Artificial Intelligence (AI) is one of the most popular technology trends in the global market right now. Rocking all major headlines, it’s expected to grow to a $70BN industry by 2020 and an estimated 2600 startups worldwide develop AI related technology. News ranges in all directions: from doomsday scenarios of robots eradicating humans, or eliminating all our jobs, to fascinating stories about the potential of AI technologies either saving the lives of millions of patients, or helping to protect our environment. The potential of AI is enormous, but it can be hard to pierce through the hype and figure out where to begin.
How to start applying AI?
Lets elaborate on the relevance. The time lines and severity of the impact of AI are ferociously discussed in the community, but whether you believe predictions of Ray Kurzweil or you are a skeptic, there is no denying that the current level of application is unprecedented. A major reason for this is the exponential growth of data and computing power — two of the driving enablers behind AI and more specifically machine learning. As usual we are drowned in statistics and speculations: By 2020, 85% of customer interactions will be managed without a human and 20% of business content is going to be created by agents as soon as 2018 according to Gartner. Bank of America argues that the rise of AI will lead to cost reduction and new forms of growth that could amount to $14-$33 trillion annually and the last years have consistently been record years in terms of AI funding with 2016 amassing $5B in venture capital spread over 550 startups using AI as a core part of their product according to CB Insights.
AI is still a poorly understood and ambiguous term, but, without going into definitions, it’s actually easier to get started than you might think. First of all, there is an increasing amount of platforms out there and one of the quickest ways to integrate AI into a new or existing business is making use of API’s. The most important question is what kind of service you want to create, or which process you want to optimize. The big players include IBM Watson, Google Prediction, Alchemy API, Wit.ai and many more. For an excellent overview written by Ray Kurzweil, check out this article.
API’s might be quick, but for long term solutions building an intelligent model yourself could be much more interesting. There has been an enormous rise of open-source AI platforms in 2016. The tech giants have their own open-source libraries: e.g. Amazon (DSSTNE), Microsoft (DNTK), or Baidu (WarpCTC). Commercial platforms with support and higher quality include Dato and H20. Other free and popular platforms include Google’s deep-learning platform Tensorflow and Elon Musk’s OpenAI released OpenAI Gym as well as open-source libraries such as Seldon, Theano, Torch and Caffe. Want to build a chatbot? Check out API.ai & Motion.ai for a quick and easy start.
Many people think you have to be a developer to start using machine learning. That’s not true.
Your affinity with math is actually more important than any coding knowledge and courses such as Machine Learning by Andrew Ng on Coursera, or Google’s Deep Learning course on Udacity are great starting points. Here is a great little overview written by Arun Agrahi that sums up easy to read or watch resources such as a human guide to machine learning or 6 areas of AI to watch closely. Or watch:
What happens when we teach a computer how to learn? Technologist Jeremy Howard shares some surprising new developments in the fast-moving field of deep learning.
Hugo Larochelle shares his observations of what’s been made possible with the underpinnings of Deep Learning.
Even though the amount of online resources to learn from and get inspired by grows exponentially, face to face knowledge exchanges are important to people. Events or gatherings enable people to connect with peers & experts whilst providing new insights. The energy you receive from personal connections often triggers the inspiration needed to get started or improve.
The [CITY AI] Value
We work with different city chapters to identify leading practitioners locally and encourage them to share their lessons learned. We want them to learn from one another and in turn also help others to get started in the right way. There is so much noise in the space and only few people are openly sharing their failures when it comes to implementing AI technologies and the business impact. Sharing lessons in person is extremely powerful and therefore we bring different AI communities together via quarterly local events. Thats also the reason we are doing a large scale dedicated AI conference this year called World Summit AI. Feel free to get in touch to start a city chapter, join one of our local events and visit our Summit in Amsterdam!
At the end of the day, even though AI seems to be all about technology, we forget that it’s people who shape our future. The mystery around AI has a fantastic effect in terms of getting more people interested in the topic. It inspires people. However, the hype also causes unrealistic expectations. What is needed is the awareness to focus on the right things (e.g. sustainable building blocks and frameworks) and transparency around failures as well as successes. In the news we are reading about the breakthroughs, but we should focus more on the people and the lessons learned. Embrace transparency and educate enthusiasts.
We don’t just need developers and data scientists, we need more people to experiment and create new use cases.
This will only happen if we make AI more accessible and take part of its mystery away whilst revealing its potential.