinds of technical personnel to reference and draw from.
For readers interested in this rich content, here is how you can obtain it. You can directly read the related content and deeply explore the practical wisdom of the Didi tech team.
