OpenAI Notes
General notes on OpenAI's interview / hiring process from the resources available on their site. Additionally collected notes on their company history / philosophy from other resources / interviews etc.
General Information
OpenAI is a research laboratory that consists of a non-profit parent organization (OpenAI LP) and a capped for-profit child corporation (OpenAI LP). OpenAI's stated goal is to develop and promote friendly AI in a way that benefits humanity as a whole. It was founded in 2015 by Elon Musk, Sam Altman, Pieter Thiel, Y Combinator co-founder Jessica Livinggston, Linkedin co-founder Reid Hoffman, AWS, Y Combinator Research, and Indian IT / information services company Infosys with a collective pledge of 1 billion dollars. In 2018 Elon Musk resigned from the board to avoid a conflict of interest with Tesla's self-driving AI development but remained a donor. In 2019 Microsoft and Matthew Brown Companies (?) invested another 1 billion dollars in OpenAI LP which was created (in part?) to recieve the investment. OpenAI's headquarters is in the Pioneer Building in San Francisco's Mission District.
Upon it's founding OpenAI pledgged to 'freely colaborate' with other institutions and researchers by making it's research / patents open to the public. In 2016 it released OpenAI Gym, a public platform for reinforcement learning research. Also in 2016 they released a platform for measuring and training general intelliggence AI using games websites and other applications (?).
In 2020 OpenAI announced GPT-3, a language model trained internet-collected text along with GPT-3's API referred to as 'The API' (?). This API represents the core functionality behind their first for-profit product (?). GPT is specifically geared towards answering text-based questions, but it is also capable of improvising and translating. In 2022 ChatGPT was released to the general public. ChatGPT is sometimes referred to as GPT-3.5, representing a stage of development somewhere between GPT-3 and GPT-4 which is already functional but not public.
In 2021 OpenAI released DALL-E and in 2022 DALL-E 2, both of which are generative image models.
OpenAI's 'capped' for profit structure dictates that it has a profit cap set at 100 times any investment. This presumably would mean that an investment of 1 billion dollars would only allow the investor to make a 100 billion dollar profit (?).
Risks to Humanity
Elon Musk has described human-level / general artificial intelligence as humanitiy's greatest existential risk, a sentiment that seems to be held by other key figures involved with OpenAI and historical figures like Stephen Hawking and Stuart Russel. OpenAI has stated that AI "should be an extension of individual human wills and, in the spirit of liberty, as broadly and evenly distributed as possible." Musk has aknowledged that the attempt to develop beneficial AI could lead to the development of destructive AI, but that the best defence is to "empower as many people as possible to have AI." Altman has stated that they do not plan to release all the source code for the projects developed at OpenAI but that they would plan eventually to have a governance board of regionally elected representatives (?).
Interestingly, GPT-2 was initially not released to the public due to a desire to "err on the side of caution" against potential misuse. This drew some criticism as being against their core mission.