Here's how AI is empowering Facebook to create a better world

Whatever you think of Facebook and its work, you have to admit, the company is doing amazing work in the field of AI and ML


Artificial Intelligence is so seamlessly seeping into our lives that I doubt most of us even know it’s there. The more vocal AIs like Siri or Google Assistant are the ones we’re familiar with, but as far as AI is concerned, we’re just playing with toys.

The real AI is still hidden.

In Facebook’s case, AI and machine learning (ML) are paving the way for an incredible future for the company, a future that will, eventually, benefit us as a whole.

Outlining the work being done in the field of ML and AI in a blogpost, Facebook explains that AI is helping them achieve breakthroughs in fields like computer vision, connectivity and VR at a rate that’s not been possible before.

The tools

Facebook specifically points to three programs, FBLearner Flow, AutoML and Lumos, all of which are powering advanced research at Facebook.

FBLearner Flow is a “platform that makes AI available to everyone”, says Facebook. They claim that this tool is so advanced and user-friendly that “nearly 70 percent” of people using it aren’t experts in AI and they’re comfortably able to run complex experiments.

AutoML can best be described as an AI for ML. As Facebook puts it, its engineers are developing over 300,000 ML models every month. AutoML takes the learnings from these models to further refine and enhance other models, recursively updating the entire ML portfolio in the process.

The last one, Lumos, is a computer vision platform that gives normal people the ability to harness the true power of intelligent vision algorithms.

AI at scale

The keyword here is scale. AI, by itself, isn’t much: A proof-of-concept in a lab is great, but what’s the point if it has no impact on people’s lives outside of it?

As an example, Facebook points to the newly announced AI-based style transfer (think Prisma) for Facebook Live. Prisma struggled to bring its technology to mobile devices at first, particularly for video, and had to rely on computing servers in the cloud. Facebook, however, condensed the entire AI model required to do this style transfer into a relatively light algorithm using a platform called Caffe2Go. This algorithm can now run on your phone and give you the same effects that previously required the power of the cloud.

This is incredible work and its benefits aren’t just limited to stylised video. The fact that a complex computer vision algorithm (Facebook calls this an industrial-strength deep learning platform) could be so condensed paves the way for future augmented reality applications, gesture-based controls, facial recognition and a whole lot more.

Real ML on a mobile device can be used to create a truly personalised experience. You might like the fact that Google Now knows where you are and throws up contextual cards today, but real ML can give you Samantha.

Facebook claims that right now, these tools have helped it translate posts in real-time, enhance the lip-syncing capability of avatars in VR, describe scenes to a blind person, enable enhanced voice interactions and more.

AI and hardware

The benefits of AI translate very well in the real world as well. Take connectivity for instance. Facebook claims to be using advancements in computer vision to analyse city layouts in real-time. The results of this analysis can be used to do things like determine the best location for a cell tower. For example. This aligns nicely with Facebook’s own projects in this field, particularly with their research into millimetre wave transmission technology.

The future: An AI with common sense

Millions of years of evolution has allowed us to take things for granted. We expect things to fall down, our thoughts are instantaneous, we can understand essays and appreciate art, walking is thoughtless; these are things that AI can’t fathom just yet.

If your professor asks you to open your textbook to page 42, you just do it. An AI would have to understand where it is, what a professor is, what subject he teaches, the books related to that subject, how to open a book and more.

Facebook’s focus — and that of every other entity researching AI — is to make a truly intelligent AI. An AI with the ability to watch and learn, to understand and appreciate context, with the ability and willingness to experiment. The era of common sense if you will.

AI is not the dominion of Facebook of course. Just about every tech company and research institute worth its salt is investing time and money into this endeavour. Facebook has just taken the trouble to give us a little snifter of their work on AI. For now, that’s more than enough.

Whatever you think of Facebook and its work, you have to admit, the company is doing amazing work in the field of AI and ML.

Find latest and upcoming tech gadgets online on Tech2 Gadgets. Get technology news, gadgets reviews & ratings. Popular gadgets including laptop, tablet and mobile specifications, features, prices, comparison.