Skip to main content

tv   DARPA Director Talks to Washington Post  CSPAN  December 7, 2018 10:02am-10:30am EST

10:02 am
, president clinton, barack obama, and donald trump. donald trump, in his heart of hearts, believes that he always wins. here is a guy who has been in new york real estate, gambling, boxing, beauty contests, television, construction, and has never been the target of a criminal investigation. that is astonishing in new york city. author,versation with peter osnos. of the defense advanced research projects agency, stephen walker, spoke with washington post columnist,
10:03 am
david ignatius. darpa's director for the last year and before that he was the acting director and was deputy director since 2012. he has been in darpa operations for a long time. our amazing government ideas laboratory. we credit them for inventing the internet. sorry, al gore. amazing lot of technology achievements. by asking you to talk about this amazing istitution and whether it adapting enough to the way the
10:04 am
world has changed. sense, darpa and the internet created a world in which darpa's old mission doesn't fit right, the private sector is so dominant, so quick. cutting edge science and technology is to be in the government, now it is outside. tell us about the new world. stephen: darpa is 60 years old this last year. we celebrated that in september. darpa was all about creating technological surprise. it was created in the sputnik era. we still look at that as our main charter. one of the ways that we
10:05 am
understand we can do that best is by creating technological .urprise for the united states is tossion of darpa create breakthrough technologies and capabilities for national security. we are focused on national security. andre a department agency the department of defense. andng the last decade, iraq
10:06 am
afghanistan, we may have got too focused on the near term. our men and women were in harm's way, so i think that was appropriate. afghanistan, we started focusing on what are the tech races we need to win again. better than any other organization in the government, bridges the gap between the commercial sector, the private sector, and the defense sector. we bring in people from industry, from universities, from other agencies and they work at darpa for a term appointment. we bring in new people all the time with new ideas. that is one of the reasons why we are at the leading edge of technology.
10:07 am
we are able to adapt to the new problems and where technology is leading us. i love your comment that the job is to avoid the being surprise in the best way to do that is by creating surprises yourself. usant to ask you to surprise . i want you to tell us a couple areas were some big things are about. stephen: i can't tell you everything, but i will focus on a few areas. what are the tech races we need to win in the 21st century? you spoke recently about ai. reason -- onehose
10:08 am
of those areas that we need to win. abouthas a long history, 50 years, of investing in artificial intelligence. some of the first language translation work was darpa-sponsored. over the last decade, we focused on self driving cars. did it out in the desert. some of those are coming to fruition now. we have a long history investing in it. today, weink about ai think about three waves of ai. the first is rules-based. turbotax. if this happens, then you do this. second wave is what people are talking about today, machine learning. games like gong
10:09 am
and doing better than humans at recognizing images. this is what people refer to as machine learning. today by thepplied commercial sector. -- where darpa is is -- working today, where machines are, giving them contextual learning in their environment. , thatchine can tell you is a cat, that is a suitcase. a machine would never understand, you can put the cat inside the suitcase. it wouldn't understand that you don't want to do that. humans understand that instinctively. how do you give that machine
10:10 am
that common sense? that is the next place darpa is headed. it is a basic research activity. it will be critical if we want machines to be partners to the humans and not just tools which is what they are today. david: if you can develop the machine version of common sense, i hope the machines will share it with us. david: i asked -- stephen: i asked the program manager, can you do this for humans? david: there are some very smart people, stephen hawking was one they look at when this prospect of generalized ai, they say that there is enormous danger to humanity in this prospect. i would ask you to respond to that. you obviously don't believe that. you're not afraid of it. you want to push us into it. what about the fear factor?
10:11 am
stephen: in the defense we don't seeday, machines doing anything by themselves. we are focused on the human-machine partnership. it,ink general dunford said how can you give the human more time to make that decision? time and speed in warfare is critical. what we know about where i was and the fragility of ai today even in the second wave, it is a very fragile capability. learning,ed machine but it is really machine trained . they are trained on a lot larger
10:12 am
data sets. if you get away from the data set, they fail pretty badly. it is not one of those things that keeps me up at night as much as biology. david: we will, to biology in a to biology in a minute. asking you about general dunford, the chinese threat of dominance in this area. i have heard it said that china is the opec of data. the chinese government captures everything that human beings do, every time they move, by anything, say anything, communicate anything, that data is being captured. vast reserve of structured data for machines to learn on. our ai companies, as brilliant
10:13 am
r scientists and you at darpa, as brilliant as you are, don't have that same structure of data available to your machines to learn on. people argue that we are in a race that we are going to lose because the other guy has the raw material that we will never hope to match. how would you answer that? inphen: we have programs place to not require so much data. we have a program called .earning classes labels the military doesn't have as much data as the commercial sector about what is going on in the military environment. we train a machine with less labels come up with less data?
10:14 am
another one is because machine learning is so fragile and requires so much data, we want to make sure that when we get an answer from the machine, we want to understand why the machine came up with that answer. now, you might get a probability that it is 85% sure it is the right answer. if we want to turn humans and machines into a partnership, we need the machine to explain to the human how it came up with that answer. we have a program that is focused on that. we're trying to work with the data issue but you are right, china collects a lot of data on its citizens, more than we do. they will have an advantage in the short term. david: briefly, i have heard
10:15 am
proposals that if the united states is going to be competitive, we need to find a way for our ai companies, begle, amazon, microsoft, to able to share, in same way, appropriately anonymized data so they are in the same competitive ballpark as the chinese. the relaxation of antitrust rules and all sorts of things. is that good or a waste of time? stephen: i think it is a good idea as long as you can anonymize the data sets are you are not focused on particular people. if we had access to data that could not go back to the individual, it would help our capabilities in this space. david: another technology that
10:16 am
-- thisterest to me audience probably doesn't know that dr. walker headed up air force research on, is hypersonics. i would like for you to explain what hypersonic technology is all about. it is said that this is an area where we really are behind. the russians and the chinese are building these hypersonic systems and prototypes. are we behind? is this a world changing military technology? give us the hypersonics 101. stephen: hypersonics is flying the -- flying five times the speed of sound or more. it is technology that enables not only speed but ranger.
10:17 am
if you think about -- but range. if you think about the pacific , standoff is important. hypersonics gives you the capability. we have been the leaders in hypersonics technology. in some areas, we still are. publicized indely the press that our competitors, china and russia, are pursuing the technology with great haste and in some cases, they are ah ead. they are motivated moreso than we have been because they want to have a capability where they can beat our defenses. inhave not been focused
10:18 am
countering hypersonics. is the defense against an object moving that fast? stephen: it is hard. that is one of the reasons why .hey are interested in it there are not a lot of good toions, but if you are going defend against something like that, you have to defeat it. and you require some improvements in our sensing capability which we are focused on now with some of the work in the infrastructure, potentially. seeing it and being able to hit kinetically,y -- is difficult. ahead, from abeen technology standpoint, turning it into the capability that we
10:19 am
want to build has not been a priority. it has become so under this last administration. they realized the threat and money is being put toward it. david: since this is one of these cool, out of the movies technologies, what about lasers and beam weapons? it was always thought that they were too heavy or too difficult to deploy. are those problems being solved? is that around the next corner? stephen: i have never met a four-star general that did not want a laser on his airplane. it would be really neat. as you mentioned, weight is an issue. a lot of the laser development focuses on the front end, and the power generation and cooling
10:20 am
in terms ofdds up weight. i think airplanes will be the last deployment of it, but we are close to having a ship-based technology. i think ground capability, from trucks, are being worked heavily. those will be closer than a las er on an airplane. has been investing in something called solid-state fiber technology. one to twove 1 to kilowatt fibers and bind them together. we are making good progress on solid-state fiber technology, being able to look at 10
10:21 am
kilowatts. david: if you could make this work, you really would have a beam, like in the movies, that can zap incoming planes or other objects, and would have potential powers against anything in space, presumably. stephen: yes. it is always easier in the movies, but certainly, you can and vision capabilities like capabilitiesion like that. i think we will see that over the next decade. david: you mentioned the question of bio research. that is on everybody's mind after the startling news from china -- it is not clear to me as. rogue he w creating new life in a test
10:22 am
tube. crossing a frontier that is so important and scary. which look at the way in biological sciences are combining with technologies, telly about -- tell me about what concerns darpa the most? where are you focusing? stephen: one of my focuses is the homeland agency and protecting against existential threats. that is where i stick the biology piece. i see it as a real threat, natural pandemics or man-made. we are focused on biomedical asian fashion bi -- we are
10:23 am
focused on bio-mitigation. this technology the darpa is focused on has so much potential for good and for curing disease, if used properly. part of it is figuring out how the technology works. we created a program called safe-change. the whole point is understanding how these technologies and gene editing systems work. and also, how would you turn it off? how could you reverse it if it were used for nefarious purposes. this is a basic program. it is all out in the open. we are working with universities on this. how tow to have success,
10:24 am
use these techniques without , affectingng effects the rest of the genome when trying to do one edit. one example of what we are doing in the bio-space is really important. especially, technologies that can be used for good and evil. not everybody shares the ethical values that we have in this country. we need to be prepared and that is what we are trying to do. david: and something about predicting pandemics earlier to deal with them more appropriately. -- so you can put
10:25 am
prevention measures in place so that disease is not spread. p3, to one is called develop a vaccine for an unknown virus that is detected, trying to develop a vaccine that could help 20,000 people or more in 60 days or less. right now, it's impossible. it usually takes about 18 months in this country for a deck seen to be developed and proven, at the earliest. it's very darpa, trying to do the impossible her,e but the program -- impossible here, but manager is is trying it. david: i have a daughter who works in infectious disease.
10:26 am
after this i'm going to call her. stephen: does she want to be a program manager? [laughter] david: i worry about a world in which people are able to, not simply to create the fake news, as we saw the russians do aggressively in the last elections, but create fake events, create digital representations of audio and video, that appear to be real, but aren't. we can see, go online, and you'll see examples of this so-called deep fake technology. i wonder if darpa has any ideas in focusing on what i call the ct, knowingof fa
10:27 am
where they come from, knowing whether they are real or fake. is that something you are working on? stephen: it is. this is something we started in .014 idea of warfare is something that is really .hallenging us the first thing challenging us is what is the truth, what are the facts? purpose of this program was to look at images on the internet and video. , butot sure about audio
10:28 am
definitely pictures and video. that we could look at these pictures and video and determine whether they have been tampered with or not. in video, when you compress and uncompress a video, if you've developedes, we have tools that can detect those changes to the video, and help a human user pinpoint where the changes have been made. a placerying to get to where we can determine what the a video on thein internet. the companies in silicon valley are pretty interested. david: that is exciting and
10:29 am
important for my business. it's time for us to wrap up, and dr. walker needs to get to another event. this reinforces my impression that darpa is the coolest job in the government. when i grow up, i want to work at darpa. if you would like to highlight , youiscussions we have had can watch it live online. thanking dr.e in walker. [applause] [captions copyright national cable satellite corp. 2018] [captioning performed by the national captioning institute, which is responsible for its caption content and accuracy. visit ncicap.org]

46 Views

info Stream Only

Uploaded by TV Archive on