I recently had the pleasure of speaking at the IoT North America conference in Chicago. This was my second time speaking at this event, and as always I was truly inspired by many of the presentations and meeting many of the IoT thought leaders.
My presentation was titled Artificial Intelligence and IoT: Making Things Smarter. It centered on the the following premise: Data is most valuable when it is new. With new data, take action and go.
Artificial intelligence is here to stay. It started in 1955, initiated by John McCarthy, who became the father of AI. From industrial automation to AI applications sponsored by DARPA, and on to smart toys, autonomous cars and smarter homes, there have been significant contributions to AI within a short period of time. Although there is further research and development needed in the AI space, it is already ready for prime time in the Internet of Things.
Let’s take a look at 3 key takeaways from the presentation.
IoT is not enough.
Artificial intelligence (AI) has now become a must in enhancing the overall user experience with IoT devices and services. With billions of data points potentially generated from IoT devices, just having a dashboard of this data is not enough. Artificial intelligence solves problems of taking action on all the data that is generated.
For example, a smart home solution will have limited use if each appliance needs to continuously be told to take action. Integrating AI would significantly enhance the experience by enabling it to automate more, and take less action. Anyone looking to make use of the data in IoT can increase its expected business value by adding AI.
It is even easier to build your own AI engine.
Back in the days before GitHub, open source and Stack Overflow came into play, we had to build everything from the ground up. With the development and creation of artificial intelligence APIs, adding AI to your IoT solution has become easy for developers who are not experts in AI.
Bons.ai offers APIs that abstract away the complexity of machine learning, and make the programming and management of AI models more accessible to developers. It takes a pedagogical approach, in which you build out curriculum using Inkling and teach your AI brain to learn from and immediately take action on that data. craft.ai, meanwhile, is a machine learning API that delivers actionable decision models from each user behavior and context history in real-time. Artificial intelligence can be applied immediately with very little data.
The possibilities of AI are just beginning.
There are three types of AI minds: artificial narrow intelligence, artificial general intelligence and artificial super intelligence. Artificial narrow intelligence has the capability of performing one task repeatedly. For example, IBM’s Deep Blue chess computer was considered a “one task” machine, although it did surpass human intelligence in defeating the world champion Garry Kasparov in 1996. Its AI software had the capability of evaluating 200 million moves a second! Another example of a machine having narrow intelligence is the Unimate industrial robot developed in 1961 by GM. It was the first industrial robot and performed the task of automating the manufacture of picture tubes.
Handling more than one task, as well as having more “human-like” capabilities, would be considered artificial general intelligence. According to AI researchers, it will most likely take several decades before we can build AGI applications and machines. Today, any solution that comes close to AGI needs human interaction.
The artificial super intelligence surpasses the ability of the artificial general intelligence and is considered to have “Terminator”-level capabilities. We have a long way to go to reach artificial super intelligence in our applications and devices. With human interaction and engineering efforts, however, AI already has the capability of supercharging your IoT solution by providing process and action on making the experience better.
If you are working in IoT, you have likely amassed loads of data from your devices and sensors. Rather than sitting on that data, try putting it to use with some artificial intelligence APIs. Build your own AI “brain” into your IoT solution, and you will be amazed to see how you can make things smarter!