Article first appeared on Dell EMC as AI Winter Is Not Coming

One of the most popular shows is Game of Thrones. Ever since the first season characters have been warning that Winter is coming. Around this time of year as the leaves change colors, we know that Winter is coming. What about in the technology world is AI Winter coming? I think not.

Since the 1950’s we have had periods of AI Summers and AI Winters. These AI Summers are times of amazing hope of what AI will do for our future. During these phases Enterprise are investing in the technologies powering AI to stay relevant in the market. In the past these AI Summers have typically been followed by an AI Winter where investments in these projects dry up because of lack of ability to execute in the summer phases. For over half a century we have been through 3 phases of AI Winter. However, since around 2000 we have been in the longest AI summer to date and this time there will be no AI Winter. Let’s look at the top 3 reasons why AI Winter is not coming.

Increased Computing Power

The first element preventing another AI Winter is the continuation of Moore’s Law where we have increased the ability to boost compute power. For years we watched as Moore’s Law has allowed for computing processing to double every 2 years and dropping the cost of that computing power for both CPU and GPU. Building and training models with Deep Learning involves large amounts of processing. Now dream projects like driverless cars (ADAS) are finally coming true because the cost to train cars to drive themselves is starting to make financial sense. In the past the investment would have been astronomical to train a car to drive itself. Remember Deep Learning has been around since the 80’s it just the cost to compute wasn’t there for massive data projects. However, the continuation of Moore’s Law is only one component for why AI Winter isn’t coming.

Digital Transformation

The second factor in holding off winter is the digitization of everything. The digital transformation is real and the data that powers it is massive. In fact, IDC predicts that by 2025 the planet will have 163 ZB of data. Mind-blowing numbers but most transaction now takes place in the digital world. For example, last night I ordered pizza from my Mobile Application, then paid for it through an online money transfer. Never once during that process did I speak to a human until the pizza arrived at my door. Every part of this process created a digital footprint of data. The data from simple transactions such as pizza ordering or GPS mapping help Data Scientist and Machine Learning Engineers build models for the next generation of AI applications in the Enterprise. Massive amounts of data ready to train models can now be can captured, accessed, and analyzed to unlock the value of this data will help hold off AI Winter.

Deep Learning Open-source

The final solution to guarding against AI Winter is the advance in Deep Learning and Machine Learning Frameworks. Today’s innovation is accelerated by the advances in the open-source data science world with frameworks like Caffe and Tensorflow. In a previous post we talked about the importance that frameworks like these were built by the world’s largest data companies on the planet and yet they decided to release these into the open-source community. Understanding the real competitive advantages comes in the massive data they have collected over years to train these models. Now these powerful frameworks are part of the open-source community where an army of developers around the globe helping to improve this technology. Open-source Deep Learning and Machine Learning Frameworks will deal the final blow to AI Winter.

How AI Transformation Begins?

Understanding that AI Winter is not coming; how can you accelerate AI innovation? Here at the Big Data Beard Podcast we are covering all aspects of the AI Transformation. 

Vikram Mahidhar talks AI applications

Rohan Kumar shares his views on the future of AI

Christian Palmaz talks about AI in the vineyard

SAS and Artificial Intelligence