The Machine Learning Inflection Point

From conversations I have regularly with execs at many of the leading tech companies today, there is no more used phrase than “machine learning.” Some people say AI, and they mean machine learning they just don’t know it. The reality is nothing we call smart today like smartphones, smartwatches, computers, smart speakers, smart lights, smart home, smart cars, etc., is smart. All our so-called smart tech is pretty dumb. A phrase I’ve been using lately is that most our technology is connected, not smart. Machine learning is the one technology that will take our technology from being connected to being smart.

Why Now?
The science, and math, behind machine learning, is not new. However, recent breakthroughs in the deep learning algorithms paired with advancements in semiconductor performance have finally put the industry in a positon to take advantage and progress machine learning in ways not possible before. We can finally teach machines to learn. This will have a profound impact on every aspect of technology.

Computers that can see and hear us are now on the horizon. Nearly all major machine learning up to this point has been text-based. Even that was rudimentary at best compared to what is possible today with big data analysis and data-drove computer training models. Laying the framework of computer vision so machines and finally have eyes was the last step. Tying text, voice, and visual data together is the inflection point we have been waiting for that is finally here.

The applications for machine learning as a new foundation for technology has far-reaching applications. In automotive is the core technology that will bring fully autonomous vehicles to market. For public safety, it can be used to detect criminal activity or public safety threats, and help law enforcement better reacts and respond to issues around public safety. In retail, imagine never having to pay for anything with a payment mechanism again as retail checkout systems recognize you and charge your payment method on file. Even our devices in our homes and on our person will get more useful as machine learning invades them as well.

Your Personal Assistant
The biggest area where we, normal consumers, will see machine learning in action is on our personal computers. As these devices learn more about us, their value will increase. Like I mentioned above, our computers will move from connected to being smart, learning about us, adapting to, and becoming even more personal than they are today. This is the framework that will lead to our computing devices becoming true personal assistants.

My mind started thinking more about this as I’ve been using the iPhone X. Face ID is a bit unique at the moment in that it learns about your face over time. Meaning the more you use it, the more accurate it becomes. This is necessary for facial changes, glasses, hats, facial hair, etc. I noticed, for example, certain faces I could make like extreme grins or frowns would cause Face ID to not work the first time. Then after logging in with my pin, which tells the iPhone it’s me, I noticed the second time I tried logging in with Face ID using those same faces that did not work caused it to work every time. I assume this is because it learned those facial variations. I’m training my iPhone to recognize me and a subtle technology Apple calls Attention Awareness could be the platform for much deeper personalization.

Attention awareness is, at the moment, a technology designed to tell the iPhone you are paying attention to the screen. The iPhone uses this when you log in with Face ID to know you are looking at the screen. It also uses this to reveal certain notifications that are hidden on the home screen once it knows it is you and you are attentively looking at the screen. The iPhone X also uses Attention Awareness to not turn your iPhone screen off if you are looking at it. Subtle, yet helpful little details. However, what if over time, the iPhone can not just detect my attention but know exactly what I’m looking at on the screen. Perhaps scrolling automatically when I’ve reached the bottom of the page as a potential use case. Or opening apps automatically when I look at them for a set period. Perhaps more interesting is if the iPhone can detect my emotion and have Siri assist accordingly. It’s hard to imagine where these use cases can go but the inevitability of our computers using their sensors like camera, microphone, etc., will all play a role in how they learn, adapt, and customize themselves to be as helpful as possible to their owners.

Privacy and Security at the Center
As this machine learning inflection point leads to more innovations in hardware, software, and services, it will go nowhere without deep security and privacy. I’m sure many of you reading this were creeped out, or worried, about these use cases and concerned that they could lead to a violation of your privacy. Machine Learning has tremendous promise but it can also become malicious. It can lead to great value to consumers but only if they trust it enough to let it learn about them, and deliver deeper personalization. That trust will only come if they believe their privacy is protected and the information the computer learns is secure. There can be no compromise here with privacy and security. There is no middle ground, only private and secure and not private or secure.

This issue will be the single biggest thing that can thwart or hold back innovations around machine learning.

Published by

Ben Bajarin

Ben Bajarin is a Principal Analyst and the head of primary research at Creative Strategies, Inc - An industry analysis, market intelligence and research firm located in Silicon Valley. His primary focus is consumer technology and market trend research and he is responsible for studying over 30 countries. Full Bio

2 thoughts on “The Machine Learning Inflection Point”

Leave a Reply

Your email address will not be published. Required fields are marked *