OpenAI had earned plenty of plaudits for its transparent and collaborative culture, but the research organization received a drubbing in MIT Technology Review for allegedly breaching the principles it was founded upon. The caustic article exposed a misalignment between the startup’s magnanimous mission and how it operates behind closed doors.
Although some doubts were raised about its mission at the time of Microsoft’s billion-dollar investment in it last year – a view that was expressed by Elon Musk, who incidentally was part of the founding team – the latest revelations have sent shockwaves through the tech industry.
Spoken anonymously, some employees felt that the energy and sense of purpose it started off with had dissipated. Instead, their accounts suggest that the San Francisco-based startup is “obsessed with maintaining secrecy, protecting its image, and retaining the loyalty of its employees.”
Not only was OpenAI’s culture put to question, the article also implied that it may be capitalizing off of panic around the existential risk from AI. The calculated release of some of its studies seemed to follow a pattern that suggested the same.
As the dust settles on the hype surrounding this story, it behoves us to reflect on these revelations, albeit through a broader lens. The startup’s approach, though not without error, seemed unique, especially when seen from the vantage point of big tech companies who were just venturing into the world of AI.
Open Collective Innovation
OpenAI has been conducting research spanning a wide range of disciplines that pursues novel ways of looking at existing problems. But unlike bigger tech companies who keep their researchers close to them, OpenAI is on a mission to collaborate with other research outfits by making its findings open to the public.
The startup, which aims to push AI as far as it will go, strongly believes that that cannot happen when researchers work in silos. According to it, if more people get together to reach a collective goal, the end result will trounce anything that would have been accomplished by a single person done in secret. It currently has 89 repositories on GitHub, opening itself to the software development website’s 40 million users. These projects offer a chance to explore research aimed at the future, and which would eventually be handed over to anyone who wants it – for free.
Such a largely open and unfettered research process is likely to accelerate the progress of AI, taking the world deeper into what it once considered science fiction. In fact, in just four years, the startup has grown to become one of the leading AI research labs in the world today as it continues to democratize AI research.
While this has spawned a slew of experimental research projects, the startup’s long-range goal has been to create an ‘artificial general intelligence’ – or AGI. This is a machine with the learning and reasoning abilities of a human; a technology that augments rather than replaces human capabilities.
The idea is that even though existing AI systems have proven superior to human intelligence, the applications of narrow AI – which gave us breakthrough technologies like digital voice assistants and facial recognition systems – are still limited. Projected to advance the continuum of narrow AI, AGI is seen as the next frontier in technology.
In theory, AGI would be able to make better decisions than humans. According to OpenAI, it can impact modern industries, including healthcare, education, and manufacturing, and address some of the most pressing issues the world is facing today.
While naysayers may question the feasibility of such an ambitious mission, AGI has created a new standard for AI and its development could mean that we may soon arrive at solutions to seemingly intractable problems.
Building The Right Ecosystem For Innovation
This has pushed the notion of openness further and has driven top tech companies to share a lot of their advanced AI research and collaborate on projects to build a secure AI.
For instance, Google open sourced its AI engine TensorFlow in 2015. This allowed experimentation with machine learning (ML) on decentralised data. It also launched a new cloud-based AI Platform that allowed users to collaborate on ML projects. Furthermore, it acquired a startup called DeepMind, which is much like OpenAI in its pursuit to develop advanced AI.
This has also led to a race to set up research facilities focused on advancing AI and Facebook also joined in with its investment in a blue-sky AI lab. Furthermore, Microsoft’s co-founder Paul Allen had also established the non-profit Allen Institute for Artificial Intelligence to conduct ‘high-impact AI research’.
With the objective of promoting and developing AI to drive many tasks of the future, such studies have already made significant headway. Soon, it can help machines understand natural language, and give it the power to learn organically, eventually helping them acquire the ability to think like a human.
In such a scenario, funding – or the lack of it – should not curb efforts to democratize AI. According to reports, DeepMind has been running at massive losses – one to the tune of $570 million in 2019 – up from $154 million three years ago. However, the deep coffers of Alphabet – which owns DeepMind – would ensure that its cogs are well-oiled.
The same could not have been said about OpenAI which, having started off as a non-profit venture, transitioned into a for-profit company to secure additional funding. Since then, it has grown an impressive list of Silicon Valley investors including LinkedIn co-founder Reid Hoffman, PayPal co-founder. Peter Thiel, founding partner of Y Combinator Jessica Livingston, former CTO of Stripe Greg Brockman, and even former CEO of Infosys Vishal Sikka.
What is more, started with nine researchers, OpenAI has an eclectic mix of the best researchers of our time, including Ilya Sutskever, an expert on ML who previously worked on Google Brain. Furthermore, this collaborative effort has also attracted a group of young, talented AI researchers from universities like Stanford, Berkeley, University of California, and New York University.
This cadre of bold thinkers and dreamers – who probably make up the smartest people in most rooms – will likely foster innovation that promises to transform the world in the years to come.