Human to AI interactions have become pretty commonplace with the ubiquity of natural language processing (NLP) devices and systems. I can ask my home assistant to set a timer, purchase things, get me a cab, or tell me a joke. I can ask my phone to get me directions or send a text for me. Being able to do these things hands-free demonstrates the power of Natural Language Processing to understand, analyze, and generate human speech.
It isn’t perfect yet. Most of us have been frustrated by a customer service phone system not understanding a word we’re saying. But over time, with more and more training data being fed into these systems, results have gotten dramatically better. And answering customer service calls is only the beginning of the benefits AI will bring.
NLP is how machines learn to understand humans, but thinking that is the end goal of this type of AI is very limiting. The future and the real power of these systems is when AIs are talking to AIs. This will introduce a whole new training problem where algorithms written by one company or municipality will have to be able to integrate and negotiate with countless other algorithms written by other companies or municipalities.
A great example of what is possible once AIs are talking to each other is a call to 911 that has to be routed quickly and correctly. Using NLP and AIs talking to AIs, a complex system with access to everything from traffic lights to hospital ERs will be a seamless part of the process. With a smart 911, you’ll know that your emergency call will get through to a human and that resources will be secured. Because the AI recognize through sentiment analysis that you are in real danger or pain, there will be less fear that you are caught in a queue behind someone’s toddler who called 911 while playing games on daddy’s phone.
Today, your call to 911 looks something like this:
There are irreplaceable human beings on the other end of the phone line who calmly, rationally, and with much compassion, help people in their time of need. But there are all sorts of systems and data available, that if connected and communicating with one another, could make those invaluable humans response times faster and result in more people being helped. With AIs talking to each other the right crew gets dispatched, the light cycles are synced to accommodate the emergency vehicles, the ER is ready, and so on. The results of these integrated systems could save lives when applied to a real emergency.
For instance, you have what you think is a heart attack and call 911. Your call could go something like this, with the same “time to human”:
It takes a lot of time and a lot of data to train an algorithm to understand whether the person on the other end of the call is in serious jeopardy or if it is a butt dial. Fortunately (unfortunately?) there is a lot of data available. 911 call centers take calls and texts from people reporting things and requesting assistance. All of this data can be used to train systems to understand sentiment - tone, syntax, content, and urgency. With the addition of other AIs providing time-saving services, like an ambulance being dispatched, you can see how AI to AI interactions can save lives.
FUN FACTS: 2018 was the 50th anniversary of the introduction of 911 as an emergency number. As of 2017, 911 call centers received around 240 million calls each year at a current estimated cost of over $12b nationwide.