The rise of publicly available machine learning software
As of last month Alphabet Inc.’s AI division, Google DeepMind, has open-sourced their new machine learning platform DeepMind Lab. Artificial Intelligence is the technology of the moment, constantly debated and attracting massive attention from investors. Despite warnings from influential figures including Professor Stephen Hawking, Google’s decision to open up their software to other developers is part of a mass movement to advance the capabilities of AI. They aren’t the only ones, either. Facebook open sourced its own deep learning software last year, and Elon Musk’s non-profit organisation OpenAI recently released Universe, an open software platform that can be used to train AI systems. So, why have Google, OpenAI and others made these platforms public, and how will this affect the adoption of Artificial Intelligence and machine learning as a whole?
Open-sourced machine learning. . . Why?
By making their machine learning platform available to the public, Google has demonstrated an increased openness about its AI research. This is perhaps a response to fears over the company’s AI-focused partnership with fellow tech giants Facebook, IBM, Amazon and Microsoft. Even though the apparently benevolent project aims to hash out the finer ethics of AI, it’s not difficult to see why such a powerful collaboration is making people nervous. There are various other possible motives for making the software accessible, for instance finding new talent and promising startups to add to the Alphabet Inc. family. At the same time, giving developers the chance to access DeepMind Lab (formerly called Labyrinth) will address one of the key issues with AI research – the lack of realistic training environments. With DeepMind Lab, games, web browsers and sites will all be used as real-world puzzles for AI to navigate. If anything, this move shows just how confident Google is about its AI software. Shane Legg, co-founder of DeepMind, openly challenged other developers to try and beat their results. Other labs are gradually adopting this open approach, including universities. Similarly to Google, OpenAI has revealed a new virtual school for AI called Universe, as well as a platform for more complex training called Gym. Universe, like DeepMind Lab, uses games and websites to train AI systems. Musk has always exercised caution when it comes to AI, so the decision to open-source both Universe and Gym will be as much about teaching AI to be good as to be clever. Making machine learning software publicly available is also a much needed move against the concentration of AI power.
Accelerating the adoption curve
The growth of open-source machine learning will facilitate a steeper adoption curve for Artificial Intelligence, encouraging developers and startups to work towards making AI smarter. The availability of these machine learning schools will expand the capabilities of AI – and by using games, websites and web browsers that were originally designed for humans, they will make AI more human-like. Accessible software platforms are changing the way that businesses develop AI, encouraging them to follow in Google, Facebook and OpenAI’s footsteps and be transparent about their research. Therefore, on the whole, the disruption caused to AI development has been largely positive. But a sceptic would ask if we can really trust that these companies are being completely honest about their research into AI. Would they really give away the details of their entire machine learning platform, especially when it’s such a powerful tool? Given that the benefits include sourcing talent, advancing the abilities of Artificial Intelligence, protecting humanity from rogue AI and preventing concentrated power. . . then yes, perhaps they would.
As AI continues to grow, partnerships like the one between Google, Amazon, Facebook, IBM and Microsoft will become all the more important in working out how to create Artificial Intelligence which will not be a threat to humankind. The creation of schools like Universe and DeepMind Lab will also be instrumental in making sure that AI development is honest, transparent and accountable. Now, companies that don’t open-source their software could run the risk of seeming suspiciously secretive. The shift towards accessible machine learning software is therefore an important step towards ensuring that AI works for everyone, instead of a handful of tech moguls. Hopefully, other businesses will follow suit and let go of their guarded confidentiality to build capable Artificial Intelligence that doesn’t want to wipe out humankind.
Does your business use an Artificial Intelligence system? Could it benefit from the use of open-source machine learning to enhance its AI? Will publicly accessible machine learning combat the development of AI monopolies? Share your thoughts and opinions.