Earlier this month Google quietly unveiled an incredibly unique opportunity for seasoned developers to explore pilot experiments based on some of Google’s frontier AI research, aptly called “AI Workshop.” Google already makes a wealth of AI research available on platforms from GitHub to its own AI Hub, complete with a searchable library of ready-to-use code examples, demonstrations and even wrappers around production systems. What makes AI Workshop so different from these other mediums is that it presents an early glimpse at selections from Google’s bleeding edge enterprise AI research that might become future product offerings, allowing the research and developer community to provide feedback that can help influence those innovations, granting a rare opportunity to help shape the future of AI in the enterprise.
The rise of deep learning has represented a unique era of collaboration between the commercial and research sectors. Many of the underlying toolkits, workflows, algorithms and even research models have all been released under open source licenses, with companies, academics and citizen researchers collaborating together to create innovative new applications and to improve the underlying infrastructure powering the modern deep learning revolution.
Google’s AI Workshop expands this collaboration to the most important process of all: translating this fire hose of innovation into the demanding world of the enterprise.
Google defines its new AI Workshop as an opportunity for “customers, partners, researchers, and developers” to experiment with “new concepts, new techniques, and new applications at the furthest extents of science and technology” in order to “connect research and practice so that we can address enterprise AI challenges together.”
Roughly translated, AI Workshop represents a place where Google will over time role out advanced experimental AI pilots it believes might have significant relevance to the enterprise. These experiments may be short lived or may evolve directly into future production products or features, based on the feedback Google receives. The goal is essentially to externalize selected enterprise-relevant AI research in a way that Google’s own research teams building those tools can interact directly with potential users to better understand how those innovations might be used in the real world, the kinds of use cases the algorithms would need to cope with and to learn more about their limitations and strengths.
Uniquely, these experiments represent a rare opportunity for companies and advanced AI users to get a glimpse of some of the directions in which Google’s enterprise AI offerings are heading and to participate in helping to shape today’s research into tomorrow’s production tools through their feedback.
As research experiments, Google’s AI Workshop pilots are aimed primarily at more technically advanced users, especially developers who specialize in creating prototypes and proof of concept applications and academic researchers.
Experiments are not typically “products in miniature” but rather research-grade pilots that may in some cases be simply a light wrapper around an advanced algorithm or analytic service that Google is prototyping. This means there will often be more work on the developer’s end to get started, but the benefit in return is the opportunity to be on the ground floor in testing these potential future new products and helping shape them into services that would be of greatest benefit to the kind of workloads most important to the developer’s company.
While Google has long worked closely with its large customers to gather feedback from early access programs, AI Workshop represents a broadening of this access to a much wider selection of Google Cloud developers. To gain access, developers will have to sign Google’s Trusted Tester Agreement and agree to an NDA to participate. They will then submit a request for access to experiments of interest, explaining their use case. The relevant Google research team behind that experiment will then determine whether the user's specific use case is a good fit for the experiment. For example, an experiment designed to optimize model training on extreme dataset sizes would be relevant only to a specific narrow subset of users.
As of this writing Google is offering seven experiments titled Turbo Image Filter, Augmented Learning for Image Classification, Interpretable Image Classification with Prototypes, Image Classification with Confidence Scoring, Mixed Integer Linear Program Solver, Semantic Similarity for Natural Language and Label Error Detector for Images.
Putting this all together, Google’s AI Workshop represents a unique new step in Google’s efforts to democratize access to AI, opening up broader collaborative opportunities to help shape the future of AI in the enterprise.