tv National Security Commission on Artificial Intelligence Conference Part 1 CSPAN November 5, 2019 7:10pm-7:50pm EST
i am joined with two close friends of mine. probably the only person i can say this in the entire world to work with and for both of them. want to disclose my conflict of interest. general shanahan went to michigan. rotc into the service of our country in 1884, he has been promoted a gazillion times and in charge of intelligence and operational activities. eventually we needed somebody operationally to implement ai in the entire dod that he was the perfect choice. worked at them in my role as chairman. and the prosecutor who came to silicon valley and worked at ebay and google snagged him
about 15 years ago. and during that time all of global policy. very significant players. what i thought we should do since you all have heard from me plenty is perhaps we should just have you make some comments so i thank you very much. it's a pleasure to be with you the public-private partnership i grew up in this community. i was born and spent the few years of my life on military bases and my father had his career in lockheed i made sure that the private defense sector and universities can
work together. before we jump in i want to take on two issues up front. i've been frustrated to hear concerns around commitment of national security and to set the record straight on those two issues. first on china africa in 2010 with an attack on our infrastructure. we have learned a lot from that experience and those that have operations in china we have chosen our operations very carefully and our focus is on advertising in the open source platform. second with regard to the more general question of national security and engagement this
is where we decided to press the reset button to develop our own ai principles with internal standards and review process. that was on, not a broader statement that the department of defense and national security administration. and throughout the valley of national security. it's important to remember the history of government technologies from radar to the internet to the wartime economist vehicles. just in the last couple of weeks we have an extraordinary accomplishment which moved forward the frontier science and technology.
that is built on research scientist at nasa. and carried out in many ways. so those kinds of exchanges that technological innovation. we feel it's weird contributing to the national community a lot of that work we have lots of people who work at google we make sure they complete their medical service and we try to take steps to make sure that is transitioning so they can make best use of military skills in the private sector and as we do that we are fully engaged in a wide variety working on
national initiatives to healthcare to business automation we are working on how to progress that operation of hardware and use that hardware software interface so taking on those things we are eager to do more to take on the certification with the range of these areas at the same time there is a great partnership to be had that were very well done and continues to lay the
groundwork by the department of defense with the directive of 3009 with that application with the work that dod has done and we have been trying to drive forward with the principles in very common and overlapping areas safety, human judgment accountability fairness they are all critical areas there are different actors and that is critically important it is a shared responsibility to get it right with that global framework and global approach it is extremely important as something we want to support we are working together to
look at the opportunities because we are proud american company committed to the defense of the united states we are eager to continue this work and where we can work together. >> take us what you are up to. >> it is great to be here i am a poor substitute but it is beyond any soundbites but undoubtedly it is the first and last time so hang on for the main event i relish the opportunity to have this conversation about public-private partnership
reflecting back in just about one year in the seat there is one overarching theme that is the importance or the necessity to strengthen bond between academia and industry. you brought it up and others have mentioned the relationship it is a triangle the form as equilateral triangle. i would suggest that is largely the form it did take beginning in the 19 fifties lasting through the early part of the decade walter isaacson writes about this in his book the innovators. that's what it is the go to for silicon valley today.
they are distorted or a little frayed in addition to being different links edward snowden and agility and general mistrust within the government and industry. the task is made much more difficult the industry is moving so much faster when it comes to adoption and integration. and the tech industry he sees no - - sees no reason and even far more than it is portrayed we don't make it easy for them
so i would reinforce the theme of the security commission report with the idea of a shared sense of responsibility and a shared vision of trust and transparency our national security depends on even those for various reasons view dod was suspicion are reluctant to accept we are in a strategic competition with china i would hope you would agree it is a critical component of our nation's prosperity and self-sufficiency. in other words where you stand with ai technology i submit we can never attain the vision outlined with industry and academia with equal
partnership. there's too much at stake to do otherwise. public-private partnerships are the very essence of america's success as a nation across the entire united states government we have to make this triangle. >> i will ask a couple of questions to both of you talk about similar. [laughter] i think it's no sick on --dash secret and is quickly evolving becoming an enterprise company that there are different protocols and different ways of engagement to go along with that that all employees have ended - - have identical views
on issues they don't but that debate and discussion is positive as well as a negative so with that constructive debate is america first on one of the leading thinkers of quantum mechanics. out of that comes incredible strength that we can have a more robust framework to help build socials trust some as he put forward our principles and the governing process in a sense a couple of pages of those principles in the implementation because you could discover a lot of the
hard problems conflict and are challenging we had debates whether or not to publish a paper on lipreading. was a great benefit people are hard of hearing but it could be misused for surveillance and other purposes. was appropriate to publish because that technology is only in the one / one phase it is an example of those discussions of facial recognition or lipreading or other challenging questions we have to come to the reality of the trade-off there is a lot of room for a collaboration of cybersecurity and logistics and transportation and healthcare that where we are
already engaged with the military. >> so tell us more. >> our intent was to go after commercial industry. do not reinvent the wheel and our approach was simple anybody in the market there was a small start up with the cybercompanies in the world wide we have to google project maven cracks we want the best ai talent in the world with extraordinarily difficult problems to go after. to have that collaboration. what was happening internal and how that plays out is a different story that all the way to the end and there are
products we were very pleased with. now those software engineers that were ostracized for the department of defense but we have tremendous support. but what we found is we lost an area very quickly. and for them to be public about what they want to do in the department of defense and we do in very general terms it was intelligent surveillance and they had no weapons and it is my weapons project with those wild assumptions a
project maven was or was not. no pun intended we actually googled it was actually controversial now has been put in permanently was not controversial to me or to the dean or not anybody right now beyond people that don't like what we are doing. this is an interesting point that i'm not sure everybody appreciates or agrees with me. i view it as a canary in the coal mine. as it emerged in the conflict we got some of that out of the way and how much the company and all the companies that we deal with want to work with department of defense. that is an important narrative. it would happen to somebody else at some point but the idea of transparency and what
each side is trying to achieve is the biggest lessons of all. >> it's a tragedy we don't wear hats anymore with my big hat i can tell you with general shanahan with soldiers and airmen that we put them in front of the mindnumbing that they watch screens all day and it's a terrible waste of the human asset that the military produces it is a huge opportunity to get them to work at a higher level. and indeed that creation. so now talking about ethics and that kerfuffle inside of google with the ethics
proposal and that produced a public document which is quite definitive and maybe you could talk about that and then the proposal to the military and you are the customer on military ai ethics for go i assume both of you are in favor as they have copied your approach in favor of this. does it really work. does google turn off or stop doing things? how does it actually work? people claim the military
won't operate under ethics principles and we cite many rules they are required to work under. >> having frameworks in place early on with the review process is a critical part of internal as well as external so to talk about surveillance and recognition tools are deployed in appropriate ways and make sure we know the scope of project development. that's a valuable thing for both sides to make sure expectations are clear across society. so another example is general
purpose facial recognition that we have more policy and safeguards. and when it comes to weapons we went to be very careful of the application of ai. we recognize the limits of our experience in that area. and working through these areas there is a remarkable degree of convergence now internationally we see the european commission and this is an interesting exercise we all have a combination for the next generation technology.
>> this may be the best starting point with this area of convergence with commercial industry and the government with the ai ethics principal so do we agree on all of these or some of these? it's a good starting point. so i can tell you with certainty that china and russia to involve public hearings and discussion they are not doing and i don't expect they will that what they are doing and why they are doing it to embark on a long process with all the different voices of artificial intelligence. so by those who have time and
attention 35 and a half years in uniform i have never spent that much time with the department of defense and eisai commended the. looking at technologies and there are differences to artificial intelligence. it is similar to every other technology in the department and here is substantive differences for go that's a pretty good framework for go you have a way of looking at this with artificial intelligence our history and approach and training of how we bring it in with the prototype so now this report has been presented so what do
you think about the report to have the best possible starting point and what will you do about it cracks this is where it becomes complicated. is of the implementation plan to make these recommendations but through my boss to make those recommendation how we implement this this is not an overnight task. and to have that starting point. 's to make that's wonderful framing i like to push back a little bit. open ai into technology that was sufficiently good they did not release it but that is an
example that is anybody putting pressure? they say no we apply with our good judgment. famously you very early said we will avoid that where will the industry and up in the self-restraint? is a common set of principles? is it common with being careful? how does this play out quick. >> you have already seen efforts on artificial intelligence with the work that is being done. looking at more of these frameworks and appropriate safeguards with a variety of
different areas. and we are on the path to doing it. that communication is background and how you use these tools so it is understandable but also that degree that is used. >> talking inside the pentagon of this new kind of warfare. the take us through in the same sense what is new and powerful about this technology in a military context what is
the language of the position quick. >> i go back to as we were formed with the secretary of defense walks in the room like yesterday. and says now you are the tea that will figure out and to get away from the research piece of it. but now we need a team that is focused on the war fighters. and what he gave us it's not an accident. becomes much easier to say. >> your acronyms will kill me. [laughter] >> so it's the idea of counterterrorism to be shocked
by the speed and the chaos so how do we envision that fight happening. if you described earlier how fast we get inside these decisions quick. >> the author of i you get to the cycle of decision-making. but if we try to do this than the other side has machines and algorithms it is unacceptably high risk of losing the conflict. so what you are getting in the future scenario how will you
be sure but that is a starting point and we have to do a lot more work on the front end but we will really be at a disadvantage if we think we are human against machine. it will be human and machine on one side and on the other but that superiority you may be facing it could be algorithmic. >> to me the key question is what happens it is human decision-making cracks so people check with their superior with a rule of engagement.
how well the military adjust so we have to give people the policies of what they need to do to have a solution to develop the algorithm in the field we can do that. and it's more than what people are comfortable with today. talk about higher risk and higher consequence. it is the idea to decentralize experimentation and innovation that is as described this morning happens at the bottom you have to push from above.
>> in addition cybersecurity and cyberdefense that we say to stabilize to work together to recognize those patterns. >> do you have a model one of the themes of the whole industry but i'm really referring to the government as a whole do you have a model how the government should work quick. >> we only talk about two important elements and first is the notion and the second is a global framework which helps of that process it is
the administrative question how to make it as easy as possible for a new company to enter into these partnerships. it's over small companies. and even to get more involved in that environment. so looking at modernizing procurement and then to make that as quick and as flexible as possible and traditionally that is fertile ground for these collaborative enterprises. looking at human resources exchanges there are authorities out there that authorize private sector to come into the government but
in practice it's harder than you would think. >> because we are making recommendations ending up in legislation one year from now are there specific things that could promote public private partnerships? are those that were closely and that extraordinary contribution so the sum of all of that do you have a model of specific things that are helpful to decrease the friction and increase cohesion for the federal government. >> so much as started to happen the last couple of years castle run and compile the combat we should say
inside the dod has an impact to change the procedures that we have to figure out how to institutionalize with the department of defense and to bring in talent from the outside and then working for start up in the last job but also the chief technical officer comes in and within 24 hours a different view of what we need to do. we need sabbaticals people coming in from academia and secretary of defense that is all beginning to happen at the next level to understand what
we are talking about. it is the are two peer relationships. >> we are priming the pump on the important areas whether models of simulation and recruiting different areas. another important component is the kernel is critical that it is difficult for jeff to have security clearances for all elements of that piece so there is that successful individual experiment to build that familiarity to make it easier to have wider adoption.
>> it is time to finish up i was to put to bed the notion silicon valley would work with the military we have clearly see as we move forward to build this collective between private partnerships can use summarize the key message or the keyword? why are you here and why did you make a special trip just to make this point we are committed to the cause of national defense united states of america with safety and security we approach that task thoughtfully as we approach other technologies to be
thoughtful with understanding as we move forward that the military and us government share we are looking forward to work more closely together in the future. >> so you are the fellow that will make this change happen across $660 billion with an enormous bureaucracy. how will you pull this off quick. >> one person at a time. [laughter] it has to be a culmination of top-down it was set on the previous panel you must have the full support of leadership from the very top it is critical there is no question they already know what that is and how we meet that in the
middle. this is a daunting task it is a multi generational problem with a multigenerational solution we won't wake up tomorrow to figure out right but just keep plowing ahead with the resources i know we will get there. >> i worked for 15 years back i could not be more proud of the impact on society and scale and reach of corporations and in general i don't think he can chosen a better person to lead this you really have moved the resources and gotten the money and the attention and delivered so thank you very much. [applause]