Read Transcript EXPAND
>>> NEXT ISN'T SOMETHING THAT HAS THE POTENTIAL TO INFLUENCE SEVERAL OF THE ISSUES WE HAVE DISCUSSED IN THE SHOW SO FAR.
FROM CLIMATE CHANGE TO PRESIDENTIAL ELECTIONS.
AND THAT IS ARTIFICIAL INTELLIGENCE.
DENNIS IS THE FOUNDER AND CEO OF ONE OF THE WORLD LEADING RESEARCH GROUPS, DEEP MIND.
HE TELLS WALTER ISAACSON ON WHY HE TAKES AN OPTIMISTIC APPROACH TO THE TECHNOLOGY.
>> THANK YOU.
WELCOME TO THE SHOW.
>> THINK YOU FOR HAVING ME.
>> IN OFFICE THERE IS A COPY OF ALAN TURING'S PAPER WHERE HE ASKED THE QUESTION, CAN A MACHINES THINK?
NOW WE HAVE A LOT OF LARGE LANGUAGE MODELS, SUCH AS GOOGLE GEMINI, WHICH YOU HELPED CREATE, CHATGPT FROM OPENAI.
HOW DO WE GET FROM A CHAT ABOUT THAT KIND OF CAN PASS THE TURING TEST, FULLY PERSON INTO SOMETHING THAT IS REALLY SERIOUS, LIKE ARTIFICIAL GENERAL INTELLIGENCE, AGI, WHAT YOU WOULD CALL THE HOLY GRAIL?
>> GREAT QUESTION.
THERE HAS BEEN UNBELIEVABLY FAST PROGRESS IN THE LAST SECOND AND A HALF, GETTING TOWARDS THE SYSTEMS THAT WE HAVE TODAY THAT CAN PASS A TURING TEST.
BUT IT IS SO FAR FROM GENERAL INTELLIGENCE.
WHAT WE ARE MISSING ARE THINGS LIKE PLANNING, MEMORY, AND TOOL USE SO THEY CAN ACTIVELY SOLVE PROBLEMS FOR US AND ACTUALLY DO TASKS.
RIGHT NOW WHAT WE HAVE IS PASSIVE SYSTEMS.
WE NEED THESE ACTIVE SYSTEMS.
>> Reporter: EXPLAINED TO ME WHAT PLANNING IS.
I KNOW YOU AND I DO IT, HOW DOES A MACHINE DO IT?
>> WE HAVE EXPERIENCED A LOT IN THE PAST WITH PLANNING USING GAMES.
ONE OF OUR MOST FAMOUS PROGRAMS BACK IN 2016 WAS OUT FOR.
A PROGRAM WE BUILT TO BEAT THE WORLD CHAMPION AT GO, THE AGENT GAME OF GO.
WE BUILT A BOARD GAME BUT WHAT KIND OF MOVES WOULD BE GOOD, IT IS NOT ENOUGH TO PLAY REALLY WELL FOR YOU ALSO NEED TO BE ABLE TO TRY OUT DIFFERENT MOVES IN YOUR MIND AND THEN PLAN AND FIGURE OUT WHICH ONE, WHICH PATH IS THE BEST PATH.
WE REALLY NEED TO BUILD THAT PLANNING CAPABILITY, THE ABILITY TO BREAK DOWN A TASK INTO A SUB TASK AND THEN INSTALLS EACH ONE IN THE RIGHT ORDER TO ACHIEVE SOME BIGGER GOAL.
IT IS STILL MISSING THAT CAPABILITY.
>> Reporter: TELL ME WHY THE USE OF GAMES IS ALSO IMPORTANT TO THE DEVELOPERS OF ARTIFICIAL INTELLIGENCE.
>> YES.
GAMES IS WHAT GOT ME INTO ARTIFICIAL INTELLIGENCE IN THE FIRST PLACE.
PLAYING A LOT OF CHEST FOR THE ENGLAND JUNIOR TEAMS AND THEN TRYING TO IMPROVE MY PROCESSES LEAD ME TO THINK ABOUT ORGANIZING INTELLIGENCE AND ARTIFICIAL INTELLIGENCE.
SO WE USED GAMES WHEN WE STARTED DEEP MIND IN 2010 AS A TESTING GROUND, IMPROVING GROUND FOR OUR ARRHYTHMIC IDEAS.
ONE REASON THAT IS SO GOOD IS BECAUSE GAMES HAVE CLEAR OBJECTIVES.
TO WIN THE GAME OR MAXIMIZE THE POINTS THAT YOU CAN SCORE IN A GAME.
SO IT IS VERY EASY TO MAP OUT AND TRACK IF YOU ARE MAKING PROGRESS WITH YOUR ARTIFICIAL INTELLIGENCE SYSTEMS.
IT IS A VERY CONVENIENT WAY TO DEVELOP THE ALGORITHMIC IDEAS THAT NOW UNDERPIN MODERN AI SYSTEMS.
>> MOST OF US HAVE NOW USED THE CHAT BOX, LIKE GEMINI OR CHATGPT.
BUT YOU HAVE TALKED NOT ONLY ABOUT MOVING US INTO ARTIFICIAL GENERAL INTELLIGENCE, IN OTHER WORDS, THE TYPE OF INTELLIGENCE THAT CAN DO ANYTHING A HUMAN CAN DO.
BUT I ALSO GUESS I WOULD CALL IT REAL-WORLD INTELLIGENCE.
ROBOTS OR SELF DRIVING CARS, THINGS THAT COULD TAKE IN VISUAL INFORMATION AND DO THINGS IN THE PHYSICAL REALM.
HOW IMPORTANT IS THAT AND HOW DO YOU GET THERE?
>> IT IS INCREDIBLY IMPORTANT, THIS IDEA OF EMBODIED INTELLIGENCE AS IT IS SOMETIMES CALLED.
SELF DRIVING CARS ARE AN EXAMPLE OF THAT, ROBOTICS IS ANOTHER EXAMPLE.
THESE SYSTEMS CAN THEN ACTUALLY INTERACT WITH THE REAL WORLD, AS YOU SAY, THE WORLD OF ATOMS, SO TO SPEAK, AND NOT JUST BE STUCK IN THE WORLD OF BITS.
THAT WILL BE A HUGE ADVANCE WE WILL SEE IN THAT SPACE IN THE NEXT FEW YEARS.
AND, YOU KNOW, THAT ALSO WILL INVOLVE THIS PLANNING CAPABILITY AND THE ABILITY TO DO ACTIONS AND CARRY OUT PLANS IN ORDER TO ACHIEVE CERTAIN GOALS.
THAT IS NOT THE ONLY AREA OF REAL-WORLD APPLICATION.
ONE OTHER AREA THAT I AM SUPER PASSIONATE ABOUT AND THE REASON I SPENT MY WHOLE CAREER BUILDING AI IS TO APPLY AI TO SCIENCE, SCIENTIFIC PROBLEMS, SCIENTIFIC DISCOVERY.
INCLUDING OUR PROGRAM TO CRACK THE CHALLENGE OF TEAM FOLDING.
>> Reporter: TELL US MORE ABOUT ALPHA FOLD.
RNA, DNA, ALL OF THESE THINGS THAT WE CAN DETERMINE WHAT APPROACHING LOOKS LIKE, BUT IT ACTUALLY IS A FOLDING OF THE PROTEIN.
HOW IMPORTANT IT HARD WAS THAT?
AND WHAT WILL IT DO FOR US?
>> THE PROTEIN FOLDING PROBLEM IS A 50 YEAR GRAND CHALLENGE IN BIOLOGY, ONE OF THE BIGGEST CHALLENGES IN BIOLOGY, IT WAS PROPOSED IN THE 1970s BY A NOBEL PRIZE WINNER --NOBEL PRIZE WINNER.
CAN YOU DETERMINE A 3-D STRUCTURE, A PROTEIN.
EVERYTHING IN LIFE DEPENDS ON PROTEINS.
ALL YOUR MUSCLES IN YOUR BODY, ALL OF THE FUNCTIONS IN YOUR BODY ARE GOVERNED AND SUPPORTED BY PROTEINS.
AND WHAT A PROTEIN DOES DEPEND ON ITS 3-D SHAPE, HOW IT STORES IN THE BODY.
AND THE CONDUCTOR WAS, COULD YOU PREDICT THE 3-D SHAPE OF A PROTEIN BASED JUST ON ITS TWO- DIMENSIONAL OR ONE-DIMENSIONAL GENETIC SEQUENCE, RIGHT?
THE STRING OF NUMBERS.
SOMETIMES IT IS CALLED THE AMINO ACID SEQUENCE.
AND CAN YOU PREDICT THE 3-D STRUCTURE, THE PROTEIN, JUST FROM ITS AMINO ACID SEQUENCE?
IF YOU COULD DO THAT, IT WOULD BE REALLY IMPORTANT FOR UNDERSTANDING BIOLOGY, THE PROCESSES IN THE BODY, BUT ALSO DESIGNING THINGS LIKE DRUGS AND CURES FOR DISEASES AND UNDERSTANDING WHEN SOMETHING GOES WRONG AND HOW TO DESIGN A DRUG TO BIND TO A CERTAIN PART OF THE PROTEIN.
IT IS A REALLY FOUNDATIONAL, FUNDAMENTAL PROBLEM IN BIOLOGY.
WE MANAGED TO CRACK THAT PROBLEM WITH ALPHA FOLD.
>> THERE ARE SO MANY LARGE LANGUAGE MODELS, ALMOST LIKE A RACETRACK IN WHICH GOOGLE GEMINI, YOURS IS UP THERE AGAINST OPENAI AND AGAINST GRONK FROM X AI.
META HAS ITS OWN, AND ANTHROPIC.
ONE OF THE THINGS THAT SEEMS TO DISTINGUISH THE LATEST MODEL OF GOOGLE GEMINI IS THAT IT IS MULTIMODAL.
MEANING THAT IT CAN LOOK AT IMAGES, YOU CAN HEAR WORDS, NOT JUST TO DEAL WITH TEXT.
EXPLAIN THAT INTO A ME --TO ME IN A WAY THAT MAKES SENSE.
>> AS YOU SAID, MULTIMODAL, IT MEANS THAT IT DOES NOT JUST DEAL WITH LANGUAGE AND TEXT, BUT ALSO IMAGES, VIDEO PROCESSING, CODE, AUDIO.
SO ALL OF THE DIFFERENT MODALITIES WE AS HUMAN BEINGS USE AND EXIST IN.
AND WE ALWAYS THOUGHT THAT WAS CRITICAL FOR THE AI SYSTEMS AND MODELS, TO BE ABLE TO UNDERSTAND.
IF WE WANTED TO UNDERSTAND THE WORLD AROUND US AND BUILD MODELS OF THE WORLD AND HOW THE WORLD WORKS AND BE USEFUL TO US AS DIGITAL ASSISTANT OR SOMETHING LIKE THAT, THEY NEED TO REALLY HAVE A GOOD GROUNDING AND UNDERSTANDING OF HOW THE WORLD WORKS.
IN ORDER TO DO THAT, THEY HAVE TO BE MULTIMODAL AND THEY HAVE TO PROCESS ALL THESE DIFFERENT TYPES OF INFORMATION.
NOT JUST TEXT AND LANGUAGE.
WE BUILD GEMINI FROM THE BEGINNING TO BE NATIVELY MULTIMODAL.
THIS WOULD HAVE HAD THAT ABILITY FROM THE START.
WE WERE ENVISIONING THINGS LIKE A DIGITAL ASSISTANT, A UNIVERSAL ASSISTANT THAT CAN UNDERSTAND THE WORLD AROUND YOU AND THEREFORE BE MUCH MORE HELPFUL, BUT ALSO IF YOU THINK ABOUT THINGS LIKE ROBOTICS OR ANYTHING IN THE OPERATING IN THE REAL WORLD, IT NEEDS TO INTERACT WITH AND DEAL WITH REAL WORLD PROBLEMS, THINGS LIKE SPATIAL RELATIONS AND CONTEXT THAT YOU ARE IN.
SO WE THINK IT IS FUNDAMENTAL FOR GENERAL INTELLIGENCE.
>> Reporter: THE BIG NEWS IN THE PAST WEEK OR TWO WAS META, FACEBOOK, COMING OUT WITH LAMA, IT'S FORM OF A COMPETITOR, IN SOME WAYS, TO GOOGLE JENNA MY AND OPENAI'S SYSTEM.
IN MARK ZUCKERBERG, WHEN HE INTRODUCED IT, MADE A BIG DEAL ABOUT IT BEING OPEN SOURCE.
TELL ME WHY THE GOOGLE GEMINI IS NOT OPEN SOURCE AND WHETHER MARK ZUCKERBERG IS RIGHT TO SAY THAT THIS IS IMPORTANT.
>> IT IS DEFINITELY VERY IMPORTANT.
WE ARE HUGE, GOOGLE "DEEPMIND" AND GOOGLE IN GENERAL ARE HUGE SUPPORTERS OF OPEN SOURCE.
WE WERE DISCUSSING ALPHA FOLD EARLIER, THAT IS OPEN SOURCE.
SCIENTISTS MAKE USE OF IT, PRETTY MUCH IN EVERY COUNTRY IN THE WORLD, WHERE THEY DO THEIR IMPORTANT RESEARCH WORK.
WE HAVE PUBLISHED THOUSANDS OF PAPERS NOW ON ALL OF THE ARCHITECTURES REQUIRED FOR BUILDING MODERN AI SYSTEMS, INCLUDING MOST FAMOUSLY THE TRANSFORMERS PAPERS.
THE ARCHITECTURE THAT UNDERLINES PRETTY MUCH ALL OF THE MODERN LANGUAGE MODELS AND FOUNDATIONAL MODELS.
WE VERY MUCH BELIEVE THAT IT IS THE FASTEST WAY TO MAKE SCIENTIFIC PROGRESS, TO SHARE INFORMATION.
THAT HAS ALWAYS BEEN THE CASE, THAT IS WHY SCIENCE WORKS.
IN THIS PARTICULAR CASE, WITH AGI SYSTEMS, WE NEED TO THINK ABOUT AS I GET MORE POWERFUL.
NOT TODAY'S MODELS, THAT IS FINE.
BUT AS WE GET CLOSER TO ARTIFICIAL GENERAL INTELLIGENCE, YOU KNOW, WHAT ABOUT THE ISSUES AROUND BAD ACTORS?
WHETHER THAT IS INDIVIDUAL OR UP TO NATIONSTATES, USING THESE THINGS, REPURPOSE AND THESE SAME MODELS, THEY CAN BE USED FOR GOOD, THAT IS WHY I HAVE WORKED ON AI MY WHOLE CAREER, TO CURE DISEASES, MAYBE HELP THINGS LIKE CLIMATE CHANGE, SO ON AND SO FORTH, SCIENCE AND MEDICINE.
HARM IF INCORRECTLY USED BY BAD ACTORS.
THAT IS THE QUESTION THAT I THINK WE ARE GOING TO HAVE TO RESOLVE AS A COMMUNITY, AND AS A RESEARCH COMMUNITY, HOW DO WE ENABLE ALL OF THE AMAZING, GOOD USE CASES OF AI AND SHARE INFORMATION AMONGST WELL- MEANING ACTORS, RESEARCHERS, AND SO ON, TO ADVANCE THE FIELD AND COME UP WITH AMAZING NEW APPLICATIONS THAT ARE BENEFITING HUMANITY, BUT AT THE SAME TIME, RESTRICT ACCESS TO WOULD BE BAD ACTORS TO DO HARMFUL THINGS WITH THOSE SAME SYSTEMS BY REPURPOSING THEM IN A DIFFERENT WAY.
THAT IS THE CONUNDRUM THAT WE HAVE TO SOLVE SOMEHOW WITH THIS DEBATE ABOUT OPEN SYSTEMS VERSUS CLOSED SYSTEMS.
I DO NOT THINK THERE IS A CLEAR ANSWER YET ABOUT HOW TO DO THAT, AS THESE SYSTEMS IMPROVE.
BUT I CONGRATULATE MARK ZUCKERBERG AND META ON THEIR GREAT NEW MODEL.
I THINK THIS IS USEFUL TO STIMULATE THE DEBATE ON THIS TOPIC.
>> Reporter: ONE OF THE THINGS THAT CAN MAKE AN AI SYSTEM REALLY GREAT IS THE TRAINING DATA THAT IT CAN USE.
ON GOOGLE, AND YOUTUBE, THIS SHOW WILL BE ON YOUTUBE, OUR SEGMENT, PRETTY SOON.
GOOGLE GEMINI TRAINS ON YOUTUBE UNLESS SOMEBODY STOPS IT.
IT ALSO CAN TRAIN ON MY BOOKS.
IT'S CAN READ ANY BOOK I WROTE.
HOW DO WE REGULATE THAT GOOGLE GEMINI JUST CANNOT TAKE ALL OF THIS DATA AND INTELLECTUAL PROPERTY WITHOUT SOME DEALS?
>> WE ARE VERY CAREFUL AT GOOGLE TO RESPECT ALL OF THOSE COPYRIGHT ISSUES AND TO ONLY TRAIN ON THE OPEN WEB, WHETHER THAT IS YOUTUBE OR THE WEB IN GENERAL.
AND THEN YOU WILL SEE WE HAVE CONSTANT DEALS AS WELL.
SO THIS IS GOING TO BE AN INTERESTING QUESTION AS WELL FOR THE WHOLE INDUSTRY, THE WHOLE RESEARCH INDUSTRY, HOW TO TACKLE THIS GOING FORWARDS.
WE ALSO HAVE GOOGLE OPT OUT SO WEBSITES CAN OPT OUT OF TRAINING IF THEY WANT TO DO THAT, PEOPLE CAN TAKE ADVANTAGE OF THAT.
AND THEN IN THE FAIRNESS OF TIME, WE NEED TO DEVELOP SOME NEW TECHNOLOGIES WHERE WE CAN DO ATTRIBUTION OR SOME FORM OF, YOU KNOW, THIS INPUTS, TRAINING INPUTS, IT HELPED IN SOME FACTIONAL WAY, SOME OUTPUTS, AND THEN DERIVED SOME COMMERCIAL VALUE FROM THAT, THAT CAN FLOW BACK TO THE CONTENT CREATORS.
I THINK THAT TECHNOLOGY IS NOT THERE YET, BUT I THINK WE NEED TO DEVELOP THAT.
IT WOULD HAVE A CONTENT I.D.
LIKE YOUTUBE.
YOUTUBE HAS HAD CONTENT I.D.
FOR MANY YEARS, FOR THE CREATOR COMMUNITY TO BENEFIT MASSIVELY FROM THE DISTRIBUTION THAT YOUTUBE GIVES.
I THINK THAT IS A GOOD EXAMPLE THAT WE ARE TRYING TO FOLLOW.
YOU KNOW, WITHIN THE AI SPACE.
YOU KNOW, AS AN EXAMPLE, THE WAY YOUTUBE IS, THE WAY THE USED TUBE SYSTEM IS DEVELOPED.
>> Reporter: IN THE FASCINATING BIOGRAPHY OF YOUR LIFE, SOMETHING ALMOST AS IMPORTANT AS BEING A GAME PLAYER AND GAME DESIGNER, THAT IS THAT YOU HAVE A PHD IN COGNITIVE NEUROSCIENCE.
YOU LOVE THE HUMAN BRAIN.
HOW IMPORTANT IS IT TO UNDERSTAND HOW THE HUMAN BRAIN WORKS IN ORDER TO DO AI?
AND IS IT SOMETHING THAT WILL ALWAYS BE FUNDAMENTALLY DIFFERENT BETWEEN THE SILICON BAY SYSTEM AND THE WORLD OF THE HUMAN BRAIN?
>> YOU ARE RIGHT.
I DID MY PHD NEARLY 20 YEARS AGO NOW, IN THE MID-2000'S.
BACK IN THOSE TIMES, THE EARLY TIMES OF "DEEPMIND", IT WAS IMPORTANT TO HAVE INSPIRATION BOTH FROM MACHINE LEARNING AND METH ADDICTS AS WELL AS NEUROSCIENCE AND INSPIRATION FROM THE HUMAN BRAIN, AS TO HOW INTELLIGENCE MIGHT WORK.
IT IS NOT THAT YOU WANT TO SLAVISHLY COPY HOW THE BRAIN WORKS.
AS YOU POINTED OUT, BRAINS ARE CARBON-BASED AND OUR COMPUTERS ARE SILICONE-BASED.
THERE IS NO REASON WHY THE MECHANICS SHOULD WORK IN THE SAME WAY.
IN FACT, THEY WORK RIGHT DIFFERENTLY.
BUT A LOT OF ALGORITHMIC PRINCIPLES, SYSTEMS, AND ARCHITECTURES BEHIND INTELLIGENCE ARE IN COMMON.
INCLUDING IN THE EARLY DAYS OF NEURAL NETWORKS, YOU KNOW, THINGS THAT UNDERPIN ALL MODERN AI WERE ORIGINALLY INSPIRED BY NEUROSCIENCE, AND SYNAPSES IN THE BRAIN.
SO THE LIMITATION DETAILS ARE DIFFERENT, BUT THE ALGORITHMIC IDEAS ARE EXTREMELY VALUABLE IN TERMS OF KICKSTARTING WHAT WE SEE AS THE MODERN AI REVOLUTION TODAY.
INCLUDING THIS IDEA OF LEARNING SYSTEMS, REINFORCEMENT LEARNING, AND SYSTEMS THAT LEARN FOR THEMSELVES, VERY MUCH LIKE BIOLOGICAL SYSTEMS, AS OUR OWN BRAINS DO.
MAYBE WHEN WE BUILD AGI, WE CAN USE THAT TO ANALYZE OUR OWN MIND SO THAT WE CAN UNDERSTAND THE NEUROSCIENCE BETTER AND FINALLY UNDERSTAND THE WORKINGS OF OUR OWN BRAIN.
I LOVE THE WHOLE CIRCLE HERE, THIS CIRCLE OF INFLUENCING EACH OTHER.
>> HERE IS SOMETHING YOU HAVE SAID.
MITIGATING THE RISK OF EXTINCTION FROM AI SHOULD BE A GLOBAL PRIORITY.
WHAT ARE >> LOOK, I THINK THAT WAS AN OPEN LETTER THAT I AND MANY OTHERS SIGNED.
IT WAS IMPORTANT TO PUT THAT IN THE OVERTON WINDOW OF THINGS THAT NEEDED TO BE DISCUSSED.
I THINK NOBODY KNOWS THE TIMESCALES OF THAT YET OR THE WORRIES OF THAT.
I THINK THE CURRENT SYSTEMS ARE STILL QUITE FAR FROM ARTIFICIAL NEURAL INTELLIGENCE.
ALSO, WE DO NOT KNOW WHAT THE RISK LEVELS ARE OF THAT.
MAYBE THEY WILL TIME OUT TO BE VERY SIMPLE, TO NAVIGATE CONTROLLED ABILITY OF THESE SYSTEMS.
HOW DO WE OUTPUT THEM AND MAKE SURE THAT WHEN WE SET THEM GOALS, YOU KNOW, THESE MORE AGENT-BASED SYSTEMS, THAT THEY DO NOT DO SOMETHING ELSE ON THE SIDE THAT WE DID NOT INTEND REX UNINTENDED CONSEQUENCES.
THERE ARE MANY SCIENCE FICTION BOOKS WRITTEN ABOUT THAT.
SO HE WANTS TO AVOID ALL OF THOSE THINGS, SO WE USE OUR SYSTEMS FOR GOOD AND FOR AMAZING THINGS.
SOLVING DISEASES, HELPING WITH CLIMATE, INVENTING NEW MATERIALS.
ALL OF THESE AMAZING THINGS THAT WILL COME ABOUT IN THE NEXT DECADE OR SO.
BUT WE NEED TO UNDERSTAND THESE SYSTEMS BETTER.
OVER THAT TIME, WE ALSO WILL UNDERSTAND THE RISKS INVOLVED ABOUT RUNAWAY SYSTEMS DOING UNINTENDED CONSEQUENCES, OR BAD ACTORS USING THESE SYSTEMS IN NEFARIOUS WAYS.
YOU KNOW, THAT MIGHT END UP BEING A LOW PROBABILITY LIKELIHOOD.
BUT, RIGHT NOW THERE IS A LOT OF UNCERTAINTY OVER IT.
SO AS A SCIENTIST, YOU KNOW, THE WAY I DEAL WITH THAT, THE ONLY RESPONSIBLE APPROACH TO THAT IS TO APPROACH IT WITH CAUTIOUS OPTIMISM.
I AM VERY OPTIMISTIC THAT HUMAN INGENUITY COLLECTIVELY WILL WORK THIS OUT, I AM VERY CONFIDENT OF THAT.
OTHERWISE I WOULD NOT HAVE STARTED THIS WHOLE JOURNEY 30 YEARS AGO FOR MYSELF.
BUT, YOU KNOW, IT IS NOT A GIVEN.
NEED TO DO RESEARCH ON AND FOCUS ON TO UNDERSTAND.
THINGS LIKE ANALYSIS OF THESE SYSTEMS SO THAT THEY ARE NOT JUST A BLACK BOX SYSTEMS, THAT WE UNDERSTAND, WE CAN CONTROL.
WE LOOK AT HOW KNOWLEDGE IS REPRESENTED IN THESE SYSTEMS, AND THEN WE WILL BE ABLE TO UNDERSTAND THE RISKS AND PROBABILITY OF THOSE RISKS AND THEN MITIGATE AGAINST THOSE.
IT REALLY IS A CALL TO ACTION TO PAY MORE ATTENTION TO THAT, AS WELL AS ALL THE EXCITING COMMERCIAL POTENTIAL THAT EVERYONE ELSE IS WRAPPED UP IN WHAT WE SHOULD THINK AT THE SAME TIME OF THE RISKS BUT STILL HE OPTIMISTIC ABOUT THAT.
BUT APPROACH IT WITH THE RESPECT THAT IT DESERVES FOR SUCH AN INFORMATIVE TECHNOLOGY OF THAT AI IS.
>> Reporter: DEMIS HASSABIS, THANKS FOR BEING WITH US.
>> THINK YOU FOR HAVING ME.
About This Episode EXPAND
Former White House National Climate Adviser Gina McCarthy discusses a summer filled with extreme weather and silence on the subject from U.S. presidential candidates. Journalists Caitlin Dickerson and Lynsey Addario talk about their reporting on migrants as they follow them through the lethal Darién Gap route. CEO of Google DeepMind Demis Hassabis on the promise and peril of AI discoveries.
LEARN MORE