What is Intelligence?
According to Yann LeCun, Director of AI Research at Facebook, intelligence is “the capacity to solve problems, to plan, to perceive and to act. Intelligence is often associated with animals and humans, but we are trying to reproduce this capacity with machines by making robots move, enabling conversations between computers and humans, or having them play strategy games, etc.” For example, in March 2016, AlphaGo was the first computer program to ever beat a professional player at the game of Go.
Artificial Intelligence and Machine Learning: Why Now?
Machine Learning is a field of Artificial Intelligence (AI) that explores the conception, analysis, and development of algorithms allowing a machine to learn from examples (data) and make predictions.
Such a machine would be able to act like an intelligent system, i.e. to observe the world, understand, collect and aggregate information, update its model of the world, then plan, decide and take actions, and finally observe again.
AI can solve lots of problems and be applied to many industries from healthcare to retail, to finance. Although Machine Learning and Deep Learning have been around for a while, it is the ever-increasing computer power that have brought these fields back on center stage (NVIDIA’s GPU-accelerated computing for instance).
Machine Learning in our Daily Lives
Did you know that Machine Learning is in all our everyday apps? From Uber and AirBnB to Facebook, Machine Learning is everywhere. It is behind Uber’s pricing model, taking into account many parameters (weather, time, traffic, etc.) to determine how much your ride should cost.
Large tech companies such as Facebook and IBM Watson are trying to democratize AI, not only inside their organizations, but also outside. Facebook not only trains 500,000 Machine Learning models every month, it also runs about 2 trillion training examples every day! The impact is important too: 25% of the engineers at Facebook leverage the algorithms built by the AI team. IBM Watson has implemented 30 APIs, divided into 4 categories: languages, speech, vision and data insights. These APIs allow companies to build applications using AI algorithms, and today in the U.S. 500 companies have solutions powered by Watson.
AI is changing the enterprise
I recently attended Applied AI Conference, where several startups presented their AI solution for the Enterprise.
Sentient ($143M total funding) is developing the largest AI in the world to help businesses tackle complex problems. At Sentient, more than 25 million decisions are made by AI every week!
Many startups are emerging, each of them tackling a specific pain point. The applications of AI are limitless. Forkable, for instance, predicts what you want to eat for lunch before you even know it and helps teams order lunch. As for Legal Robot, it uses AI to make legal documents less painful for everyone and improve access to justice.
Human vs Machine
One of the main concerns related to AI in our daily lives is the impact on jobs. Even though machines can excel at analytics (Natural Language Processing, pattern identification, etc.), they can’t beat humans at judgment calls, common sense, morals, compassions, and creativity.
Indeed, AI could initiate a sales conversation with a potential client, but it could not close the sale. We still need good salespeople. This would apply to customer support too. Would an angry customer want to speak with a machine? Certainly not.
To sum up, machines will play bigger roles in analytics, and the trend would be to use a combination of AI and human intelligence. This would result in an augmented intelligence, i.e. what Garry Kasparov, the grand chess master, calls “centaurs”. The idea, in chess for instance, is that a human chess player will listen to the moves whispered by the AI and decide to follow or override them (judgment call). Today the best chess player alive is not a human, nor a robot, it is in fact a centaur: Intagrand, a team of humans and several different chess programs.
What about other industries? Centaur doctors will have an improved capacity to spot cancers in medical images.
If you want to learn more about successful Machine Learning platforms for your organizations, Open Innovation and Machine Learning, and AI startups for the Enterprise, join us on July 13 at our Machine Learning Techmeeting! We will start with a panel of big players such as Uber and SAP, and then move into a startup pitch session with Maluuba, Neura, Otosense, Rhythm, Trifacta, among others. See you there!