OpenAI, now valued at an impressive $80 billion as of 2024, has become a global leader in AI technology since its inception in 2015.
OpenAI starts in an attempt to achieve Artificial General Intelligence(AGI) and make research accessible for public good.
Originally established as a nonprofit organization, OpenAI aimed to achieve Artificial General Intelligence (AGI). Its goal was to democratize AI by making cutting-edge research accessible for the public good. This ambitious vision was supported by some of Silicon Valley's most influential figures, including Ilya Sutskever, Greg Brockman, and co-chairs Sam Altman and Elon Musk, alongside a significant $1 billion funding commitment from notable backers such as AWS and Infosys.
The foundation of OpenAI was built on an ethos of open collaboration and sharing. This approach led to the development of pioneering initiatives like OpenAI Gym and Universe, which attracted top AI talent and set a new trajectory for the future of AI. Their research blog shows how they created the transformative generative models from the early days.
The Departure of Elon Musk
In February 2018, Musk proposed taking control over the company and left OpenAI after his proposal was not accepted. By then, as a nonprofit, OpenAI had raised only $130 million of its $1 billion goal, leading to concerns about achieving AGI without shifting to a for-profit model.
In February 2019, OpenAI announces GPT-2, and how it won't be completely open-source anymore.
From OpenAI's GPT-2 launch blog:
OpenAI shifted from its open-source model, citing concerns over potential misuse of the technology, marking the first time the organization withheld a model (GPT-2) from being fully open-sourced.
The creation of GPT-2, formation of for-profit company and a subsequent $1 billion investment to the for-profit company from Microsoft in later 2019 marked a turning point for OpenAI, enabling the licensing of GPT-3 and spurring developments in major AI projects like ChatGPT and DALL-E.
We can only suspect that this is when people working closely with the project were convinced that the LLM models they were working on will generally get better with more compute and resources. Marking a significant achievement towards AGI.
Former OpenAI members starts Anthropic in 2021
In 2020, disagreements over AI governance led to the departure of senior OpenAI members, including Vice President of Research Dario Amodei, who, along with his sister Daniela Amodei, founded Anthropic in 2021.****
OpenAI uses 'reinforcement learning from human feedback' to align its AI with human values, continually training the AI with human critiques and rewards. Anthropic's AI, however, follows a 'constitution' that dictates its principles and values, allowing for self-critique without constant human oversight.
There's no doubt that this technology is creating massive headlines.
A closer look at OpenAI's recent blog posts: OpenAI and Elon Musk reveals the complexities of its relationship with Musk, who initially proposed a $1 billion funding commitment in 2015 to surpass the efforts of tech giants like Google and Facebook.
Before his departure in 2018, Musk's total financial contribution was about $45 million. (For reference, 2018 is the year where Tesla missed Model 3 production targets and Elon tries to take Tesla private.) And OpenAI realized that Artificial General Intelligence (AGI) proved more resource-intensive than anticipated, requiring billions of dollars annually. This leads to creation of OpenAI LP, a new “capped-profit” company in 2019, shifting the company towards a for-profit model.
All in all, The story of its development, marked by the saga with Musk and Anthropic, represents a significant chapter in the ongoing narrative of AI's advancement. Although dramatic, such disagreements seem necessary for keeping the balance and achieving greater good for the public. Impactful technology demands careful handling and diverse perspectives.
As always, here are some interesting reads on this topic: