Author Reinhard Schlieker, From German: Oculus News
Mira Murati escaped the whirlwind of intrigues at ChatGPT. Once a child prodigy from Albania, she is now one of the most powerful women in the world, building her own AI empire. Donald Trump’s AI program is likely to bring a significant influx of money into her coffers.
Meanwhile, ChatGPT also recognizes its co-creator, Mira Murati—a child prodigy and technical genius. But it wasn’t always this way. During her time at OpenAI, the artificial intelligence and programming tech company, the spotlight on her was nearly nonexistent. Perhaps she preferred to remain private, speculated ChatGPT. Not anymore: the 36-year-old is now the founder of her own tech startup and has become a media sensation, not just among those in the AI space. Her work and influence are set to benefit greatly from Trump’s announcement of plans to invest billions in AI data centers.
Murati: A Math Genius
Mira Murati, the star from Albania and a math prodigy, now an American with a prestigious Canadian education, ranks among the 100 most powerful women in the world. Not only did she contribute to the practical application of new technologies during her time at OpenAI, but she also spearheaded development as the company’s Chief Technology Officer (CTO). Her work included projects like DALL-E, the image generator, and other popular AI language learning applications.
When she left OpenAI last September, alongside other top executives, it caused a stir within the industry and beyond, marking the end of an intense period of friction at the otherwise progressive company. Now, she is back in the spotlight. It seemed unlikely that such a highly intelligent researcher, who designed a racing car as a student and contributed to shaping Tesla's “Model X,” would quietly fade into the background.
Superior to Human Intelligence in Daily Life
According to reports, Murati’s new project will focus on Artificial General Intelligence (AGI)—an algorithm-based application that is said to surpass human intelligence in daily life, at least when it comes to specific tasks like passing standardized tests or performing examination tasks. Microsoft, a partner of OpenAI, has already claimed this label for the latest version of ChatGPT (4), which is one of the first AGI applications.
Not long ago, the euphoria of Apple co-founder Steve Wozniak, known as “Woz,” brought us back to reality. According to Woz, he’s still waiting for a computer to visit a random citizen’s house and make coffee in the kitchen.
Of course, as noted by the American magazine Wired, “there are fears that AI could worsen economic inequality, spread racist stereotypes as memes, or harm our ability to recognize authentic media.” However, it also stated: “Based on what’s been made public about the algorithm, GPT-4 doesn’t want to be any more alive than your Texas Instruments 89 calculator wants to take on human form.”
Murati Seeks $100 Million in Seed Capital
Mira Murati is already poised to push boundaries. According to Reuters, she is attempting to raise $100 million in seed capital for her new company. Among those who left OpenAI months before her was Ilya Sutskever, a co-founder, former head of research, and presumed rival of Sam Altman. Sutskever has already raised $1 billion for his new project, "Safe Superintelligence," which Murati likely observed closely. Another rival, Anthropic, founded by former OpenAI employees, has secured billions in funding, including investments from SAP and $8 billion from Amazon alone.
Even though competition in this sector is tightening, Murati’s $100 million target appears modest. Yet it seems sufficient to sign notable names for her startup, including key figures from OpenAI or Google DeepMind. For instance, Jonathan Lachman, a former Chief Technology Officer at OpenAI, is reportedly on board. Murati’s company doesn’t even have a name yet, let alone a product lineup.
Relationship Between OpenAI CEO Sam Altman and Murati Remains Intact
Despite the various "talent raids," the relationship between OpenAI CEO Sam Altman and Murati appears unaffected. Both praised each other in farewell statements last September. This followed months of palace intrigue and rebellion within OpenAI. Disagreements over whether OpenAI should remain non-profit to avoid greed-driven abuse or embrace for-profit operations to tackle future challenges were at the heart of disputes between Altman and parts of the advisory board. Observers at the time referred to this conflict as “The Denver Clan in California.”
Return to OpenAI Drama: Murati's Role in Turbulent Times
Without Murati, it’s fair to assume that OpenAI would not have achieved much. The palace-revolt-turned-soap-opera in November 2023 was likely a chilling spectacle for sensitive board members. OpenAI’s founder, Sam Altman, 39, was abruptly informed via video call that he was being dismissed by his co-founder, Ilya Sutskever, 38, who was on the board at the time. Another veteran, Chairman Greg Brockman, 37, was similarly ousted without much ceremony.
Billion-Dollar Deals at Stake
Publicly, the board claimed to have lost trust in Altman, but behind the scenes, the issue revolved around billion-dollar deals and the company’s direction. Initially established as a non-profit organization, market constraints necessitated a shift to a more conventional, profit-driven corporate structure. This commercial arm of OpenAI now accounts for a significant portion of the umbrella organization’s operations, with shares divided among Microsoft (reportedly holding over 49%), other Silicon Valley investors, and employees.
Most board members represented a faction focused on human-centered AI ethics, while a minority, led by Altman, aimed to attract investors to accelerate AI development. These disagreements had already led to the departure of several long-serving employees. Elon Musk, too, had previously clashed with Altman. In this sense, Silicon Valley seemed more like a treacherous canyon than a thriving valley.
Murati’s Brief Tenure as Interim CEO
While Altman was on a roadshow promoting OpenAI’s chip production in early November 2023, the palace revolt was planned and executed. However, the backlash was swift and overwhelming. Employees, alongside Microsoft, resisted Altman’s dismissal. A mass resignation threat from over 700 employees—out of 770—made waves, with key figures publicly expressing their dissent on platforms like X (formerly Twitter).
In response, the board appointed Mira Murati as interim CEO. She accepted the role, emphasizing continuity for the company, though this likely wasn’t the revolution the board had envisioned. After three days, Murati returned to her team, satisfied to step back.
Altman Hints at His Return
A major investor urged the board to end the chaos and reverse its decisions. Altman hinted at his imminent return, receiving widespread support on X, including from Murati. Soon after, Altman and Brockman were invited back to OpenAI’s headquarters to discuss the company’s future. Murati’s conciliatory approach offended the board once again, leading to her dismissal from the interim CEO role.
A Messy End for the Board
The powerful board was left in shambles, realizing that competitors like Google and Salesforce were starting to poach their employees. On November 21, 2023, they relinquished their authority. The new board included a Microsoft representative, former U.S. Treasury Secretary Lawrence Summers, and the reinstated Sam Altman and Greg Brockman. Altman returned triumphantly as CEO.
“The Times” aptly summed up the saga: “Only humans could create this level of chaos.”
Meanwhile, Mira Murati has worked and continues to work on further developing artificial intelligence in the sense of a function that serves humanity. It must be clear to the new multimillionaire that one must also make money with it, and thus, after her unfortunate experiences, it was expected that she would combine both fields with her sense of responsibility.
In an article for the specialized journal Daedalus in 2022, she articulated her goals and considerations: “As we make models like GPT-3 more competent, we also need to bring them closer to human values, which means they should be more truthful and less harmful.”
Tracing the Origins of Thought
OpenAI researchers have now trained language models that can follow users’ goals much better than GPT-3 by being more truthful and less harmful. These models, called InstructGPT, are trained with human involvement so that people can use reinforcement to guide the models' behavior in the direction we desire, reinforcing good outcomes and preventing unwanted behaviors.
The mathematician and programmer, holding a degree in mechanical engineering, is increasingly concerned with questions of philosophy and neuroscience. How is language created? Which areas of the brain are active when we speak, and how do they function when specific content is expressed? The ongoing research process, Murati says about her goals, aims to lead her to the origins of thought while simultaneously decoding the laws of word usage and word finding—whether through studying infant language development or understanding poetic works: “Technology, culture, civilization—none of these happen without language. Language is both the pinnacle and the foundation of human intelligence.”
How Do Languages Work?
But there is a problem: What exactly are languages? How do they work? We can think of language as a response to context and environment. But if we cannot write down the rules of language, how can we teach them to a machine? This problem has puzzled thinkers for a century, and now answers are beginning to emerge,” said the scientist. To understand that Murati “speaks English only with a noticeable accent,” you would need a German publication (FAZ 2023).
On the other hand, the former prodigy from the Albanian provinces seems to be only at the beginning of a remarkable career. For her, the sacrifice at OpenAI was just a step forward. Even though her new company may appear shrouded in mystery, she clearly defined the direction she wanted to pursue two years ago.
The original version of this article, “This Woman Profits More Than Donald Trump,” comes from WirtschaftsKurier.