Superintelligence will accelerate all scientific domains, says Michael Kratsios

>> THE WHITE HOUSE IS INVESTING OVER $6 BILLION TO BOOST U.S. SCIENCE, ACCELERATE STEM DEGREES AND DRIVE INNOVATION IN SPACE QUANTUM TECH AND BIOLOGY. JOINING US RIGHT NOW IS WHITE HOUSE OFFICE OF SCIENCE AND TECHNOLOGY POLICY ADVISOR MICHAEL KRATSIOS. HE'S ALSO THE ASSISTANT TO THE PRESIDENT. GOOD MORNING TO YOU. LET'S TALK ABOUT WHAT THIS INITIATIVE IS AND WHERE ALL THE MONEY IS GOING TO GO. >> WELL, GOOD MORNING. THANKS SO MUCH FOR HAVING ME. I THINK FIRST AND FOREMOST, THE THING WE'RE MOST EXCITED ABOUT IS APPLYING SUPERINTELLIGENCE TO SCIENTIFIC DISCOVERY. WE FUNDAMENTALLY BELIEVE THAT ALL SCIENTIFIC DOMAINS, WHETHER YOU'RE WORKING IN MATERIAL SCIENCE AND CHEMISTRY AND MATH AND PHYSICS, ARE ALL GOING TO BE ACCELERATED BY THE USE OF C. AND YESTERDAY WE ANNOUNCED A $2.4 BILLION CONTRIBUTION BY THE PRIVATE SECTOR TO A CONSORTIA, WHICH WILL THEN BE ABLE TO DIVVY OUT THAT MONEY TO AMERICAN SCIENTISTS ALL AROUND THE COUNTRY. AND JUST AS YOU GUYS COVER ON YOUR SHOW ALL THE TIME, PRIVATE SECTOR COMPANIES PAY A LOT OF MONEY TO ACCESS COMPUTE TO DO THEIR WORKLOADS. WHAT'S EVEN HARDER IS FOR ACTUAL AMERICAN SCIENTISTS TO GET ACCESS TO THAT COMPUTE. SO YESTERDAY WAS A HUGE ANNOUNCEMENT IN PROVIDING THIS BIG UNLOCK FOR FOR AMERICAN SCIENTISTS. >> WHO ARE YOU BUYING THE COMPUTE FROM? >> SO THE COMPUTE IS ACTUALLY BEING DONATED. SO THE BIGGEST CONTRIBUTOR IS NVIDIA WITH $1 BILLION. AMD CAME IN WITH ABOUT 500 MILLION AND A NUMBER OF OTHERS. ANTHROPIC DID 150, OPENAI DID 200, GOOGLE DID 150. AND THESE ARE COMPANIES THAT UNDERSTAND THAT THE FUTURE OF AMERICAN LEADERSHIP AND SCIENTIFIC DOMINANCE IS IS ALL ABOUT ACCESS TO COMPUTE. AND WE'RE THRILLED THAT THEY WANT TO BE PART OF THIS INITIATIVE. >> WHEN YOU THINK ABOUT THE BIGGEST OPPORTUNITIES, PEOPLE OFTENTIMES, ESPECIALLY WHEN THEY TALK ABOUT AI OR I SHOULD SAY S I IN THIS IN THIS CONTEXT, THEY TALK ABOUT SORT OF THE ABILITY TO DO EXTRAORDINARY SCIENCE. SOME PEOPLE TALK ABOUT CURING CANCER IN FIVE YEARS, BUT THEN YOU TALK TO OTHER PEOPLE IN THE SCIENTIFIC COMMUNITY AND THEY SAY, YOU KNOW WHAT? THERE'S GOING TO BE CERTAIN CERTAIN THINGS THAT CAN BE IMPROVED. BUT CANCER, FOR EXAMPLE, MAY NOT BE ONE OF THEM. WHERE DO YOU LAND ON THAT? >> I'M MUCH MORE OPTIMISTIC, AND I THINK WE HAD AN ALL DAY EVENT YESTERDAY WITH SCIENTISTS FROM ALL THESE DIFFERENT DISCIPLINES WHO CAME TO TALK ABOUT THE IMPACT THAT THESE EMERGING TECHNOLOGIES ARE HAVING ON THEIR FIELD. AND THE SENSE OF OPTIMISM WAS, WAS, WAS QUITE PALPABLE. I THINK EVERYONE WAS VERY EXCITED ABOUT ABOUT WHAT'S TO COME. AND I THINK WHAT WE'RE SEEING IN THE ECOSYSTEM, WHICH IS VERY UNIQUE FROM YEARS BEFORE, IS THAT WE'RE STARTING TO INTRODUCE NEW INSTITUTIONS OF DISCOVERY. SO PREVIOUSLY, IF YOU THINK ABOUT WHERE THESE BREAKTHROUGHS WERE MADE, A LOT OF THEM WERE MADE INITIALLY IN ACADEMIA, AND THOSE WERE OFTEN THEN PICKED UP BY THE PRIVATE SECTOR AND COMMERCIALIZED. WHAT WE'RE SEEING TODAY IS THAT THIS DISCOVERY IS HAPPENING ALL OVER OUR ECOSYSTEM. IT'S HAPPENING IN THE PRIVATE SECTOR. IT'S HAPPENING IN PHILANTHROPY, IT'S HAPPENING IN ACADEMIA. IT'S HAPPENING IN OUR NATIONAL LABS. AND THAT'S LIKE A REAL REFLECTION OF THE ECOSYSTEM WE HAVE. WE HAVE TODAY. ANOTHER GREAT ANNOUNCEMENT WE HAD YESTERDAY WAS ABOUT $310 MILLION, CONTRIBUTED AGAIN, BY PRIVATE SECTOR PLAYERS TO SUPPORT WHAT ARE KNOWN AS X LABS. AND X LABS ARE AGAIN, NONACADEMIC AND NON-GOVERNMENT LABS THAT ARE BEING SET UP BY PHILANTHROPY AND BY OTHERS TO WORK ON DISCRETE SCIENTIFIC PROBLEMS. SO AGAIN, IT'S THIS REFLECTION THAT'S I IS TRULY UNLOCKING DISCOVERY ACROSS SO MANY PARTS OF OUR OF OUR ECONOMIC ECOSYSTEM. >> HOW DO YOU THINK ABOUT THE FLIP SIDE OF ALL OF THIS? THERE IS A DEEP WORRY THAT SOME OF THESE TECHNOLOGIES COULD BUILD BIOTERROR THINGS WHERE, YOU KNOW, WE'RE ALL WORRIED ABOUT THE PLAGUE IN RUSSIA, FOR EXAMPLE, WHICH WHICH, BY THE WAY, MAY NOT BE RELATED TO AI. >> I THINK THE POWER OF S I IS TO ALSO HELP US ON THE ON THE DEFENSIVE SIDE. AND I THINK WE'VE SEEN THAT, PARTICULARLY IN PLACES LIKE CYBER, WHERE PEOPLE DO SOMETIMES HAVE CONCERNS ABOUT THE CYBER CAPABILITIES OF THESE ADVANCED MODELS. WHAT THESE MODELS ALSO ARE, ARE THE BEST CYBER DEFENDERS IN THE HISTORY OF THE WORLD. SO WE CAN APPLY THOSE S I MODELS TO SORT OF HARDEN OUR OUR CYBER INFRASTRUCTURE. AND I THINK THE SAME GOES FOR, FOR THE WORLD OF, OF BIO, IF YOU CAN ACTUALLY SORT OF CREATE THE RIGHT MODELS TO BE ABLE TO ASSESS WHERE SORT OF POTENTIAL BAD ACTORS WILL BE DOING THINGS, YOU CAN CREATE THE NECESSARY CONTROLS ON THE MATERIAL SIDE. SO THOSE THINGS CAN ULTIMATELY NEVER BE PRINTED. >> YOU KNOW, WE'RE IN THIS BATTLE WITH CHINA. HOW FAR AHEAD DO YOU THINK WE REALLY ARE? >> I'M OPTIMISTIC. I THINK AT LEAST IF YOU TRY TO MEASURE SORT OF THE WHERE WE ARE ON FRONTIER MODELS THEMSELVES, WE'RE USUALLY KIND OF IN THE 3 TO 6 MONTH RANGE AHEAD OF THEM. AS WE'VE TALKED ABOUT OFTEN, MANY CHINESE FIRMS USE A TECHNIQUE CALLED DISTILLATION, WHERE THEY ESSENTIALLY TAKE THE THE TRACES OR THE OUTPUT FROM OUR MODELS AND USE IT TO TRAIN THEIR OWN. I THINK THAT ALLOWS THEM TO BE A SOMEWHAT FAST FOLLOWER AND KIND OF 3 TO 6 MONTH RANGE, BUT IT ALSO IS A SEVERE AND SEVERE CAP ON HOW MUCH PROGRESS THEY CAN MAKE BECAUSE THEY USE DISTILLATION, THEY CAN NEVER BE BETTER THAN OUR BEST MODEL. >> I'VE GOT A DIFFERENT CHINA QUESTION FOR YOU. YOU KNOW, WE'RE ALL IN THIS RACE AND WE'RE ALL WORRIED ABOUT SAFETY. AND THERE'S THIS VIEW THAT, YOU KNOW, CHINA IS JUST GOING TO BARREL AHEAD. I'VE ALWAYS THOUGHT THAT AT SOME POINT, CHINA ACTUALLY COULD BE THE ONE THAT PUTS THE BRAKES ON ALL OF THIS. AND THE REASON I SAY THAT IS HISTORICALLY, THEY'VE OFTENTIMES BEEN VERY PERMISSIVE ABOUT TECHNOLOGY EARLY ON. CRYPTO, YOU KNOW, SOME OF THE BIG TECH COMPANIES, ALIBABA EARLY ON. BUT THEN, YOU KNOW, AT SOME POINT WHEN THEY THOUGHT IT WAS TOO MUCH, YOU KNOW, THEY BROUGHT DOWN THE GATES. WHAT DO YOU THINK OF THAT? >> YEAH, I THINK THAT'S A THAT'S A VERY LIVE DEBATE HERE IN WASHINGTON AMONG PEOPLE WHO TRACK SORT OF CHINA POLICY, WHICH DIRECTION WILL THEY GO? I THINK RIGHT NOW THE FRONTIER CAPABILITIES ARE ONES WHICH POSE SOME RISK TO, TO, TO THE REGIME OR TO, TO BROADLY LEADERS WHO HAVE THIS TECHNOLOGY IN THEIR COUNTRY. BUT IT'S NOT QUITE IT'S NOT QUITE THERE YET. THAT WOULD THAT WOULD BE VERY, VERY BAD. BUT I THINK