By Nick Chase, Head of Technical and Content Marketing for Mirantis

Hang on tight, 2020 will take us to the top of the roller coaster

In 2019, we saw the tech industry try its hand at a number of trending technologies, but hang on to your hats folks, because 2020 will see a number of those trends begin to come to a head -- and not always in a positive way.

As with the past couple of years, everything in the tech community will be affected by the growth of Machine Learning and Artificial Intelligence, even if the direct link doesn't seem immediately obvious.

Two years ago I said that user-friendly toolkits like Tensorflow would lead to a massive uptick in developers creating ML and AI solutions, and that's come true. Following on from that, this year we will see more complete solutions being offered. Using a ready-made API for, say, image recognition is already possible, but in 2020 we'll see developers using solutions like Google's bot to have a conversation on the phone. In fact, this is already happening with frighteningly convincing spam calls, and at some point during the year we'll see a successful Turing Test, upending the notion of what's acceptable -- and legal.

I also said that hardware vendors would gear up for machine learning, and Intel just spent $2 billion buying Israeli AI chip maker Habana, just a couple of months after Google claimed "quantum supremacy" -- where a quantum computer can perform tasks a regular computer cannot. This means a couple of things for tech in general.

Most obviously, look for multiple developer-accessible quantum computing services, if only on a very basic scale. More subtly, vendors will begin to take advantage of these capabilities, though we probably won't see much of it come to fruition until late 2020 or 2021. Until then, watch for vendors to make overblown claims and try to support them later.

One place where I'm pulling back a little is with regard to machine learning in the data center; I still believe we'll see machine learning-enabled configuration and operations, but at the moment, most companies don't have enough data -- or more specifically, enough clean and usable data -- to create the models machine learning needs to be effective. Look for this to start to pick up in 12-24 months.

The millions of Internet of Things devices have the potential to provide a lot of that data needed to train machine learning models -- particularly with the arrival of 5G -- but they suffer from some problems:

1. There is a lack of standards for how they communicate and collect data
2. There is a lack of privacy standards that will finally start to get noticed in the next 12-18 months as large-scale hacks begin to happen because...
3. There is little to no security, and unprotected devices such as freezers and vacuum cleaners live on networks with personal devices such as laptops
   

Meanwhile, ethics and transparency will become more crucial as the destructive nature of deepfakes begins to really sink in and rather than being an amusement the general public will realize this is a problem as it becomes almost trivially easy to replace the dialogue in a video, or worse. Expect one or more incidents during the US election season in which either a fake damaging video is created by an opponent or a real damaging video is claimed to be fake. 

As the Chinese "social credit" system, in which users are constantly surveilled and even the actions of their friends and families has bearing on whether they can do something as basic as book a train ticket, becomes more well known, the rest of the world will begin to realize what's at stake, and users will begin to stake out positions on both sides.

This will lead to a greater push for "explainable AI", and we'll see advances around mid-year, as it becomes more and more crucial to understand how algorithms are making decisions.

All of this will lead to an environment where control becomes more important, and we'll see a pullback from "nebulous" infrastructures such as serverless computing, as it becomes more important to know exactly what's running and where. That said, the emphasis on multi-cloud architectures will pick up steam as companies become less and less willing to be tied to a single cloud vendor.

A natural outgrowth of the rise in multi-cloud will be a corresponding rise in Edge Computing, which focuses on putting processing power closer to where the work actually needs to be done. Adoption will be slowed, however, by the need for companies to change their entire way of thinking in order to best take advantage of it. In 2020 we'll see at least one massive Edge cloud debacle, but towards the end of the year and the beginning of 2021 companies will finally start to get the hang of it and we'll see Edge gradually move towards being the default over the next two years.

All in all, in 2020 industries will try to tool up and scale up for technologies that have been around for a few years, but as science projects. Now that they're serious contenders, look for companies to struggle at staffing up for them, but soldier on because they can no longer be ignored. They won't always get it right, but as the pace of change picks up, the price for standing on the sidelines will be greater than the price of failure.

##

About the Author

Nick Chase is Head of Technical and Content Marketing for Mirantis, and is deeply involved with cloud computing using Kubernetes and OpenStack. A former release team member for Kubernetes, he is a frequent speaker on technical topics and author of hundreds of tutorials and over a dozen books, including Machine Learning for Mere Mortals.