For my first blog post of 2016, I thought it time to take a cursory look at OpenAI, the non-for-profit organisation being set up by serial entrepreneur Elon Musk. This aims to provide a non-commercial basis for furthering research into Artificial Intelligence, publishing and widely disseminating the output of research carried. To this end, $1 billion worth of funding has been pledged (or ‘donated’) by a number of Silicon Valley luminaries, including PayPal’s Peter Thiel and Y Combinator’s Sam Altman.
At first, I thought that this makes economic sense. In the Open Source software movement, money is not made by developing and licensing complex enabling software. Similarly, bringing together the economic resources of a diverse range of organisations to broaden the AI knowledge pool, can create technology that can be easily used by companies large and small. Just as Internet giants such as Facebook, Amazon, Uber, Google, all use some form of artificial intelligence to automate their services, making AI capabilities more freely available can only stimulate further innovation, often in the most unexpected places. We have seen how the Android open source platform has, together with a variety of proprietary additions, allowed for an explosion of devices including smartphones, low-cost phones, wearables, TVs and tablets, all from companies who would never have had the resources to create a platform of the required complexity to be viable. An open AI initiative can provide similar benefits, not least to Tesla, Musk’s electric car company which is developing self-driving cars, which depends on artificial intelligence to allow them to navigate our roads safely. Machine Learning APIs such as IBM’s Watson Analytics and Microsoft Azure Machine Learning have already gone a long way to make artificial intelligence much more widely available.
However, on closer examination of the Open AI Press Release accompanying the launch of this initiative, things appear slightly different. Amongst the reasons given for this initiative is “to have a leading research institution which can prioritize a good outcome for all over its own self-interest” because “it’s hard to imagine how much it (AI) could damage society if built or used incorrectly.” Now this is a bit odd. Elon Musk has already described Artificial Intelligence as one of humanity’s biggest threats, and together with Bill Gates and Stephen Hawking have warned about the inherent dangers of AI-powered weapons.
So while a fear of the inherent danger of Terminator-style killing machines is logical, as they can be considered as an inevitable evolution of today’s armed drones, what does the OpenAI initiative do to address this? Accelerating the evolution of Artificial Intelligence can only bring forward the date by which autonomous killing machines are technically possible, and much like nuclear weapons, once invented, cannot be uninvented. Moreover, unlike nuclear weapons, once invented, would-be Dr Evils would not be restricted by factors such as scarcity of enriched uranium, meaning that such technology could easily be developed or fall into what our society would consider as being the wrong hands. Making AI software and algorithms more widely and freely available, will in my mind, do very little to address this valid fear.
To the cynic in me, this initiative has more to do with rebalancing where the power resides in artificial intelligence research and talent development. To date, most of the best AI brains can be found at Google, Amazon, Microsoft, Facebook and IBM, and as far as the capability, technology and talent is locked within these organisations, they will create an unsurpassable competitive advantage against any other tech newcomers. By making best-in-class, cutting-edge AI available to all, and creating a more accessible pool of AI talent, Musk and his friends will have effectively reduced this gap. This rationale for OpenAI is a lot easier to understand, and to my mind, a lot more credible. The only downside is that an ambitious project to further one’s own business goals does not make as good a story as saving the world from killer robots.