If you’re a company entrenched in an arms race for artificial intelligence (AI) technologies, you could do worse than tapping into a pool of thousands of data scientists to augment your digital products and services.
That’s the pole position Google holds after acquiring crowdsourcing platform Kaggle last week for an undisclosed sum. Some 600,000 professional data crunchers use Kaggle to build prediction models for such heady challenges as cancer detection and heart disease diagnoses. And experts say Kaggle could help Google facilitate broader adoption of AI technologies.
“Data science and machine learning is now global and this is a validation of the idea that Google recognizes that most of the smartest people in the world work for somebody else,” Neil Jacobstein, who chairs the artificial intelligence and robotics track at Singularity University, told CIO.com. “This is potentially a very positive move, I think, that could make everybody more competitive.”
Kaggle’s team will be integrated with Google’s other AI assets, such as its developer library for TensorFlow, its open source machine learning software, said Fei-Fei Li, chief scientist of artificial intelligence and machine learning at Google Cloud, at the company’s Google Cloud Next event last week.
Computing systems have become exponentially more powerful over the past decade, enabling companies to collect vast amounts of data. Despite advances in analytics software intended to sift through such data, companies require people who can identify and glean insights from the data and turn it into actionable information that provides a competitive advantage. Accordingly, data scientists, those individuals who can leverage predictive models and machine learning algorithms to arrive at those answers, are highly coveted and in short supply.
Democratizing machine learning
Fortunately for Google rivals and others looking to tap Kaggle’s talent base, the community will remain open to all data scientists, companies and technologies, Kaggle CEO Anthony Goldbloom said in a blog post announcing the deal. This strategy is in line with Google’s plans to democratize AI for the world by “lowering the barriers of entry, and making it available to the largest possible community of developers, users and enterprises,” Google’s Li said.
If Google’s openness pledge holds, the acquisition could help further democratize AI for a world that has yet to leverage such technologies to any appreciable effect, according to Singularity University’s Jacobstein. He expects Google will try to foster a closer relationship with the community, helping to host the competitions and providing technology, including scalable infrastructure and the ability to store and query large data sets.
Jacobstein is optimistic Google will keep its promise, noting that Google tends to act as a “responsible and enlightened player in the AI community,” open-sourcing such key technologies as the TensorFlow machine learning library. And DeepMind, the renowned AI company Google acquired in 2014, open sourced its flagship platform, DeepMind Lab, a 3D game platform tailored for AI research, last year.
Whether Google can harness these technologies to make the world a better place, a goal shared by Facebook, Microsoft, Baidu and other tech giants, remains an open question.
“The real challenge is deploying AI in everyday human affairs in a way that allows us to augment our intelligence and manage our planet intelligently,” Jacobstein said. “We have a huge opportunity to deploy better decision-making capabilities across the board and we are underutilizing those tools.”
Necessary progression of AI
For Google, acquiring Kaggle is part of a necessary progression for AI technologies, including natural language processing, predictive analytics and machine learning tools intended to achieve breakthroughs, said Forrester Research analyst Brian Hopkins.
“Google’s acquisition of Kaggle essentially enables that innovation to be exponential because instead of saying there is one set of people at Microsoft or Facebook that can do this, Google can say that we are going to crowdsource this and create a set of emerging technology innovation,” Hopkins said. “Being able to crowdsource and bring the best minds in and incentivize them to do that work is the way it has to happen. There is no other way it can happen. I’m pretty sure Google sees that.”
However, Hopkins said it remains to be seen whether Google can effectively commercialize anything from its AI assets. Its so-called moonshots, from self-driving cars to internet-enabling balloons, have yet to pan out.
“Google is positioned to potentially repeat what they did [for search and AdWords] in the AI space because it is a big area for them, but they have competition,” Hopkins said. “They need to find the commercialization a-ha of how they’re going to turn AI into something profitable just like they [turned] search into profitability,” Hopkins said.
The reality is that most enterprises haven’t figured out quite what to do with machine learning and associated AI tools yet, let alone how such technologies can give them a competitive edge.
Most organizations are only piloting AI and machine learning tools, including virtual assistants and robo-advisors, said Gartner analyst Whit Andrews, who tracks machine learning and AI technologies. Andrews said that a poll of 923 corporate employees with a title of vice president or higher revealed that 76 percent planned to try some form of AI or machine learning over the next 12 months.