BBVA API Market
Barely any ground-breaking technology firms consider artificial intelligence utterly devoid of interest. Most of them have a foot and half in fields such as machine learning, natural language processing and products that support conversations between machine and user, with APIs and bots representing the end tech product.
Google, Facebook, WhatsApp, Telegram, etc., have all grasped that the way in which products and services are used is naturally evolving from applications to conversation interfaces via smartphones or other devices.
We are on track to see the plot of the Spike Jonze movie ‘Her’ at least partially become a reality. In the film, Joaquin Phoenix plays a lonely man who falls in love with an operating system. While many aspects of ‘Her’ remain the realm of the distant future, trends are moving in the direction depicted in the movie: chatbots capable of understanding the natural language spoken by users and learning as they interact with them, conversation APIs that respond in natural language rather than JSON format… In this future scenario, users do not require dozens of applications to search for information, but instead engage in a technological relationship with artificial intelligence that is capable of responding to their needs in real time.
Google has been working with artificial intelligence for some time. Fruit of this hard work was the recent launch of Allo. Launched on September 21, 2016, this is an instant messaging app that connects users with friends.
It joins a crowded market jostling with more established players, such as Facebook Messenger, WhatsApp, Telegram, WeChat, Snapchat and more. Each of which have their own particular features. They allow users to send each other messages, use emojis or giant stickers in chats, and to share all kinds of content (news, images, video or audio) with their contacts. But none of this is the added value that Allo has to offer.
What is truly novel and interesting about the Google release is the step that it takes in the direction suggested at the top of this article. Users interacting with machines, specifically via the search giant’s virtual assistant. Anyone using Allo can start chatting with the assistant via its user handle @google, by posing the assistant a question or making a request.
For example: @google, can you find me the best Italian restaurants nearby? In said new relationship, not only does the assistant answer the user, but it also asks questions in order to better understand the user’s tastes and requirements, and thus provide the best possible response. It is, to some degree, the sweet and seductive voice (although @google does not speak) of Scarlett Johansson, the American film star who voices the operating system called Samantha in the movie ‘Her’.
Unsurprisingly, this relationship between user and Google assistant is only available in English as yet, but the company aims to rollout the service in the other languages that its products support. The Mountain View company hopes to harness the potential of the Google assistant and extend it to new challenges, such as Google Home. This would represent a further step toward the development of connected homes, capable of reacting to user requests and requirements with sufficient intelligence to understand what is said and respond with specific actions (turning on lights in a room, playing a song at the user’s request… ).
DeepText is one of the latest projects from Facebook in the world of natural language processing and machine learning. The social network is equipping its text engine with the tools that it needs to precisely understand many thousands of messages per second in natural language, and in more than 20 languages!
Facebook harnesses a set of neuronal networks that are trained using a model based, on the one hand, on FBLearner Flow, and on the other on Torch, the toolbox used for machine learning and data mining. DeepText is used by Facebook, for example, for its Messenger platform.
In late 2014 Facebook decided to overhaul the entirety of the machine learning platforms used until that time by the social network. Its aim was to adapt the platforms to advances made in artificial intelligence and establish certain criteria to universalize ML work for all engineers:
● Each machine learning algorithm must be reusable.
● Training channels for said algorithms must also be reusable and serve to improve the performance of other algorithms.
● Model training must be accessible to engineers with differing levels of experience in machine learning and be fully automated.
The studies conducted by the company, revolving around three truly powerful principles, led to the idea for FBLearner Flow: a platform capable of reusing algorithms for different products and conducting thousands of simultaneous and personalized training experiments, while also being simple to use.
FBLearner is used by 25% of Facebook engineers. More than a million models have been trained and the platform currently makes more than 6 million predictions per second. FBLearner Flow is a platform that works with dozens of algorithms at once, based on a workflow launched by an engineer, with the algorithms and model then shared with all the other engineers.
Amazon is another company interested in machine learning and natural language processing as a means of understanding the market and gaining user insight. It is possibly the world’s largest e-commerce platform. And like many tech companies, it wants to understand the role that chatbots will play in the way users relate with businesses, specifically in the e-commerce sector.
Which is perhaps why in September 2016 Amazon struck a deal, halfway between an acquisition and a partnership agreement, with Angel.ai. The latter company chiefly creates services for third parties based on natural language processing and chatbots that can understand human language and establish trust relationships with clients. It is evident that major developments are underway in the world of artificial intelligence, with bots set to change the mobile ecosystem and how users interact with businesses.
APIs allow companies to offer their own e-wallets, building a new user experience that drives and contributes to customer loyalty.