The idea of “a machine that thinks” can be traced to Ancient Greeks but it was an IBMer who created the term Machine Learning in 1959. Arthur Samuel was the pioneer in the field of Computer Gaming and Artificial Intelligence. He made contributions to a first-of-a-kind Checkers playing program that was among the world’s first self-learning programs. That’s 62 years ago! Why is AI having such growth in adoption and development only in recent years?
AI has a long history of being the next big thing and has had several “AI Winters”, which are periods of reduced funding and interest. Hype is something very common in emerging technologies and AI experienced several periods like that, with over-inflated promises and unnatural high expectations from businesses and the media.
Despite the multiple periods of rising and falling of AI’s reputation, it has continued to develop. In the past few years, we’ve seen incredible use cases and we are surrounded by AI applications and algorithms that interact with us every day. The question is...why now?
Key Factors for the AI growth now
- More data: the massive proliferation of connectivity and mobility combined with thousands of elements generating data every second and the much lower cost of storage in the cloud made it possible to have the primary source to make AI smarter, which is large data sets
- More compute power: AI requires computers to learn, and the power of those computers will enable the training of more complex Machine Learning models. Hardware specific to AI such as GPUs or TPUs is making AI faster to train and consume. Cloud is making that specialized hardware available to everyone from any regular laptop.
- Better Algorithms: advancements in Neural Networks, higher-level libraries that abstract the complexities of building those networks, or technologies like Transfer Learning to train custom models with fewer data.
What’s missing to see even more adoption?
There is only one thing missing to see an ever higher rate of adoption of AI: skills. Production AI systems that you can trust and rely on need specialized people to train them and be managed at scale. I believe every single business will have 100s or 1000s of production models, to manage all aspects of the business such as forecasting, targeted marketing, customer 360, customer support, and providing hyper-personalized customer experiences. It will be very critical to have people that understand how to monitor and manage production AI so it continues to operate well in our very dynamic and fast-changing world.
There are two ways to tackle the skills gaps:
- More education: we’ve seen new degrees in AI, Data Science, and Machine Learning and tons of Online Education to give the basics of AI to a whole new set of people entering the job market or upskilling data analysts or application developers into ML roles. We also need business leaders to understand how AI works and champion cultural change and drive a good Data and AI strategy.
- No-code AI tools: the democratization of tools that we are experiencing right now will bring millions of new users to the AI field. Technologies such as AutoML will help automate some tasks and hide the complex math behind the algorithms.
A combination of better tools that can be used by non-programmers and more education will bring us a bright and exciting future. The sky's the limit! Just remember: use this technology responsibly.