>>> PRODUCTION FUNDING FOR "QED WITH DR. B" IS PROVIDED BY AMERICAN ELECTRIC POWER FOUNDATION.

BOUNDLESS ENERGY FOR BRIGHTER FUTURES.

AND BY VIEWERS LIKE YOU.

THANK YOU.

>> I'M DR. FREDERIC BERTLEY, IMMUNOLOGIST AND EDUCATOR.

SCIENCE IS EVERYWHERE AND FOR EVERYONE.

AND IT'S ALL AROUND US, SHAPING OUR LIVES EVERY SINGLE DAY.

IN THIS SERIES, WE'LL LOOK AT CUTTING-EDGE RESEARCH, TALK TO THE SCIENTISTS WHO ARE CHARTING NEW FRONTIERS AND SOLVING TODAY'S PROBLEMS TO MAKE ALL OUR LIVES BETTER.

WHEN A SCIENTIST OR MATHEMATICIAN DEMONSTRATES PROOF OF CONCEPT IN THEIR WORK, THEY OFTEN USE THE TERM Q.E.D.

QUOD, ERAT, DEMONSTRANDUM.

THAT ROUGHLY TRANSLATES TO, "QUITE EASILY DEMONSTRATED."

WELCOME TO "QED WITH DR.

B."

>> HEY, DR. B.

HOW ARE YOU DOING TODAY?

>> GOOD.

HOW ARE YOU DOING, DIANA?

!

!musiC@!!!musiC@!

WHAT ARE WE GOING TO TALK ABOUT TODAY?

>> WE'RE TALKING ABOUT ARTIFICIAL INTELLIGENCE.

WHAT IS THE DEFINITION OF ARTIFICIAL INTELLIGENCE?

>> YOU KNOW, IT'S A GREAT QUESTION AND A LOT OF PEOPLE STRUGGLE WITH THAT.

AND I THINK THE OPERATIVE WORD IS "INTELLIGENCE."

WE THROW THAT AROUND AS HUMANS.

WE ALL HAVE SOME SEMBLANCE OF WHAT IT MEANS.

THAT FUNDAMENTAL DIFFERENCE, HOWEVER, BETWEEN WHAT WE TALK ABOUT AS HUMAN INTELLIGENCE VERSUS ARTIFICIAL INTELLIGENCE IS WE AS HUMANS ARE CARBON-BASED SPECIES.

IF YOU'RE A LIFE-FORM, 7.7 BILLION PEOPLE ON THE PLANET, WE'RE CARBON- BASED.

RIGHT?

AND THE ONLY DIFFERENCE BETWEEN THAT AND ARTIFICIAL INTELLIGENCE IS ARTIFICIAL INTELLIGENCE IS COMPUTER-BASED, I.E.

IT'S SILICON-BASED.

BUT ULTIMATELY, HOW THEY THINK, HOW THEY ASSOCIATE THINGS, HOW THINGS MOVE FORWARD, HOW YOU PROCESS INFORMATION, NOT MUCH DIFFERENT.

WE GET OUR INSTRUCTIONS TO DO WHAT WE DO THROUGH OUR DNA.

COMPUTERS OR SILICON-BASED PRODUCTS GET THEIR INSTRUCTIONS THROUGH AN ALGORITHM THAT WE, AS HUMANS, PROGRAM INTO THEM.

>> OKAY.

WELL THEN, WHAT IS AN ALGORITHM?

>> AH, THAT'S THE MILLION-DOLLAR QUESTION.

AND THAT'S THE POWER OF THESE COMPUTERS.

AN ALGORITHM, IN ITS SIMPLEST FORM, IS JUST A SERIES OF INSTRUCTIONS.

AND IF YOU WANT A EVERYDAY, KIND OF ANALOGY, MY FAVORITE IS, MY MOM WAS AN INCREDIBLE BAKER.

YOU KNOW, SHE CAME FROM THE CARIBBEAN, SHE WOULD MAKE THIS INCREDIBLE THING CALLED COCONUT BREAD.

AND A RECIPE IS AN ALGORITHM.

IT'S A SERIES OF STEP-BY-STEP THINGS THAT YOU NEED TO DO TO PRODUCE THAT -- IN THE CASE OF A RECIPE, THAT FOOD PRODUCT, IN THE CASE OF MY MOM, THIS COCONUT BREAD.

WELL, IMAGINE YOU DID THAT RECIPE ONCE AND YOU MADE YOUR COCONUT BREAD.

GREAT.

IMAGINE YOU DID THE RECIPE FIVE TIMES.

PROBABLY BETTER THE FIFTH TIME THAN YOU WERE THE FIRST.

IMAGINE YOU MADE COCONUT BREAD 10,000 TIMES.

IMAGINE YOU MADE IT A MILLION TIMES.

HOW GREAT IS THAT COCONUT BREAD GOING TO BE ON THAT MILLIONTH TIME?

NOW IMAGINE YOU DID THAT EVERY SECOND.

AND THAT'S POWER OF AN ALGORITHM, AND THAT'S THE POWER OF ARTIFICIAL INTELLIGENCE.

IT PROCESSES INSTRUCTIONS, BUT AN INCREDIBLY FAST RATE.

AND EACH TIME IT LEARNS, EACH ITERATION, IT PROVIDES FOR INCREASED KNOWLEDGE OR INTELLIGENCE.

AND THAT'S THE POWER OF THE ALGORITHM.

>> THERE'S ACTUALLY SCIENTISTS WHO ARE DEVELOPING SOME NEW ARTIFICIAL INTELLIGENCE AND SOME NEW WAYS TO USE IT.

>> ABSOLUTELY.

IT'S SUCH A PLEASURE TO SPEAK WITH DR. HOWARD.

SHE'S THE DEAN OF THE OHIO STATE UNIVERSITY'S COLLEGE OF ENGINEERING.

SHE'S ALSO A WORLD-RENOWNED ROBOTICS AND ARTIFICIAL INTELLIGENCE EXPERT.

ARTIFICIAL INTELLIGENCE, IT'S A RAPIDLY GROWING FIELD WHETHER WE LIKE IT OR NOT.

NOT ONLY IS IT HERE TO STAY, IT'S EVERYWHERE.

CAN YOU TALK ABOUT HOW AI IMPACTS US AS HUMANS, AND MAYBE EVEN USE SOME EXAMPLES OF YOUR OWN RESEARCH IN YOUR WORLD?

>> YOU KNOW, THIS WHOLE FIELD OF ARTIFICIAL INTELLIGENCE HAS BEEN AROUND FOR A LONG, LONG, LONG TIME.

WHAT'S DIFFERENT NOW IS THAT IT'S IN APPLICATIONS THAT WE USE EVERY DAY.

IF YOU DO ANYTHING IN TERMS OF BANKING, IF YOU'RE GOING AND YOU'RE WORRIED ABOUT SOMEONE STEALING YOUR IDENTITY OR FRAUD ALERTS OR SOMEONE TAKING YOUR CREDIT CARD NUMBER AND GOING SHOPPING WITH IT.

RIGHT?

ALL THE ALGORITHMS THAT ARE BEHIND THE SCENES ARE BASED ON ARTIFICIAL INTELLIGENCE PROCESSING, YOU KNOW, KIND OF ANOMALIES, TRYING TO FIGURE OUT, ARE YOU JUST ON A TRIP TO HAWAII OR THIS REALLY AN ANOMALY OR A FRAUD?

IT'S THERE IN TERMS OF, YOU KNOW, JOBS.

LIKE THE NEXT TIME YOU APPLY FOR A JOB, GUESS WHAT, AN A.I.

ALGORITHM IS PROBABLY GOING THROUGH YOUR APPLICATION, TRYING TO FIGURE OUT IF YOU ARE THE CORRECT MATCH BEHIND THIS.

AND THEN THERE'S EVEN THINGS SUCH AS THE AIRPORTS NOW ARE USING A.I.

WHEN YOU SCAN YOUR I.D.

TO SEE ARE YOU IN THE SYSTEM OR NOT.

>> HMM.

WHEN YOU SAY "THE SYSTEM", CAN YOU UNPACK THAT A LITTLE BIT?

>> YEAH, SO ALL OF OUR DATA IS COLLECTED, RIGHT?

AND SO THERE'S MULTIPLE SYSTEMS THAT ARE OUT THERE.

THERE ARE SOME THAT ARE GOVERNMENT HOSTED, THERE ARE SOME THAT ARE HOSTED BY COMMERCIAL AGENTS -- AGENCIES THAT THEN PROVIDE INFORMATION ALGORITHMS OUTPUTS BASED ON THE COLLECTING OF DATA.

AND SO OUR DATA IS STORED IN MULTIPLE, MULTIPLE PLACES.

WE DON'T EVEN KNOW ALL THE PLACES THAT OUR DATA IS STORED.

>> SO, SHOULD WE BE EXCITED ABOUT THAT?

OR SHOULD WE BE A LITTLE WORRIED?

>> BOTH.

WE SHOULD BE EXCITED, BECAUSE THERE ARE BEAUTIFUL OPPORTUNITIES OUT THERE.

ONE EXAMPLE I'LL GIVE IS HEALTHCARE.

SO I WORK IN ROBOTICS, REHABILITATION HEALTHCARE, PRIMARILY CHILDREN WITH SPECIAL NEEDS.

IMAGINE THAT YOU'RE A CHILD, YOU HAVE TO GO THROUGH THERAPY IN TERMS OF MOVEMENT, EXERCISE.

YOU TYPICALLY GO TO THE CLINIC.

NOW IMAGINE YOU'RE IN YOUR HOME ENVIRONMENT AND YOUR ROBOT PLAYMATE, THAT LOOKS LIKE A HUMANOID, THAT IS A HUMANOID, IS BASICALLY PLAYING THE EXERCISE GAME WITH YOUR.

RIGHT?

AND YOU'RE HAVING FUN, 'CAUSE YOU'RE PLAYING, YOU'RE BEATING YOUR ROBOT, AND YOU'RE EXERCISING.

BUT YOU'RE REALLY DOING INTERVENTIONS, YOU'RE REALLY DOING REHAB.

THAT'S WHAT WE DO.

AND SO, WE LOOK AT, HOW DO MODEL A HUMAN?

HOW DO YOU ENGAGE A CHILD?

HOW DO YOU DO GAMIFICATION OF THERAPY?

AND THE ROBOT, OF COURSE, IS THE PLAYMATE.

IF I CAN COLLECT DATA ABOUT HOW KIDS WORK WITH ROBOTS, I CAN USE THAT INFORMATION TO RETRAIN OTHER ROBOTS THAT CAN WORK WITH OTHER KIDS.

SO THAT'S A GOOD THING.

>> YOU ARE AMAZING.

>> THE NEGATIVE IS IS THAT THERE'S NOT ENOUGH, KIND OF, RULES AND REGULATIONS TO PROTECT US AS A SOCIETY ABOUT THE USE OF THAT DATA.

SO, JUST LIKE THE SAME DATA CAN BE USED TO HELP IN HEALTHCARE, IT CAN ALSO BE USED TO MARKET ME AND REALIZE THAT, YOU KNOW, TODAY IS A GOOD DAY TO SHOW ME AN AD SO I'LL BUY THE NEXT ITEM FROM WHATEVER COMPANY.

SO THAT'S THE THINGS WE SHOULD WORRY ABOUT.

>> I GUESS IT CAN BE WORSE, WHERE YOU MIGHT NOT GET THAT LOAN.

OR YOU MIGHT NOT GET THAT JOB.

THAT'S A LITTLE MORE -- YOU KNOW, IN TERMS OF HAVING IMPACT ON YOU, HAS A LITTLE MORE GRAVITAS.

IS THAT AN ISSUE, OR ARE THEY -- HAS THE REGULATION FIGURED OUT THAT STUFF TO AT LEAST BLOCK SOME OF THAT?

>> NO, THAT'S STILL AN ISSUE.

LAST SUMMER, THERE WAS A CASE WHERE A.I.

ALGORITHM, FACIAL RECOGNITION -- IT CAME TO FRUITION THAT WE FOUND OUT ABOUT IT.

THERE WAS A MAN IN DETROIT, AFRICAN-AMERICAN MAN, WHO AN A.I.

ALGORITHM BASICALLY PULLED HIS PICTURE AND WAS LIKE, THIS IS -- THIS IS THE PERSON THAT COMMITTED THE CRIME.

AND HE WAS LIKE, I WASN'T EVEN IN TOWN THAT DAY.

RIGHT?

BUT IT WAS LIKE, WELL, THE A.I.

ALGORITHM CAN'T LIE.

AND SO THAT'S THE NEGATIVE, IS THAT, YOU KNOW, YOU CAN HAVE THESE IMPACTS OF OUR CIVIL LIBERTIES, OF OUR FREEDOMS.

YOU KNOW?

AND YOU CAN'T -- IT'S VERY HARD TO BASICALLY, YOU KNOW, COUNTER.

BE LIKE, HEY, WAIT -- THE A.I.

ALGORITHM IS WRONG, BECAUSE OF THE FACT THAT I LIVE IN THIS NEIGHBORHOOD THAT IT TARGETED.

IT'S REALLY HARD TO PROVE -- PROVE ANYTHING NEGATIVE, EVEN THOUGH YOU MIGHT KNOW THAT IT IS.

>> BUT DOESN'T THE PROGRAMMER PLAY A ROLE IN THAT?

I THINK WITH ONE OF THESE KIND OF FACIAL RECOGNITIONS, AT THE END OF THE DAY YOU HAVE TO HAVE INPUTS FOR THE ALGORITHM TO ITERATE OFF OF.

AND IF -- LET'S JUST PICK ON DEMOGRAPHIC -- THE INPUTS ARE ALL OF A TYPICAL WHITE, MALE 40-50 YEARS OLD, CAN THE ALGORITHM PROPERLY READ AN ASIAN, OR AN AFRICAN, OR A LATINO PERSON?

>> NO.

AND I WILL SAY THAT ONE OF THE THINGS THAT WE KNOW IS THAT -- AND WE'RE CALLING IT BIASES.

SO, BIASES IN THE ALGORITHMS.

IT'S ON THE DEVELOPER SIDE.

IT'S ON THE DATA SIDE.

IT'S ON HOW THE DATA IS CODED.

IT'S IN THE ALGORITHM ITSELF.

AND IT'S ON THE OUTPUT AND HOW IT'S USED.

AND SO IT'S REALLY SYSTEMIC THROUGHOUT THE ENTIRE ECOSYSTEM OF AN ALGORITHM.

WE CAN TALK ABOUT THE DATA.

LIKE, IMAGINE YOU COLLECT ALL THE RIGHT DATA, BUT THEN THE DON'T KNOW HOW TO CODE THE DATA RIGHT, YOU'RE GONNA HAVE SOME DIFFERENCES.

AS AN EXAMPLE, THERE WAS A STUDY THAT SHOWED CODERS ARE MORE LIKELY TO LABEL AFRICAN-AMERICAN WOMEN AS ANGRIER.

RIGHT?

AND, YOU KNOW, YOU ALWAYS HEAR THAT.

BUT NOW IMAGINE THAT DATA IS ENOUGH DATA THERE, THERE'S A BUNCH OF DATA THAT REPRESENTS, YOU KNOW, THE WORLD, BUT NOW YOUR CODERS ARE LABELING IT IN A DIFFERENT WAY.

>> SO IT DOES SOUND LIKE, THOUGH, THAT THERE'S A CONSCIOUSNESS ABOUT THAT.

AND AT LEAST THE FIELD AS A WHOLE IS STARTING TO RECOGNIZE BIASES AND SAYING, HEY, WAIT A MINUTE.

WE MIGHT NEED TO GO BACK AND RECODE OR REPROGRAM, ET CETERA.

IS THAT FAIR TO SAY?

>> I WOULD SAY THAT IT IS A GROWING MOVEMENT WITHIN THE COMMUNITY.

>> OKAY, SO WHERE IS A.I.

GOING IN THE FUTURE?

>> I THINK A.I.

IS GONNA CONTINUE TO GROW AND BE MUCH MORE PERVASIVE.

>> YEAH.

>> WHAT DOES THIS MEAN?

IT MEANS THAT WE'RE GONNA HAVE A LOT MORE SERVICES THAT IMPROVE AND HELP US, IN TERMS OF EDUCATION, IN TERMS OF JOB PLACEMENT, IN TERMS OF RETRAINING.

IN TERMS OF HEALTHCARE, IN FACT.

THERE'S INTERESTING THINGS IN HEALTHCARE AND A.I.

THAT'S THE POSITIVE.

THAT'S THE GOOD STUFF.

I WOULD SAY THE NEGATIVE IS JUST LIKE WITH LAW -- LAW TAKES A WHILE TO CATCH UP WITH TECHNOLOGY.

I THINK IN, WITH RESPECT TO A.I., THE LAW AND REGULATIONS ARE GONNA TAKE A WHILE.

AND SO WE'RE GONNA SEE A LOT OF MISTAKES HAPPEN IN THE NEAR TERM -- YOU KNOW, LIKE WITH FACIAL RECOGNITION -- BEFORE WE CATCH UP IN TERMS OF, OH, THIS IS HOW YOU'RE SUPPOSED TO DO IT.

THIS IS, YOU KNOW, THE FDA FOR A.I.

TO MAKE SURE WE DON'T DO HARM, WE DO MORE, YOU KNOW, IMPROVEMENTS AND BETTER PEOPLE'S LIVES.

SO THAT'S GONNA TAKE A LITTLE BIT OF A WHILE TO CATCH UP.

>> THAT'S DEEP AND THAT'S FANTASTIC.

DR. HOWARD, THANK YOU SO MUCH FOR BEING ON "QED WITH DR.

B."

A PLEASURE TO CHAT WITH YOU.

>> THANK YOU.

!

!musiC@!!!musiC@!

!

!musiC@!!!musiC@!

!

!musiC@!!!musiC@!

!

!musiC@!!!musiC@!

>> I REALLY LOVE THAT SHE SAID THEY COULD PROGRAM ROBOTS TO BE ABLE TO DO PHYSICAL THERAPY ALONG CHILDREN, AND THEN TAKE THAT DATA AND TRAIN OTHER ROBOTS TO ALSO DO THAT SAME THING.

SO WHAT ARE THINGS THAT A.I.

CAN DO THAT HUMANS CAN'T?

>> SEE, I LOVE THAT EXAMPLE YOU TALKED ABOUT.

BECAUSE THAT'S REALLY, KIND OF, WE TALK ABOUT THIS IN EDUCATION -- EACH ONE, TEACH ONE.

AND SO NOW ARTIFICIAL INTELLIGENCE ALLOWS ROBOTS TO TEACH OTHER ROBOTS, WHICH IS GREAT.

BUT ONE OF THE POWERFUL THINGS THAT A.I.

OFFERS THAT WE CAN'T DO IS, IT'S REALLY A COMBINATION OF TWO THINGS -- SPEED AND AMOUNT OF DATA TO PROCESS.

COMPUTERS JUST CAN PROCESS THINGS INFINITELY FASTER THAN WE CAN, AND AT A MUCH HIGHER VOLUME.

SO INSTEAD OF US BAKING SOMETHING, LIKE MY MOTHER REFERENCING HER CAKE, SHE MAY HAVE BAKED IT 50 TIMES, A HUNDRED TIMES, MAYBE EVEN A THOUSAND TIMES IN HER LIFE.

WE'RE TALKING ABOUT 100 MILLION TIMES BAKING A CAKE.

AND SO INFINITELY MORE DATA POINTS, PROCESSING THAT IN A TINY AMOUNT OF TIME.

AND THAT'S POWERFUL, AND THAT'S WHY THE OUTCOMES CAN BE TRANSFORMATIVE.

>> AND KNOWING THAT ABOUT ARTIFICIAL INTELLIGENCE, IT SEEMS LIKE THERE'S SOME BIG APPLICATIONS IN MEDICINE.

>> YEAH.

I MEAN, THAT'S THE COOL THING ABOUT A.I.

AND WE'LL GET INTO IT.

THERE'S SO MANY APPLICATIONS ACROSS THE BOARD, BUT IN THE CASE OF MEDICINE, WE HAVE FANTASTIC IMPROVEMENTS IN ALL KINDS OF FIELDS.

AND OUR NEXT GUEST, DR. REGINA BARZILAY, REALLY TALKS TO THAT BOTH ON A PROFESSIONAL LEVEL AND WITH HER PERSONAL EXPERIENCE.

>> YOU'RE A WORLD EXPERT IN ARTIFICIAL INTELLIGENCE.

SO MANY ACCOLADES, WORKING REALLY HARD.

AND THEN THERE'S A PERSONAL STORY THAT DEVELOPS IN YOUR LIFE THAT FURTHER SHINES A LIGHT ON HOW YOU THINK ABOUT ARTIFICIAL INTELLIGENCE, AND HOW WE CAN USE IT.

DO YOU MIND SHARING YOUR PERSONAL STORY, THAT CONNECTION?

>> YEAH.

SO TO GET INTO MY PERSONAL STORY -- SO IN 2014, I GOT SICK, AND I WAS DIAGNOSED WITH BREAST CANCER.

IT WAS THE FIRST TIME I WAS REALLY SICK IN MY LIFE.

AND I DIDN'T KNOW MUCH ABOUT THE HEALTHCARE SYSTEM BEFORE, AND, YOU KNOW, HOW PEOPLE AND DOCTORS MADE THE DECISIONS.

I REALIZED THAT THERE'S LIKE REALLY SO MUCH POTENTIAL A.I.

CAN DO TO CHANGE HEALTHCARE FOR ALL OF US.

AND IT REALLY MAKES SENSE TO BRING THIS VERY POWERFUL MACHINERY TO HELP PEOPLE WHO SUFFER.

YOU KNOW, IT'S OUR OBLIGATION AS SCIENTISTS TO BRING OUR SCIENCE TO THE POINT THAT IT CAN REDUCE HUMAN SUFFERING.

>> THAT'S INCREDIBLE.

SO TALK ABOUT HOW YOU LOOKED AT A.I.

AND HOW THAT HELPED IMPROVE YOUR HEALTH.

WHAT ACTUALLY HAPPENED?

>> SO, ONE THING THAT I DISCOVERED WHEN I WAS GOING THROUGH THE TREATMENT IS HOW MUCH MONEY AND LABOR IS INVOLVED.

IF YOU WANT TO TAKE OUT SOME OF THE PATIENTS THAT HAPPENED, YOU KNOW, IN THE LAST 10 YEARS AND ASK, "WHAT HAPPENED TO PATIENTS LIKE ME WHO TOOK TREATMENT 'A' OR TREATMENT 'B'?"

YOU NEED TO HAVE A HUMAN WHO WOULD GO THROUGH THE CHARTS AND EXTRACT RELEVANT INFORMATION.

IT'S JUST TRULY UNBELIEVABLE THAT IN THIS COUNTRY, WITH ITS GREAT ACHIEVEMENTS, WHEN PEOPLE, YOU KNOW, WENT TO THE MOON, IT'S STILL DONE MANUALLY.

CAN YOU BELIEVE IT?

>> WOW.

>> IT IS DONE MANUALLY.

SO THE BIGGER POINT IS WHAT A.I.

CAN DO IF YOU CAN TAKE ALL THIS INFORMATION, ALL TO MAKE EXTRACTION OF RELEVANT BEATS AND PREDICT THE TRAJECTORIES.

YOU CAN REALLY BE MUCH MORE PRECISE AND PERSONALIZED.

AND WHAT WE LEARN, YOU KNOW, ACROSS MULTIPLE AREAS OF CLINICAL SCIENCE, THAT PEOPLE BASED ON THEIR BIOLOGY, BASED ON THEIR INDIVIDUAL HEALTH, RESPOND VERY DIFFERENTLY TO TREATMENT.

AND YOU WANT TO BE ABLE TO PERSONALIZE THE CARE TO THEM.

AND THAT'S WHAT A.I.

CAN DO.

ANOTHER APPLICATION RELATES TO PREDICTING HOW DISEASE CHANGES.

AND TODAY, EVEN IF WE COLLECT A LOT OF INFORMATION ABOUT PATIENTS, THIS INFORMATION, YOU KNOW, DEPENDS ON HUMAN FOR INTERPRETATION.

IF YOU'RE LOOKING AT A MAMMOGRAM, OR ON ANY IMAGINE, YOU KNOW THERE IS LIKE A CLEAR SIGN OF LIKE A SPOT.

YOU CAN SAY, "YEAH, THIS IS CANCER."

THAT'S WHAT WE CAN SEE.

BUT MAYBE THERE ARE MANY OTHER SUBTLE THINGS THAT ARE THERE IN THE IMAGE.

OR IF YOU HAVE SEQUENTIAL IMAGES, MAYBE THERE IS SOME CHANGES THAT HUMAN EYE JUST CANNOT PERCEIVE, BECAUSE THE IMAGES ARE SO BIG AND THEY ARE SO DETAILED.

HUMANS CANNOT REALLY DO IT, NO MATTER HOW GREAT AS A RADIOLOGIST THEY ARE.

ON THE OTHER HAND, MACHINE CAN TAKE, YOU KNOW, MILLIONS OF IMAGES WITH OUTCOMES, AND UNDERSTAND WHAT IS THE LIKELIHOOD OF LET'S SAY THIS PATIENT -- WHICH NOW, WE CANNOT SEE CANCER IN THEIR BODY.

BUT WHAT IS THE LIKELIHOOD THAT WITHIN A FEW YEARS, THEY WILL DEVELOP CANCER?

SO THIS KIND OF QUESTIONS HUMAN RADIOLOGISTS, YOU KNOW, JUST CANNOT EVEN ASK.

BUT WE DEMONSTRATED WITH MACHINE LEARNING YOU ACTUALLY CAN DO IT.

>> SO THIS IS REMARKABLE.

IT SOUNDS LIKE THERE ARE TWO MASSIVE FRONTS ARTIFICIAL INTELLIGENCE CAN IMPACT OUR HEALTHCARE.

ONE IS TREATMENT.

YOU KNOW, MASS ANALYZING COPIOUS AMOUNTS OF DATA, PROCESSING IT, AND SAYING OKAY, NOW WE CAN GET MORE PERSONALIZED AND SAY THIS MIGHT WORK BETTER FOR PERSON A, THIS MIGHT WORK BETTER FOR PERSON B.

BECAUSE OF OUR CAPACITY TO CRUNCH ALL THAT DATA AND LOOK AT OUTCOMES.

BUT THE SECOND PIECE YOU'RE TALKING ABOUT IS ON A DIAGNOSTIC FRONT.

IT SOUNDS LIKE, AGAIN, THE HIGH-POWERED CAPACITY OF ARTIFICIAL INTELLIGENCE, AND THE CAPACITY TO CRUNCH SO MUCH DATA THAT WE'RE JUST LIMITED AS HUMANS, CAN ACTUALLY LEAD TO EARLY DIAGNOSES -- AND IN THE CASE OF CANCER, THAT IS CRITICAL FOR LONG-TERM OUTCOMES.

IS THAT CORRECT?

>> YOU'RE VERY, VERY CORRECT.

THAT TODAY, EVEN THOUGH YOU KNOW THE CURE CAPACITY INCREASES EVERY YEAR, WE STILL HAVE A BIG, BIG NUMBER OF DISEASES.

LIKE PANCREATIC CANCER, IS JUST ONE EXAMPLE, WHERE WE REALLY DON'T HAVE ANY CURE AT THIS MOMENT.

WHAT WE DO WANT TO HAVE IS THE CAPACITY TO LOOK AHEAD.

NOW WHAT DOES IT EXACTLY MEAN LOOK AHEAD?

WE KNOW THAT FOR INSTANCE, CANCER AND NEURODEGENERATIVE DISEASES, THEY DON'T DEVELOP AT THE MOMENT THAT YOU DIAGNOSE IT.

SO LIKE YOU'RE DIAGNOSED TODAY, AND IT WAS HAPPENING THE LAST MONTHS.

YOU KNOW, THE BODY UNDERGOES A SEQUENCE OF TRANSFORMATION FOR QUITE A LONG TIME.

AND WHAT WE DID WITH THE MAMMOGRAMS, WHICH ARE X-RAYS OF BREASTS, WE TOOK THE IMAGES OF WOMEN FOR WHICH WE KNOW THE OUTCOME.

AND NOW I KNOW THAT THIS IMAGES, EVEN THOUGH ORIGINALLY DIDN'T ANY CANCER, WITHIN FIVE YEARS THIS PATIENT DEVELOPED CANCER.

THIS PATIENT DIDN'T DEVELOP CANCER.

CAN YOU TRAIN THE MACHINE TO IDENTIFY THESE VERY SUBTLE THINGS THAT HAPPEN THERE?

AND THE ANSWER IS ACTUALLY YES.

WE'VE DONE IT AND IT WORKS VERY WELL ACROSS DIFFERENT POPULATION.

AND IT GIVES YOU A LOT OF OPPORTUNITIES, BECAUSE FIRST OF ALL, MAYBE THE PATIENT CAN HAVE SOME CHANGES IN HER LIFESTYLE.

THERE ARE ALSO CHEMO PREVENTION.

YOU CAN GET SOME, YOU KNOW, MEDIAL TREATMENT LIKE DRUGS THAT CAN HELP YOU.

YOU CAN ALSO SCREEN THEM WITH A HIGHER FREQUENCY.

SO THERE ARE LOTS OF THINGS YOU CAN DO PREVENTIVELY IF YOU KNOW WHAT'S TO COME.

!

!musiC@!!!musiC@!

[ LAUGHTER ] >> SO THOSE ARE COME VERY SOPHISTICATED WAYS TO USE ARTIFICIAL INTELLIGENCE, BUT IT'S ALSO USED IN SOME EVERY DAY WAYS.

CAN YOU GIVE US SOME EXAMPLES?

>> SURE, I MEAN, SEARCH SITES LIKE GOOGLE.

YOU'RE WONDERING, WHY ARE THEY SENDING ME THESE NICE COMMERCIALS FOR THESE BRAND NEW CHUCK TAYLORS?

THEY HAPPEN TO KNOW DR. B LIKE CHUCK TAYLORS.

LIKE, HOW DID YOU KNOW TO DO THAT?

IT'S AN ALGORITHM THAT'S PROCESSING THAT.

>> AND THERE'S ACTUALLY SOME REALLY GREAT TECHNOLOGY COMING OUT NOW WITH OUR NEXT PERSON THAT WE'RE INTERVIEWING.

>> ALEX TIMM REALLY TALKS ABOUT HOW WE USE THIS IN A PRACTICAL SENSE IN THE INSURANCE FIELD.

ALEX TIMM, SO GOOD TO HAVE YOU HERE AS PART OF "QED."

THANKS FOR BEING WITH US.

>> THANK YOU FOR HAVING ME.

>> ALL RIGHT, SO AS YOU KNOW, TODAY'S EPISODE IS ABOUT ARTIFICIAL INTELLIGENCE.

WE'RE GONNA ACTUALLY TALK ABOUT INDUSTRY, IN PARTICULAR YOUR INDUSTRY.

SO YOU'RE IN THE INSURANCE AREA, CORRECT?

>> CORRECT.

>> ALL RIGHT, AND YOU USE ARTIFICIAL INTELLIGENCE?

>> ABSOLUTELY.

SO WHAT WE BASICALLY DO IN INSURANCE IS WE TRY TO PREDICT THE FUTURE.

AND, YOU KNOW, CRYSTAL BALLS DON'T REALLY WORK, SO WE'VE GOT TO USE SOMETHING ELSE.

AND IT TURNS OUT THAT IF YOU CAN JUST INGEST A LARGE QUANTITY OF DATA, YOU CAN GET VERY GOOD AT PREDICTING WHETHER OR NOT FOR US PARTICULARLY, WHETHER OR NOT YOU'RE GOING TO GET INTO AN AUTO ACCIDENT -- >> OKAY.

>> OVER THE NEXT SIX MONTHS.

AND THESE ALGORITHMS ARE INCREDIBLY POWERFUL WHEN IT ACTUALLY COMES TO PREDICTING WHAT'S GOING TO HAPPEN IN WAYS THAT WE HAVE NEVER SEEN BEFORE.

WE USE THOUSANDS OF ALGORITHMS THAT INGEST JUST REALLY LARGE AMOUNTS OF DATA.

BEHAVIORAL DATA, DATA OFF OF A CELL PHONE.

I MEAN WE'RE TALKING -- WE'RE PULLING DATA MULTIPLE TIMES A SECOND.

WE PULL THOSE IN TO THOUSANDS OF DIFFERENT ALGORITHMS AND THEY TRY TO TELL US SOMETHING THAT'S ACTUALLY HAPPENING IN THE PHYSICAL WORLD THAT CAN THEN ALLOW US TO PREDICT WHAT'S GOING TO HAPPEN IN THE FUTURE.

>> SO, YOU USE THE ALGORITHM IN YOUR INDUSTRY, IN THE INSURANCE INDUSTRY AND IT DEALS WITH, KIND OF, GATHERING PEOPLE'S DATA TO CUSTOMIZE, PREFERABLY GIVE THEM A BETTER PLAN THAT SUITS THEM.

IF THEY'RE AN ACE DRIVER, THEY SHOULDN'T BE WORRIED.

IF THEY'RE MORE KIND OF -- NOT SUCH A GOOD DRIVER, THAT MIGHT BE A DIFFERENT ISSUE.

HOW DOES THE CUSTOMER RELATE TO THAT?

ARE THEY COMFORTABLE SHARING THEIR DATA, AND HAVING YOU MINE IT TO THEN CUSTOMIZE?

AND IS THAT WHERE THE FIELD IS GOING IN GENERAL?

>> THIS IS DEFINITELY WHERE THE FIELD IS GOING IN GENERAL.

THIS IS REALLY CUTTING EDGE RIGHT NOW.

REALLY WHERE THE CONSUMER FEELS SAFE IS WHEN YOU ACTUALLY JUST EXPLAIN TO THEM, HEY, I'M GOING TO USE YOUR DRIVING DATA TO TRY TO GET YOU A BETTER PRODUCT.

AND COMPARED TO WHAT THE INDUSTRY IS DOING TODAY, USING YOUR CREDIT SCORE, YOUR GENDER, YOUR AGE, RIGHT?

THIS IS ACTUALLY A MUCH FAIRER SYSTEM AND IT'S BETTER FOR THE WORLD.

SO THERE'S A GOOD SIDE HERE TO A.I.

>> WHILE THAT'S REALLY USEFUL, BUT IT ALSO TOUCHES UP A LITTLE BIT AGAINST PRIVACY.

SO HOW DO WE, AS KIND OF JOE AND JANE PUBLIC, HOW DO WE PROCESS THAT?

THAT IT'S GREAT ON THE ONE HAND, VERSUS THEY KNOW A LITTLE BIT TOO MUCH ABOUT YOUR PRIVATE INTERESTS?

>> THE REAL MOST IMPORTANT THING IS THAT IT'S TRANSPARENT.

AND THAT WE MAKE SURE THAT IT IS VERY TRANSPARENT HOW WE'RE USING THIS DATA.

AND THAT THERE ARE REGULATIONS AROUND THAT.

SO THAT YOU'RE NEVER SORT OF GIVING UP INFORMATION FOR A REASON THAT YOU DIDN'T KNOW YOU WERE GIVING IT UP FOR.

WHEN IT'S NOT TRANSPARENT AND OBVIOUS, THAT'S WHAT WE HAVE TO PREVENT GOING FORWARD.

>> OKAY.

AND DOES THE POLITICAL SECTOR, IS THERE ENOUGH KIND OF GUARD RAILS TO MAKE SURE THAT'S BEING RESPECTED?

OR IS IT KIND OF, WE'RE BUILDING THE PLANE AS WE'RE FLYING IT SITUATION?

>> IT'S A BUILDING A PLANE AS WE'RE FLYING IT SITUATION, AND WE'RE BEHIND.

THE REGULATION IS BEHIND AND THESE THINGS ARE ACTUALLY MUCH MORE COMPLICATED THAN I THINK WHAT OUR REGULATORY SYSTEM IS REALLY SET UP TO UNDERSTAND.

>> TALK TO US ABOUT THAT LINE THAT AN ALGORITHM AND ARTIFICIAL INTELLIGENCE, AND HOW THAT MIGHT IMPACT SOCIETY -- WELL, FIRST HOW IT MIGHT IMPACT THE BUSINESS SECTOR IN GENERAL, AND THEN BROADER THAN THAT, HOW IT'S GOING TO IMPACT ALL OF US.

>> SO THE BIG THING ABOUT HOW IT'S GOING TO IMPACT THE BUSINESS SECTOR IS THERE'S GOING TO START TO BE THE A.I.

HAVES AND THE A.I.

HAVE-NOTS.

>> SAY MORE.

THE A.I.

HAVES AND THE HAVE-NOTS?

>> THE A.I.

HAVE-NOTS.

THE A.I.

HAVES ARE GOING TO BE THOSE THAT ARE LEANING IN, GATHERING A LOT OF DATA.

AND THE REALLY REMARKABLE THING IS, THE MORE DATA YOU HAVE THE BETTER AND BETTER YOU GET.

>> HMM.

>> AND SO THERE'S A NATURAL MOAT TO THESE BUSINESSES WHEN THEY BEGIN TO COLLECT MORE AND MORE DATA.

>> AND JUST TO CLARIFY, WHEN YOU SAY THE BETTER AND BETTER WE GET, YOU MEAN WHATEVER WE'RE TRYING TO DEVELOP GETS STRONGER AND MORE POWERFUL OR -- >> THE BETTER THE ALGORITHM GETS.

>> OH, THE BETTER THE ALGORITHM GETS.

>> YES.

SO THE ALGORITHM GETS EXPONENTIALLY BETTER AS YOU GET MORE AND MORE DATA.

AND WE'RE NOT TALKING ABOUT SMALL AMOUNTS OF DATA.

WE'RE TALKING ABOUT GIANT AMOUNTS OF DATA.

>> SURE.

>> AND SO ONE YOU GET TO THAT SCALE, IT BECOMES VERY DIFFICULT FORT HE A.I.

HAVE-NOTS TO EVER COMPETE.

BECAUSE AT THIS POINT THE ALGORITHM IS SO GOOD, AND YOU DIDN'T START EARLY ENOUGH THAT YOU'RE GONNA GET BLOWN OUT OF THE WATER.

>> THAT'S REALLY COOL.

AND SO YOU'RE AT THE CUTTING EDGE OF THAT FOR THE INSURANCE WORLD.

IS THAT GOING TO APPLY AND SPILL OVER INTO OTHER SECTORS AS WELL DO YOU THINK?

>> ABSOLUTELY.

I THINK INSURANCE IS RIPE FOR THIS KIND OF APPLICATION, BUT BANKING IS PROBABLY RIGHT BEHIND IT, AND OTHER SEGMENTS WILL FOLLOW.

!

!musiC@!!!musiC@!

!

!musiC@!!!musiC@!

>> WHERE WOULD YOU LIKE TO SEE ARTIFICIAL INTELLIGENCE GO IN YOUR LIFETIME?

>> YOU KNOW, THAT'S A GREAT QUESTION.

THERE'S SOME FUNDAMENTAL THINGS THAT STILL IMPACT US AS HUMANS ON THE PLANET.

BIG ISSUES LIKE INFECTIOUS DISEASES, POVERTY, EDUCATION EQUALITY AND CLIMATE CHANGE.

THESE ARE FUNDAMENTAL THINGS THAT WE'RE STILL STRUGGLING WITH AS HUMANITY TO ADDRESS.

IMAGINE AN EMPATHETIC, INCREDIBLE ARTIFICIALLY INTELLIGENT SPECIES OR BEING THAT COULD HELP US SOLVE THOSE PROBLEMS TO MAKE THE WORLD -- YES, I'M A CORNY SCIENTIST -- TO MAKE THE WORLD A BETTER PLACE.

>> I LOVE IT.

SO REMEMBER, THAT'S ALL QUITE EASILY DEMONSTRATED.

>> "QED WITH DR. B", THAT'S ME.

FIND US ON FACEBOOK, TWITTER AND INSTAGRAM.

AND WE'LL SEE YOU NEXT TIME.

YOU KNOW, A LOT OF TIMES AS SCIENTISTS OR ENGINEERS, WE TALK ABOUT THAT "A-HA" MOMENT.

ALONG WITH KNOWING THAT YOU WANTED TO BE A ROBOTICIST SLASH INSPIRED BY THE BIONIC WOMEN -- WHICH IS TOTALLY COOL, BY THE WAY -- DO YOU HAVE ANOTHER "A-HA" MOMENT YOU CAN SPEAK TO IN ARTIFICIAL INTELLIGENCE OR ROBOTICS?"

>> THE "A-HA" MOMENT WHEN I REALIZED THAT PEOPLE ARE SPECIAL AND THAT I ACTUALLY LIKE PEOPLE -- WHICH, YOU KNOW, I DIDN'T REALIZE I ACTUALLY LIKED PEOPLE.

I HAD GADGETS.

BUT I LIKE PEOPLE WAS -- I WAS DOING A WORKSHOP, DOING A CAMP.

AND THERE WAS THIS ONE YOUNG LADY WHO WAS PART OF CAMP AND HAD A VISUAL IMPAIRMENT.

AND I REMEMBER SITTING THERE AND I'M GOING LIKE, "OH, THIS IS HOW YOU DO IT."

AND SHE'S LIKE, "THERE'S NOTHING ON THE SCREEN FOR ME.

THERE'S NOTHING THERE."

AND I REMEMBER SITTING NEXT TO HER -- AND SHE WAS SO BRIGHT AND JUST WORKING THROUGH IT.

AND AT THAT MOMENT I WAS LIKE, "YOU KNOW WHAT?

I CAN ACTUALLY USE MY ENGINEERING TALENT TO DO SOMETHING."

LIKE, TO DO SOMETHING THAT COULD HELP THIS PERSON THAT I'M SITTING NEXT TO IN A PROFOUND WAY.

AND THAT WAS THE "A-HA" MOMENT THAT I NEEDED TO USE MY ENGINEERING TALENT FOR PEOPLE THAT WERE IN MY BACK YARD.

AND THAT'S WHEN I ACTUALLY LITERALLY SWITCHED MY FOCUS OF RESEARCH FROM, LIKE, SCIENCE-DRIVEN ROBOTICS GOING TO THE ARCTIC, TO WORKING WITH CHILDREN WITH SPECIAL NEEDS.

I MEAN, LITERALLY IN LIKE A YEAR TIMEFRAME.

>> WHAT ARE YOU TOP FIVE A.I.

MOVIES?

>> GOTTA TALK ABOUT THE TRILOGY OF NONE OTHER THAN "TERMINATOR."

THEN THERE'S "THE MATRIX" SERIES.

"WALL-E." "I, ROBOT."

>> WHAT AM I?

>> "BLADE RUNNER."

>> PRODUCTION FUNDING FOR "QED WITH DR. B" IS PROVIDED BY -- AMERICAN ELECTRIC POWER FOUNDATION.

BOUNDLESS ENERGY FOR BRIGHTER FUTURES.

AND BY VIEWERS LIKE YOU.

THANK YOU.

!

!musiC@!!

!musiC@!