Thanks to Brilliant for supporting  this episode of SciShow.

If you’re looking to start the  year by building new habits, there’s no better place to channel  all that productive energy than   with today’s sponsor, Brilliant; check them out at   Brilliant.org/SciShow. [♪ INTRO] Driving can be pretty dangerous. In the U.

S., over two and a   half million people get injured in traffic accidents each year…that’s   about one person every twelve seconds. And hundreds of thousands of those accidents   happen because of a distraction. Like, you could get distracted by   stuff going on around you, talking to another passenger,   or just changing the car’s temperature, until something unexpected happens.

So, to keep our brains on  track and our bodies safe,

scientists are turning to artificial  intelligence to learn how our minds handle distractions and help keep us in a  mental sweet spot we’re while driving. While you’re driving on a busy highway, you might be simultaneously watching out   for other cars, turning on the music, and talking to someone in the passenger seat,   all while you’re trying to maneuver multiple tons of steel going   at least 80 kilometers an hour. And juggling all those   tasks at the same time is what researchers call cognitive load or workload.

Generally, you’ll want to keep your  workload from getting too high. Otherwise, you’re less likely  to be able to respond to a sudden threat by doing things like  changing lanes to get out of the way. And scientists are trying to find ways  to measure this workload to design and engineer better cars and roads.

But this isn’t as straightforward as asking   drivers directly how they’re feeling. Because getting an already preoccupied   driver to perfectly describe how much mental energy they’re using   might affect their workload by adding to it. After all,   asking someone every couple of seconds if they’re distracted can be… a bit distracting!

But a workaround to this problem might  be looking out for physical signs that someone’s mind is working  hard, like increased heart rate, and getting an idea of the size  of their workload that way. Like, in a study from 2014, researchers  had twenty-two participants in a driving simulator doing an “n-back” test. That’s when you recall things from your memory based on a sequence of items, like numbers.

So, for example, someone says the numbers “three,   seven, five” and participants are then asked to recall which number   was read out two steps before, in this case,   “three” when the number “five” is read. The idea was to use the n-back test to   increase the workload of each driver. While this test was happening,   researchers measured participants’ heart rates and how much electricity their skin conducted.

And scientists found that both the  participants' heart rate and skin conductance increased when the task was made harder. And their driving behavior,   like how well they stayed in a lane, changed as well, but not as drastically. In another 2014 study, the same  researchers flipped the script.

They recruited even more participants  to drive on real highways while performing the n-back test  on and off at periodic intervals. Then, they used the data they collected  to train an algorithm to recognize when the participants were doing a task  and experiencing a higher workload. The algorithm used heart,  skin, and driving data for when the task was being done and when it wasn’t.

So it could tease out features in the data that helped distinguish between  high and low workloads. And it worked! The algorithm was able to determine whether a driver was or wasn’t experiencing  a high workload with 90% accuracy.

But if strapping an electrocardiogram to  measure your heart rate while you drive sounds stress-inducing,  scientists are exploring other ways of measuring workload too. For instance, scientists could use functional   near-infrared spectroscopy, or fNIRS, which shines   infrared lights near a person’s skin and measures how much light bounces back. The amount of light absorbed  by the person’s blood can tell scientists the amount  of oxygen in the blood.

And the whole setup can be made small  enough to fit into a headband, which someone could wear  so that scientists could measure the oxygen content  of the blood in their brain. If there's a lot of oxygen around,  their brain is working hard. Using the data collected by the fNIRS headband, the same researchers developed another algorithm   that could also be used to measure the driver’s workload in a   simulator while they used the n-back test.

And the algorithm worked pretty well   in this scenario too, with an accuracy of around 80%. With effective ways of measuring  a driver’s workload, the hope is that carmakers can test out  different interfaces in future cars with real drivers, to find designs  that are simple and safe to use. In fact, new cars could even include indicators  for how distracted a driver might be in real-time so that they can change  their behavior or slow down to manage their cognitive  workload while driving, sort of like a fuel efficiency  gauge, but for the mind.

All these studies covered when  drivers get overwhelmed, but being underwhelmed can be just as bad! Like when you’re on a familiar road   or when you’re simply driving for a really long time, it becomes easier to tune   out from what’s happening on the road. And as cars are being developed   to become smarter and do more of the driving for you,   the risks of boredom on the road get higher.

Because when we do less and less,

it becomes harder for our minds to snap  back and take control when we need to. To tackle this, researchers  are tinkering with ways to artificially add some workload  when drivers are understimulated. For instance, a 2017 study had participants  drive in a simulator with and without a kind of smartphone “game” that encouraged  them to stick to changing speed limits.

The phone was positioned in a legally  allowed holder, and the game was designed to avoid being too showy or distracting. When the game was introduced, the participants   drove with less risky behavior, like speeding, and stayed more   engaged on the otherwise quiet, boring roads in the simulator. Which is not to say that  playing games while driving is something that you should do,  but it goes to show boredom can affect how much attention  you’re paying to what you’re doing.

So, while some machines doing our  jobs might make us tune out, there’s a hope that soon, they’ll  also be able to help us tune back in. And if you want to learn more about the  nitty-gritty of how machines tune us in, you should check out today’s sponsor, Brilliant. They are an online learning platform   with courses about science, engineering, computer science, and math.

With their interactive course,  Algorithm Fundamentals, they can help you learn how to program  without having to dig through the weeds of coding syntax through these  fun, interactive challenges. You can shift around blocks of "pseudocode,” and then you can get immediate   feedback on your results! If you'd like to check them out,   head over to brilliant.org/scishow or click on the link in the description.

The first 200 people will get 20% off  Brilliant's annual premium subscription, and checking them out also helps us, so thanks! [♪ OUTRO]