Sam Altman is back as CEO of OpenAI after his abrupt dismissal on Friday, Nov. 17. The move follows immense pressure from employees and investors on the board that ousted him some days ago. To comprehend the origins of this development, it is essential to backtrack and gain insights into Sam Altman's identity and his position within OpenAI.
1. Who is Sam Altman?
Sam Altman's story is a fascinating journey through the intersection of technology, entrepreneurship, and visionary leadership. Born on April 22nd, 1985, Altman demonstrated an extraordinary propensity for innovation from an early age. He studied computer science at Stanford University but dropped out without earning a degree. Nevertheless, this provided the foundation for his future career, which would see his ascension as a prominent figure in the world of technology. At the tender age of 19, Altman embarked on his entrepreneurial journey by co-founding and assuming the role of CEO at Loopt, a geolocation-based mobile social networking application. His keen insight and ability to foresee market trends swiftly propelled him to a prominent position within the entrepreneurial landscape. Forbes magazine recognized his achievements by naming him top investor under 30, while Businessweek magazine acclaimed him as one of the "Best Young Entrepreneurs in Technology."
However, it was his involvement with OpenAI that marked a turning point in his career.
2. OpenAI
On December 11th, 2015, within minutes of its registration, a document signed by, among others, Sam Altman, Elon Musk, and Peter Thiel, as well as donor companies such as Amazon and Y-Combinator, was uploaded to the Openai.com domain. This action marks the commencement of one of the most controversial and fascinating initiatives of the past decade: OpenAI.
The project is presented as follows: «OpenAI is a non-profit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact.»
Therefore, OpenAI was founded in 2015 as an artificial intelligence research laboratory, and with the stated goal of developing and promoting friendly AI so that humanity can benefit from it.
However, in December 2018 OpenAI transitioned from a nonprofit organization to a capped-profit company, attracting venture funding and granting stakes to employees. A few months later, Microsoft invested a billion dollars into the venture, and OpenAI went from being an artificial intelligence laboratory to a company actively engaged in its production.
2.1 Corporate structure
OpenAI’s corporate structure is organized around two entities: OpenAI, Inc., which is a single-member Delaware LLC controlled by OpenAI non-profit, and OpenAI LP, which is a capped, for-profit organization.
The OpenAI LP is governed by the board of OpenAI, Inc, which acts as a General Partner. In an LP, the General Partner has unlimited liability, yet complete management and control of the business. On the contrary, Limited Partners are not involved in the management of the LP. They also have limited liability, which is usually commensurate with the size of the investment. Limited Partners comprise employees of the LP, some of the board members, and other investors like Reid Hoffman’s charitable foundation, Khosla Ventures, and Microsoft, the leading investor in the LP.
3. Partnership with Microsoft
Diving deeper into Microsoft’s complicated relationship with OpenAI, a commercial partnership was first announced in July 2019 along with the previously mentioned billion-dollar investment. The companies felt that joining forces would be the best way to achieve OpenAI’s lofty goal of creating artificial general intelligence (AGI) while making sure that these breakthroughs result in AI that is safe, reliable, and ethical. This came with the caveat that further work must be done on Azure, with OpenAI having to port its services to Microsoft’s cloud computing platform. The two companies would together develop supercomputing technologies for Azure, which in turn would enable developers to create the next generation of AI applications. This helped differentiate Azure in an era where cloud computing was becoming increasingly commoditized, as integrating AI into the very fabric of the platform gave it a crucial edge over more standardized platforms. Today, Azure is one of the leaders of the cloud industry. It has become the primary platform for companies to develop AI models that they wish to incorporate into their workflow.
In 2023, following reports of Microsoft investing $10 billion more in OpenAI, another chapter in this influential partnership was announced. This multi-year, multi-billion-dollar investment is focused on further advancing Azure’s AI infrastructure and the development of specialized supercomputers to power OpenAI’s independent AI research. The terms of this investment stipulate that Microsoft has certain profit-sharing rights until it breaks even, after which it retains a 49% profit share until a cap is reached. This 49% also represents Microsoft’s stake in OpenAI, a figure that helps evade regulatory attention. Usually a company like Microsoft would have just assimilated a startup like OpenAI and further developed its technology in-house. With this partnership on the other hand, Microsoft can influence OpenAI while it remains independent, and OpenAI has dependable funding with some level of freedom. Although this may be a symbiotic relationship of sorts, the synergies aren’t exactly perfect. There have been several disagreements with Microsoft employees criticizing the reduced spending on their internal AI, especially since OpenAI has withheld access to specific details of their technologies. One of these points of contention was how quickly OpenAI launched ChatGPT a year ago.
3.1 ChatGPT
Launched on November 30th, 2022, the large language model-based (LLM) chatbot took the world by storm to such an extent that even OpenAI engineers themselves were surprised at the scope of the response. Human beings are social creatures, and maybe it is not a coincidence that the first AI to explode like this was a chatbot. Based on OpenAI’s proprietary generative pre-trained transformer (GPT) models, it is commonly used as a reference for routine writing tasks, prompting concern from educators, scholars, and academics about the future of academia. It is not limited to literary tasks, and has further been used to write code, assist in giving legal verdicts, answer medical questions, and even to develop video game characters. The other side of the coin is how it has reduced human creativity and increased misinformation, whether intentionally or unintentionally. It has most prominently, however, contributed to existential fears about the rise of AI and raised questions like how many jobs will soon be replaced if AI keeps developing at this rate. Other controversies include numerous data protection and privacy concerns, which led to it being banned in Italy for some time. Google researchers also found that they could cause it to malfunction and extract restricted details by using keywords and issuing absurd commands with $200 worth of questions. They fear what a dedicated and well-funded bad actor could achieve if this flaw was not corrected. While the world waits for AI regulation to catch up, it is up to us to decide whether to incorporate it into our workflows or not. It doesn’t necessarily have to replace jobs, and could be leveraged to complete tasks more efficiently, like when we google something to look something up. ChatGPT could be used in a similar capacity without sacrificing all agency in our work. At the end of the day, this can be one of many tools we use, or it can be a crutch we constantly lean on until one day we have forgotten how to stand without it.
4. The Future
Merely 4 days after his dismissal, Sam Altman was reinstated as CEO on November 21st. He has stated his main goal is to “get the company back on solid ground”, but one has to wonder how he can be held accountable going forward when almost every employee is ready to follow him out the door. The reason for ousting him given by the board was a lack of candor in his activities, possibly related to the mysterious OpenAI project Q*. Little is known about this project beyond its name and it has been claimed to signal the imminent advent of AGI, despite outright rejections of this claim by Microsoft President Brad Smith. Its name Q* is speculated to be derived from Q-learning, a technique used to teach AI via reinforcement learning. Reuters has also reported that Q* was able to solve math problems at a grade school, which is especially critical since LLMs are trained on pattern recognition. Hence, they understand languages but have often struggled with math, resulting in some people calling ChatGPT a “stochastic parrot.” Simple word problems with slightly different values are comprehended as the same question, prompting the AI to reply with the same answer without any calculation or method. However, if Q* can solve these kinds of problems, then it must have an innate capacity to reason beyond seeing patterns in how people speak. Whether or not Q* will result in the creation of AGI is up for debate until we know more about it, but in the near future, what is one small click for a man could be a giant leap for mankind.
Bibliography:
https://fourweekmba.com/openai-microsoft/#:~:text=Microsoft%20further%20commercializes%20OpenAI%27s%20products,tasks%20required%20by%20large%20models.
Comments