Will AI batteries Take Over the World?

We recently published a white paper on Battery AI. 

>> ICYMI: Download the Battery AI White Paper

While the topic has generated a lot of interest, confusion also arose.  

Here are some common responses: “Eeeh, I didn’t know batteries can use AI. I don’t know how AI works… let alone AI for batteries, can you explain?” or “AI is chatGPT, why would I want to chat with my battery?!” 

So, let’s clear that up.

No, we don’t make batteries that will pass the Turing test. They haven’t and won’t read the entire internet. They won’t try to take over the world, at least not Minority Report or I, Robot style.

But yes, we use artificial intelligence. Telemetry is at the core of our technology — the data registers the behaviors and characteristics of physical processes in the electrochemical cells, power electronics, applications, and more. We also consider how environmental parameters like temperature affect battery behaviors and longevity.

It’s like collecting statistics about traffic, retail behaviors, propagation of infectious diseases, etc., and asking an expert to interpret their implications.

Our technology ingests massive amounts of field data and seeks trends to build and hone models. After identifying a trend, such as “higher temperatures seem to cause accelerated battery wear,” the software iterates the mathematical model that connects temperature with longevity billions of times to get progressively better at predicting battery behaviors.

After establishing models with sufficient fidelity, we can extrapolate and forecast. For instance, we know heat kills batteries. But how hot? For how long? How does overheating manifest itself? Is it a gradual process? Can I measure it? How much capacity have I lost? And so on.

An AI-powered battery system can tell you point blank, “This forklift, of which you sold 14 units to a customer in Trenton, New Jersey, will deteriorate 27% faster if it were deployed in Phoenix, AZ.” The battery AI has become smart enough to use what it has learned to forecast how the cells will behave in a different context.

The process is like meteorologists using weather data to create models to inform forecasts. However, we can draw more definitive conclusions because battery data and modeling lend themselves more to computer analytics.

Yes, a robot has been obsessing over reams of data in our virtual backroom. It can tell you what a battery will do based on thousands of inputs and produce predictions faster and more accurately than any human can. 

That’s pretty smart, but since it is only obsessed with battery telemetry, it won’t make a good conversation partner for a typical human nor take over the world, enslave humans, and harvest their electrolytes.