Kris Bondi, CMO, NKris Bondi pngeura

As CMO of Neura, Inc. Kris Bondi leads the company’s efforts to increase adoption and overall mindshare.

Neura delivers customer awareness that anticipates users’ needs and preferences. Its AI engine increases engagement and retention for some of the world’s most cutting edge apps and LoT devices.

 

AI for IoT: Are We There Yet? 

If you listen to the hype, Internet of Things (IoT) environments promise to do everything from help cities navigate traffic and prevent accidents, offer doctors and medical staff real-time information as they treat patients, improve manufacturing and make our homes better understand and adapt to our personal preferences.

The buzz around IoT makes it tempting for product developers and marketers alike to promote it as a feature set of any connected device. While all of that sounds amazing, to date the development of the IoT universe has focused heavily on the hardware, sensors and connectivity required to make it a reality.

Despite this emphasis on hardware and connectivity, the reality is that IoT environments will never meet their potential without Artificial Intelligence (AI). As another trending buzzword, marketers and product developers alike can’t resist the pull to bring some level of AI functionality into everything from speakers to toothbrushes. The reality is that IoT and AI are different sides of a similar coin. IoT can’t work without AI, but how easily can we define AI for IoT?

Perhaps instead of asking what AI is inside of an IoT environment, a better place to start would be to ask “What is it not?”

The True Purpose of AI

Despite all of the hype, the ability for devices to connect to one another or the capabilities behind voice-activated commands are not actually AI. While both may be super-cool experiences that make devices more useful, neither requires any extra intelligence to perform tasks. For example, if you ask your in-home personal assistance device what the weather is like, an internet search is initiated. That if-then action has been programmed – there’s no actual intelligence taking place.

For all of the excitement that comes with each new connected ‘smart’ device, the reality is that we’re still operating predominately with ‘dumb’ appliances that do nothing more than what they are told. It’s time technology started learning how to adapt to humans, rather than the other way around.

True AI for IoT requires machine learning capable of informing the IoT device. The result is a device that, over time, learns both your usual practices and can adapt when you don’t follow your normal routine. All of this ‘learning’ really amounts to vast repositories of data, and only AI is capable of sifting through that data to make it available, understandable and – most important – actionable.

The Transition from Smart to Intelligent

Much like a seasoned adult can tell the difference between knowledge and wisdom, there is a pronounced difference between current ‘smart’ environments and ones that can become truly intelligent. As the current ‘smart’ environments take hold, it is the perfect time to leapfrog from connected devices to AI-enhanced IoT devices. The difference in the two is the ability of the devices or applications to learn how the human-element interacts with it. For example, switching from a

For example, switching from a programmed or app-activated thermostat in a smart home environment to an AI-enhanced thermostat means the difference between programming your home to change temperature or having a device that is capable of ‘knowing’ when family members are on their way home and turn on, or when to adjust for people leaving or going to bed.

If we continue with the home analogy, a more complex example of the value of an intelligent environment is illuminated during a typical bedtime routine. As the user gets in bed, AI-enhanced IoT products that have learned, understood and made predictions on his behavior can then confirm the front door is locked, the thermostat is adjusted, the appropriate music or white noise is turned on, the alarm is set, his current book is pulled up to read and he receives a reminder to take his nighttime meds.

This scenario is possible, not in some future sci-fi driven universe, but in the current here and now. It simply requires a transition in programming – from the current time-based, or rules-based model to embracing machine learning that can create API calls at precisely the right meaningful moment for the user.

Similar scenarios can play out across healthcare, connected cities, manufacturing plants and a variety of other arenas where connected intelligent devices can work smarter based on the behaviour of users. Artificial Intelligence turns the billions of data points being gathered in IoT environments into actionable insights, transitioning to the world where our devices finally offer predictive value that will significantly disrupt the way we live, work, play and navigate the world around us.

Contributed article from Kris Bondi, CMO @ Neura