As an introductory note, if you are interested in the technology, people, and companies working to create those technologies, the Andreeson-Horowitz podcast covers interesting subjects with some excellent guests. Here is the link: https://soundcloud.com/a16z/ai-go
A recent podcast included Frank Chen and Steven Sinofsky (both software development veterans and partners at Andreeson-Horowitz) discussing AlphaGo, Google’s deep learning project. Chen and Sinofsky cover AlphaGo’s recent wins over two of the world’s best players and the technologies that made the accomplishment possible.
The successes of AlphaGo, much like Watson and Big Blue, is a demonstration of what can be accomplished with the thoughtful application of technologies. What’s really interesting about this march of progress is the increasing reliance on an ensemble of technologies.
As Chen and Sinofsky point out, AlphaGo is not a singular new technology but a number of technologies, both cutting-edge and quite old, assembled to accomplish a specific task. This is also true of IBM Watson’s run on Jeopardy, where Natural Language Processing is just the beginning of a long list of software, hardware, and data storage, all combined to quickly answer questions that are not only difficult, but worded in a way that would be particularly difficult for machines to interpret.
These company projects are often as much about marketing as they are research, but as Chen and Sinofsky point out, anyone with a real-world problem to solve will combine whatever tools are available to them in order to get the job done. Modern data solutions require this same kind of inter-operability. In our data-centric future the problems we face will increasingly be solved by combining different resources to solve each problem. At Emcien we like to talk about these technology ensembles in terms of the analytics stack.
For many of our customers, Emcien functions as the central analytics component between their data source and the alerts or actions that help drive their business. And because we see this first hand we’ve come to appreciate how many different technologies can be combined to turn lots of disparate data into something actionable. While some projects are very similar and require the same analytics stack, building a complete solution is simply a matter of putting the right pieces together.