That’s the question they will be looking to answer—for the first time in history—at next week’s PAPIs AI Startup Battle in Valencia, Spain. If artificial intelligence has a continued track record of enhancing and advancing our decision-making skills, it presents itself as an interesting way to better determine which startups are safer to invest in. Telefonica Open Future has partnered with machine-learning platform provider BigML to extract historical data from the application programming interfaces or APIs of websites like Crunchbase and AngelList—it’s proprietary so we can’t know the exact secrets—and combined it with data like the LinkedIn profile of funders to find a quantifiable correlation among successful startups.
These five finalists will perform live startup pitches where the machine decides who wins based on making live calls to the BigML API:
- Intranetum SaaS self-learning knowledge tool
- Emotion Research Lab emotion-recognition software
- Novelti streaming analytics for the Internet of Things
- Datatrics predictive marketing analytics
- restb artificial intelligence based on a REST API
The only requirement to be a part of this competition is that artificial intelligence is at the startup’s core. But what makes the machine capable of accessing these analytics and what makes it easily available for both techie and layman is the predictive API that connects to that data. The predictive API enables both startups and large companies to leverage machine learning and AI at little cost and little effort. Instead of hiring highly specialized data scientists, these companies can access the same information through a couple of API calls.
ProgrammableWeb views these APIs, where developers essentially gain on-demand programmatic access to expertise that would otherwise be prohibitively expensive to employ, as PhD APIs.
BigML’s VP of predictive applications Atakan Cetinsoy told ProgrammableWeb, "One of the biggest draws of BigML's Machine Learning Platform is its REST API that makes it much easier to build and deploy machine Llearning driven smart applications programmatically. What used to take weeks and months is now possible in hours and days. This has enormous implications the most important of which is that we are entering a new era, where apps without any self-learning will be an exception and not the norm.”
But this doesn’t mean that the API solves all the problems of making AI for the people. The question becomes what sort of data should you be calling on to make these insights.
Conference founder and machine learning PhD Louis Dorard told us that it took “a lot of work coming up with the right historical startup data and enriching it with external sources of data.” Once identifying the right data, they put it into the BigML engine which comes up with a predictive model on new startups.
Next Monday March 14 kicks off the Startup Battle as part of PAPIs Connect which addresses a broader audience of decision makers, who might not be developers but can get the value out of predictive technology, which Dorard says includes machine learning, which in turn is a part of artificial intelligence.
“They just wanted to figure out what machine learning can do for them, what were the use cases in business and how predictive APIs make it easier to use these technologies and making machine learning more accessible.” Dorard went on to say that people want to see the actual products in action.
He says that C-level members of companies need to figure out that it’s not just a technical concern but a strategic one. “The decision-making angle is figuring out what exactly they make predictions for and how that will be converted into business value. A lot of time when they are starting out, the technical teams build predictive models and then it’s not even clear that the business guys use it at all.”
Dorard went on to tell ProgrammableWeb that the conference objective is “to show innovators and strategists why they should care about this stuff and what’s the value for them right now, not to just see it as a long term thing where machines will replace people.”
Given how predictive APIs are still emerging as a technology, Dorard wants to show use cases in action right now. Also, predictive APIs offer the benefit of machine learning without language barriers—data scientists are usually coding in Scala, Python or R, which aren’t common in Web app development—which is a compelling reason for smaller companies and startups to attend.
Keynotes with feature Nuria Oliver from Telefonica’s research and development talking about using predictive APIs for social good and Alex Housley on the intersection of open-source machine learning and open-source predictive APIs. Different sectors like retail and banking will also talk about how they are using machine learning powered by predictive APIs.
Sign up for PAPIs Connect 2016 in Valencia March 14 and 15, just in time for the dramatic Las Falles festival and then tell us below how you are using predictive APIs and machine learning to drive your business.