Be Smart

(peaceful music) (Joe sighs) - I shouldn't do this.

I'm just gonna take one quick peek at the comments.

Okay.

Oh.

"I always learn so much from Joe."

That is so nice.

"Is it me or is he getting old?"

Old!

You're gonna die!

You're gonna die!

Hey, old man.

You're ancient.

Old.

Death, death, death.

(Joe screams) (blender whirring) Hey, smart people.

Joe here, but for how long?

I'm just trying to cheat death with this life-extending smoothie recipe.

I found it on TikTok.

(Joe gulps) Tastes like youth and burning plastic.

Let's face it.

We're all a little bit like Barbie.

We think about death a lot.

According to a 2022 survey, half of all Americans think about death monthly, and one out of three Gen Z'ers thinks about death daily.

Should probably start doing these without the box.

Death anxiety, or thanatophobia, is a perfectly natural human feeling.

'cause, let's face it, being alive is pretty cool when you consider the alternatives.

The global health and wellness market is estimated at $1.8 trillion.

Yes, that's trillion with a T. Truth is, no matter how hard we try to cheat death, it could happen at any moment.

We can't predict our death, or can we?

Come on, folks.

This is a science show.

You didn't really think that I was gonna.

Right now, there are people out there predicting your death and mine, distilling our lives into data points, feeding it into lifeless machines, and calculating with an uncanny level of accuracy when someone exactly like you or me is gonna die.

I'm talking about predictive analytics.

(lively theme music) Predictive analytics is a branch of mathematics that uses historical data to make predictions about future outcomes, and it's everywhere.

Shopping, sports, social media algorithms, fraud detection, politics, and deciding if you'll see this YouTube video, because if a government or business can know what's gonna happen before it happens, that's pretty useful.

It turns out we've been using math to predict people's deaths for centuries.

By the 1600s, humans were shipping goods around the world and you could make serious bank doing it as long as your ship didn't sink.

Captains and the people who funded their voyages had more to worry about than just weather.

The late 1600s were also the golden age of piracy.

To the Lloyd's Company of London, these hooks-for-hands hooligans looked like an opportunity.

They started crunching numbers, using past data to help predict how dangerous a particular sea voyage would be.

Then, Lloyd's would offer insurance to help cover the risk of the trip.

The more risk the calculations showed, the higher the insurance would cost, and lo and behold, the insurance industry was born.

Today, Lloyd's of London is one of the largest insurers in the world, and the predictive analytics they pioneered are still used to predict risks and outcomes today, only now it's powered by computers and they're good at it.

If you're thinking, "Hey, I'm a complex and free will-having snowflake.

You can't predict me."

Think again.

- We tend to think that we're pretty unpredictable, but if you think about it, if I were to guess for you or almost anyone else where you'll be at 4:00 AM tomorrow or a month from now, or a year from now, you're gonna be at your house in bed, and it's every single day.

Where are you gonna be in the daytime?

Well, you're gonna be at your work, but that's one example of where we have a lot of predictability, but we're kind of blind to it.

- That's right.

Every day, you leave invisible breadcrumb trails of data and behavior that you don't even think about.

Like, you might have apps on your phone that track how many hours you slept last night, or a metro card you use to catch the bus, and you ordered coffee on the way to the bus stop.

I hate to tell you this, but somewhere out there, somebody knows about all the websites you've visited.

Yes.

Even that one.

That's all data, and it turns out, so are we.

- The idea, Joe, is that, you know, any one person is just a data point.

Probabilities of mortality or longevity are gonna play out for that person individually.

- In order to make predictions about us walking human data points, the computers have to pile us together and create a sort of statistical Franken-human that represents the whole bunch.

This is the mathematical theory called the law of large numbers.

Basically, the larger your data sample is, the more likely it is that the average of that sample will reflect what actually happens.

- Again, the benefit might be in studying, you know, 10,000, 100,000 people who have potentially some similar characteristics to you.

They might be of a similar age.

They might be male.

They might have a general same health and wellness.

- This is where things get really complicated.

In order for us to get accurate predictions of the future based on past data, we first have to figure out all the potential outcomes that could happen around an event.

Here's how predictive analytics works in the simplest terms.

Say I have a bag with 20 marbles in it, where some are red, some are yellow, and others are blue.

If I pull one marble out, I can't accurately predict which color I'll grab, but if I were to draw 100,000 times, I could calculate the likelihood of anyone pulling a particular color with extreme accuracy, as long as I don't lose my marbles first.

To be even more accurate with our prediction, we could even start factoring in other data like the weight or size of different marbles and how that may affect their distribution in the bag.

The point is that multi-factored data, considering all of the different factors and how big or small their influence is on the outcome, that can improve our ability to predict the future.

So marbles are great, but when are we gonna die?

There are so many potential factors we have to predicatively analytic-size.

Have a chronic illness?

Love scuba diving?

These are potentially negative factors.

Exercise regularly?

Eat well?

Got access to good healthcare?

Looking at you here if you're up in Canada.

Well, these are all positive factors in the mathematics of mortality, but some of these things are more likely than others to put you in a speedboat across the River Styx, so some factors get more weight in different scenarios.

If that sounds like there are an almost overwhelming number of factors to consider when predicting the future, you're right.

The future is complicated.

At least, I think it will be.

That is why scientists are using machine learning to look at more and more complex factors and figure out which ones are actually important.

When they said AI was gonna be responsible for our death, I don't think this is what they meant.

So what kind of data do you need in order to construct an algorithm like this?

- So if you're predicting how long someone will live, you look at their age and lots of other properties that the insurance companies, you know, have tallied out, but the algorithm that we do, we do something different.

We basically say to you, we're gonna put your whole life in the mathematical model, and then the model will tell what's important.

So you can put in lots of stuff that you might not think was important, but that the model will then learn that actually is one of the things that tells you something about our future behavior.

- [Joe] Typically, human analysts would select the factors that they think are likely to predict some outcome, and they'd test how much weight they should be given in the calculation, but what factors are selected or not selected may be affected by human bias.

That machine-learning algorithm instead feeds all possible factors into the system and lets it select and weight factors, free from human bias.

The algorithm analyzes a person's life the way a large language model analyzes words.

Where a language model calculates patterns of words that are likely to be associated with each other and uses those to create future language, Sune's mortality model looks for patterns of behavior and demographics that are likely to be associated with death and instead of language, it writes the story of a life.

- If you look at, let's say, income, it would say that, if all other things are equal, if we kind of take you and increase the income for your data point, then you have a higher probability of surviving, and that lines up with what we know from existing social science, that if you're wealthy, you basically have a better chance of living a long life.

- They first trained this model on a large multi-factor data set pulled from health and demographic statistics in Denmark and compared this to actual death records to gauge its accuracy.

They then tested the model by feeding it a set of people where half survived and half died.

If we were to randomly predict if a particular one of these people survived, we would expect to get the answer right 50% of the time.

Their AI predictive mortality model was able to guess right 8 out of 10 times.

Right now, this AI mortality model is being used as a research tool to create better models in the future so I can't ask it when I'm gonna die.

So I decided to ask an actual actuary.

Well, as an actuary, have you ever missed a flight?

- I think I am 100% on making my flights.

- So I sent Dale a bunch of information about me, like my height, age, and some of my habits.

All good ones, mind you, and Dale crunched the numbers.

- I'm gonna estimate, Joe, that you're around, you know, say, 40 years old, give or take.

- That's a good estimate.

It's fine.

It's close, yeah.

- And put that in there.

I'm going to then select that you're a male.

You do not smoke.

So I have, you given this longevity illustrator, to be around a life expectancy of 86.

You have a 37% probability of actually living to age 90.

You also, by the way, have an 8% chance of living to age 100.

- This is the best news I've heard all day.

I thought you were gonna say like 70, 75, something like that.

- Well, remember, some of those life expectancies that you hear quoted are life expectancies at birth, and so you've had the benefit of surviving the first 40 or so years and past some of the hazards or risks that might unfortunately lead to some early deaths, and so I would encourage you to do a little bit of thinking of, all right, what are some of the planning I might want to do should I live to that age?

- This is fantastic.

My gym membership has gotta be the greatest investment I've ever made in my entire life, and I hope all the YouTube commenters are listening.

You hear that?

I don't look old.

Okay, anyway.

And even with all the data in the world, there will still be outlier events that we could never see coming, so-called black swan events that are moments of totally unpredictable chaos.

That said, these things are accurate, almost scary accurate.

As for me, I'm glad that I met with Dale and that he gave me a number.

As a scientist, I love numbers, and as a person who's alive, I love that I'll probably get to stay that way for a long time.

The most important thing I learned is that, even though the mathematical tools that predict our lives and our actions are uncannily accurate, we still have power to make choices that can change those predictions, to leave new breadcrumb trails of data that might lead to different destinations.

At the end of the day, all of us only have a little time on this blue rock we call home.

Math and science and predictive analytics can help us make the most of it.

At the very least, it'll suggest some good videos to watch while we're waiting.

Stay curious.

"Hank Green is my favorite."

I'm not.

I'll take it.

- [Producer 1] Those are great.

- [Producer 2] Yeah.

(crew chuckles)