tv Charlie Rose PBS July 7, 2016 12:00pm-1:01pm PDT
>> rose: welcome to the program, i'm major garrett of cbs news sitting in for charlie rose who is away this week. we begin tonight with a look at the fallout from the fbi's investigation to hillary clinton's e-mails with ari melber and eric lichtblau. >> we ask other people to respond to them. i actually think the fbi process and director comey deserves better than that, that is how i approach my fact checking. i will tell you something else, within minutes of his presentation, unusual as it was, you had politicians coming out and saying that obviously it should have been the reverse. that is dangerous for a process-- process. speaker paul ryan said this was clearly wrong and erroneous and so this was a problem for the rule of law. quite the opposite, i think, if he hadn't digested all the
information, didn't know what they know. >> we continue with a discussion of domestic politics in the presidential race with bob cost blanca candidate. if anything, he has doubled down on being an unconventional candidate. and he's, from the get-go, profited by the element of surprise. everyone is political friends and adversaries, us, everyone, have to wait and see what he is going to say or do next. >> we conclude with a look at artificial intelligence with lucy suchman, nathan yal popper and zeynep tufeki-- tufeki. >> one of the things i'm-- is the language that is used. now could see that as a way of trying to make things more understandable to us. but i think it actually
mystifies and obscure was is really going on. so the algorithms that zeynep was describing which are very powerful, but we could hear from her description that they are basically reproducing the things that they're operating with, are parameters that have been introduced by us. so and many of those are stereotypes. >> rose: more dns. >> more hillary clinton e-mail investigation, and >> rose: funding for "charlie rose" has been provided by the following: >> and by bloomberg, a provider of multimedia news and information services worldwide. captioning sponsored by rose communications
from our studios in new york city, this is charlie rose. >> good eveningk i'm major garrett filling in for charlie rose who is away. we're going to start this evening with a conversation about the legal aspects of fbi director james comey's decision announced yesterday in washington not to bring criminal charges against presumptive democrat i cannot knee hillary rodham clintonry clinton. joining me eric lichtblau and ari pel ber. let me start with you. equal justice under the law, did what james comey said yesterday meet that standard? >> yes, based on the evidence they provided and what is publicly known, according to a very volume nows review of the e-mails, the fbi which means director comey and also the investigate ares and line agents read every e-mail, interviewed the key players, looked at the security issues including legitimate questions about what it means to use a private server and what exposure that provides
classified information, and then came to a judgement that i thought was quite clear in the way he stated it. that number one, there wasn't the kind of evidence to support the bringing of federal charges. full stop, that is enough. but then number two he clearly wanted to go further and say this wasn't a close call. indeed, no reasonable prosecutor, he said, would go forward on these facts. >> eric, and yet there is this ongoing debate that people who have been in lesser positions within the government have been charged for lesser offenses. is that a valid point of view of criticism of the conclusion comey reached? >> well, i think it's a valid question to raise. you certainly have, as you say, outside lawyers, some of them even former prosecutors who were saying there are lesser offenses that have been charged, involving the handling and mishandling of classified evidence, even if there wasn't intent, which is really the big thing that comey focused in on
in his briefing, was saying that they found no evidence or at least not convincing evidence that she had intentionally violated any statutes. other people, though, who were prosecuted who also were not found to have intentionally violated statutes. >> and rudy geulianee who is a trump supporter who say partisan but never the less was an effective and aggressive prosecutor said one of the things that prosecutors look at, ari, is if you do it over and over again, that suggests intent. >> it could be, if prosecutors look for all sorts of aggressive theories. i think what jim comey was talking about with intent which is the legal concept that is important to note here, is that they in their review didn't find an intent to either distribute or endanger this classified material. that this was a bug bud, not a feature. he also spoke to the lack of judgement or width of in setting up this type of system. but as others have noted, there are many aspects of the government process that have unsecured material.
alberto gonzalez was reviewed because he took materials, put them at home and in a safe in his office. some people might think that sounds some what safe, it wasn't. that is an unauthorized place. it was a federal review of that and ultimatey determined was this case, that what he did was unauthorized but it wasn't a desire to break the law when he put it in a safe. and so i think another big question here is, what do people think happened? do they think that there was something focused on mishandling classified information or something else. because a lot of what we have seen in the other prosecutions that did go to trial, which did include, yes, lower level officials, there was usually what i would call a plus factor. there was some endangerment of the classified material and something else, sometimes that was an expressed intent to give it to a reporter, which as journalists, a lot of us do end up getting this material and we have a broad view of why that could be in the so called public interest. but again, the federal criminal standard that comey is applying, it taking it and giving it away to an unauthorized person or
what general petraeus ultimately plead to is a plus factor. then there is compliance which a lot of normal people don't care about, the law and prosecutors care a lot about that. some of theers other cases have been cited involved also the person who mishandled the classified information misleading investigators, general petraeus admitted to doing that. that is a free standing federal crime there was in allegation of that either. here. i think director comey did something positive by sharing a lot of information but that leaves a lot of people picking part dispar at aspects of it. >> i want to pick up on what ari said, because many people raised the petraeus case as something that fits within this-- is comparable, and therefore evidence that the system did not work as it ought to have worked from a legal perspective. would you care to validate or invalidate that theory? >> sure, i mean i think you are right that the petraeus case was critical in this. the impact really was to box in the fbi because the feeling at the fbi was that yes, while
there were similarities between the two cases, high profile people mishandling classified material, the petraeus case was much more egregious even from the outside. here you had a general, four star general who at the time was the head of the cia who knowingly gave his mistress what he calls his black book journals with all sorts of classified information. in them was tape recorded saying they knew these were classified and lied to the fbi about it. so those facts, i think from the outside, seem more egregious than those in the clinton case, as much as you can criticize her. and he was allowed to plead to a misdemeanor. the case the fbi really went to bat with or the justice department and jim comey was overruled by eric holler in that case. so the feeling was if general petraeus doing all that he did was allowed to plead to a misdemeanor, the lowest offense, where does that leave us with hillary clinton. i think that was a big part of why you saw them not recommend
charges against her. >> rose: there's. >> there's a perception that something went wrong here, that maybe there was an unnecessary or erroneous expectation that the law would come down harder on hillary cln ton than the fbi direct are did. do you think there is anything about this case that raises-- raises legitimate questions about, for lack of a better term, the sank tit of the rule of law? >> i don't think there is anything about this case that is normal. everything, every step of the way has been usual from the profile of the-- one of the subjects of the investigation to the media scrutiny. and yesterday to the remarkable press conference that comey held. so it's really difficult to compare this in any reasonable way to other cases. you can certainly make a case as many are that wasn't her neighbor, prom thens, that got her lenient treatment.
people are making that case. on the other side you have democrats who are saying that she got worse treatment than other people might because you are the fbi director up there as is rarely ever done talking publicly, laying out facts, blisserring facts about her, her behavior and the decision not to-- not to bring charges. >> there is a lot there in the system that assures independence. so it is quite a serious burden for people who have political conspiracies to provide the evidence. why did this bush deputy attorney general want to go to bat so hard for hillary clinton if that is your conspiracy. why do all the fbi line agents who raid all the individual e-mails, why would they be doing this. it doesn't really wash based on what we know, as a lawyer and journalist, i'm always open to the evidence. there may be things we don't know. but we do live in a world where it is easy to throw out major charges and allegations and ask other people to respond to them. i think the fbi process and director comey deserves better than that, that is how a proach
my fact-checking. i will tell you something else, within minutes of his presentation, unusual as it was, you had politicians coming out and saying that obviously it should have been the reverse. that is dangerous for our process. that speaker paul ryan said this was clearly wrong and erroneous and so this was a problem for the rule of law. quite the opposite, i think, if he hadn't digested all the information, didn't know what they know. there are a lot of things that voter may want to punish, including potentially the way hillary clinton acted in her tenure as secretary of state, that are simply not illegal. they may be unwise, they may be unsafe, and they may be judgements. i think it's a good thing think being the connective tissue between the law and politics, that the investigation is over and voters can factor this in as information not a buy nary system of whether this was illegal or not, they say they didn't find that. >> the investigation is over but the investigation of the investigation has not even started. >> that is true. >> the process was
abnormal-- abnormal. so was it abnormal good or bad, people will debate that. typically you don't have an announcement like this prior to the d o.j. macking what is supposed to be their final call. so that is going o to be debated. i think at this hearing that the house republicans are leading this week, this is a legitimate area of debate. the fbi director regularly testifies, there is regular oversite, and now he can't say this is an open inquirery, i can't talk. he will have to answer those questions so that is also legit and relevant. >> eric to the point you were making, democrats might well say, mr. fbi director, under normal circumstances you might have issued a release, a staim or you might have just come to the microphone and said we looked into this, it doesn't warrant a charge. thank you very much, i'm leaving the stage. >> we rarely see democrats or republicans united in congress. i think they might be united at this hearing tomorrow, both asking really tough questions of the fbi director from different perspectives. the republicans want to know why did you not recommend charges.
and why did you make that public. the democrats are going to say why did you go public with all this criticism and was that proper. so he is going to get it from both sides am but i think he's prepared for that. you are right he did not take questions yesterday. i was sitting in the room and we certainly wish he had. but i think his aides knew that he was going to get hit from all sides on this. and the reaction has not been unexpected given how remarkable this case has been from start to finish. >> as you were sitting there, what was the question you must wshed you had a chance to ask the fbi director before he left? >> that's a good one. well, i think i would have asked him, given the standard that he laid out, why he didn't consider or recommend misdemeanor charge which a lot of people are now questioning as well, if there was, without intent, may be gross nedges, would that have been proper here? >> i think we are actually circling one of the important long-term issues here, which is there is a tension between
uniformity that we expect under the law. please treat everyone the same, please approach everyone the same, regardless of who they are. union fit. and transparency, tell us what is going on so that we know and can factor in, especially if you have a subject like this, who is in an election and voters want to make decisions. that is a tension to pick up very different legal example, we saw that play out in a lot of these grand jury proceedings around officers who were accuse-- accused or suspected of improper use of force. and we saw a very similar tension where some of these prosecutors who show were clearing police officers, involved in controversial shootings put out more information than usual about grand jury proceedings which are typically as we all know secret. and they are supposed to be secret and that is considered a good thing. >> in ferguson, where it was a big controversy, and so that immediately goes back to that tension. and i do believe the internet and the public scrutiny is part of this. things that we used to accept more as a culture or society as being done over there, we don't have information, are under pressure, not legal pressure
because grand jury proceedings are legally secret, period. but under a type of extra legal or political or public pressure to say we're used to hearing more about other things, why don't we hear more about this. this say very different scen ar yovment obviously my analogy is not to the underlying conduct but to the procedural tensions. but jim comey was trying to do something there, he says that was for the good of the institution of the fbi, just as some of those prosecutors said we don't usually release grand jury material but we are going to because of exceptional interest. but again, you cannot meet, i don't believe, meet that goal without the tension-backed unionity where other people are saying well, gosh, other people don't get this sort of grand jury treatment that these officers got. and that then seems unequal. that say tension that is very hard to solve, as for the question, i think the biggest question would be even if you can't describe all the classified materials to us by definition, director comey, did you see the type of thing that
really gave you pause as to how anyone could do this. or did you see the type of thing that was looked bad in retrospect but wasn't actually at that high level. because as reporters we still haven't seen the underlying classified material. >> eric, one thing that the fbi director said that caught a lot of people's attention was well, we here at the fbi are not involved in whether or not you should continue your classified access status or you would receive administrative punishment. many people would have said if hill roe clinton were in the federal government, she would have her classified access denied and she might be fired or demoted. do you think that aspect of this will continue to be either debated or screut niezed? >> sure, in fact, we're writing about that question today. and we saw paul ryan calling for her classified briefings to be denied as a presidential candidate, the major presidential candidate by custom, get presidential-- get
classified briefings. >> after the conventions, right. >> in light of this, she should not. so there will certainly be ramifications, the state department has been looking at administratively whether those people are still employed who might have been involved in this should be in any jeopardy. and i think mrs.'s no doubt that there will be ongoing questions involving security status for people involved. >> because that is an important process, about not necessarily legal question but a process question. with gravity. >> let's be clear. we were talking a lot about the law. when you talk about policy, what is going on in someone's mind who is in public service with account ability to the government and a system that relies on the expertise of other officials who is so consumed with secretary resee or routing around the normal rules, legal as it may be, that they need to construct an entirely physically separate solo operation to control everything.
as a policy question, that raises major questions about that person's trust in the system, ability to work with others and not that privacy always means you have something to hide but are you not entitled typically for that kind of privacy for work material. i think that's a big question about why is this individual, hillary clinton, so prone to secretary resee. >> all right, thank you so very much. >> thank you. >> we continue this evening with politics, specifically presidential politics. joining me from cincinnati, bob costa of "the washington post" and in washington colleen nelson and-- bob corker takes himself out of the running as the potential running mate, where
does the short list for running mated reside. >> great to join you, here in cincinnati, vice presidential speculation continues for donald trump. he will be at an event here in the evening with newt gingrich, the former house speaker but hours before then, i got a call from senator corker, the chairman of the powerful senate foreign relations committee, and he said even though he spent eight hours with donald trump on tuesday, meetings at trump tower new york, and an event in the evening in north carolina, he believes because of his temperment, he's low key, his focus on policy, he decided to formally bow out of consideration. this immediately narrows the list for trump. it is a sudden departure because many people thought corker was being seriously considered. he was under vetting by the trump campaign lawyer. but this shows you some reluctance people like corker and others have, perhaps of being on the ticket, of being with someone like donald trump, a celebrity presence, a gre garious presence and someone who is controversial. >> does this really bring this list down to two, chris
christie, new jersey governor and the former speak of the the house, newt gingrich? >> no, i think the list is more expansive than that. i think one person to really watch right now is indiana governor mike pens. he had a very warm meeting with trump over the weekend at his golf course in new jersey. because of his house experience, timing in indiana who can bring heft to the trump particular. if trump wants to pick someone more like trump, someone who is bat combative, tv savey than christie and gingrich are in the conversation. but my sources tell me inside of the campaign that paul manafort. campaign chairman, he is urging donald trump to pick someone with deep washington experience and someone who may also reassure those in the party, especially the leaders in the party who remain wary of trump. >> and mike pence has a very strong profile among evangelicals an prolife voters, that could be an asset for trump, could it if the. >> is he someone with deep roots in the evangelical community.
someone would you always see at conservative conferences, especially when he was in the house. is he someone who grew up in the midwest. aes he got the cadence of a talk radio host, someone who is soft spoken yet very articulate. he actually was a talk radio host in indiana before he ran for office. his nickname, his mondayicer was rush limbaugh on decav, that is what they called mike pence back if the day. >> what do you think is the most important political, not legal but political takeaway for hill-- hillary clinton. >> i think the political takeaway is this isn't over. we saw republicans moving very swiftly yesterday and more so today, to make sure that that is the case. this will continue to be a political problem for her are, could potentially be administrative problem, and republicans are doing all they can, certainly, within their purview on the hill and elsewhere to try and focus on the first part of what director comey said yesterday, which was
a very, very strong indictment in every way. but the legal sense of the way secretary clinton had conducted business on her own and with her aids-- aides at the state department when it came to security practices. it was a very scathing report he delivered. and that is going to be the substance of the republican charges going forward. it's also a political liability for her outside the strictly partisan sense in that it is confirm tore for a lot of people, including some who will end up voting for her, that she has a trust problem. that there is always kind of a lingering question, has she said everything that she needs to say, is she telling the truth. and that whole set of issues that are raised by the e-mail problem will continue for her. >> colleen, what is the atmosphere within clinton's innercircle?
>> well, i happened to be at clinton campaign headquarters yesterday as director comey was speaking. and the atmosphere there definitely was very happy, very relieved. they didn't have a heads up that this was coming, so they were watching this in realtime, getting the results from director comey in realtime as everyone else was. and they were unequivocally happy about this, they really did not focus on as ann says, what was the scathing comments from director comey. the message from the clinton team so far has been this is resolved. we are happy this is resolved and we are moving on. but it may not be that simple. and this, even though there wasn't an indictment, there still was damage done. and this is something that they're going to need to keep answering for. >> there has been some hand ringing already in republican circles that this was a golden moment for donald trump, that he mishandled yesterday, that there wasn't enough, donald trump surrogates there weren't nuch
prominent republicans, he himself didn't deliver a strong consistent message on that. is that something the clinton campaign might find a narrow advantage in in the aftermath? >> yeah, i think it might even be more than a narrow advantage. and it just goes to z the very different kind of campaign operations that the two can gats have felt. and hillary clinton started in a very deliberate manner, well more than a year ago, to build an organization that includes large state operations, an enormous fundraising operation, a vast headquarters operation with surrogates with political intelligence agents, basically, all over the country, giving her advice and tips. and who can be deployed to go out and speak on her behalf. all of those things take a long time to build. they also have an advantage in clinton-world in that a lot of
that network, a lot of those systems already exist after three decades as it. so he's got a lot of makeup work to do if he wants to have the same kind of resources at his finger tips. and what you saw yesterday may be indicative of the fact that he simply doesn't have that material. he doesn't have all of those networks, all those con stactes-- contacts and all those networks at hand. and there is the fact of his own temperment. i mean he could have simply focused like a laser on the first part of what comey said and hammered it for 20 minutes. and he didn't. and that is something that his campaign is going to have to decide whether they want him to do. >> bob, do you sense unease about the trump reaction, specifically, an unease about will the trump campaign still
isn't as it approaches its convention, meaning organized at the field level, cohecht in terms of messaging and strong enough in terms of fundraising to compete with hillary clinton? >> i think there is also a sense of resignation among people many people in the upper ranks of the gop says that trump, if he has any chance of winning this general election, it's going to be because of his presence, his ubiquity on television, his ability to blend his pop lism and celebrity into something that is in utter contrast to secretary clinton. the republican national committee is doing things on the ground. it's spread across the country. but they do not have the organizational depth at the trump campaign that you see with clinton. and you also don't see trump's presence on the airwaves in terms of paid advertisements. but because it's july, major, there is a sense that maybe this can't be fixed and have to rely on the force of personality in their own candidate to try to see them through.
>> and this sort is of distraction that trump created for himself with his speech in north carolina, where he diverted to sideways praise of saddam hussein, this seems to me to be something that fits in the larger sense of anxiety, republicans have, not only about trump and what he says but his potential orientation to the vast po wrt-- powers of the presidency. >> he praised that he could get it done. cokill people in essence without any kind of trial that is what impressed trump. this sent off shock waves throughout the foreign policy establishment of the republican party. they say who is this. who is the nominee. but is he the nominee and that is not going to change. trump is so resistant to advice from party leaders that this idea that he could kind of hold back a little bit more, i think that's a realization that is probably not going to happen. >> colleen, let me ask you something about an item, data point that caught my eye in the
"u.s.a. today" suffolk university poll that came out yesterday. it said only 12% of those surveyed were undecided and that nine out of ten of trump supporters and nine out of ten clinton supporters said nothing could happen to dislodge them from their support of either clinton or trump. how important in that con feks-- context do you think this week has been? >> well, that say great question. and voters certainly have had decades worth of exposure to hillary clinton. so many of them have made up their minds, made up their minds about hillary clinton years ago. and are not going sto be movable whether they like her or they don't like her. and donald trump obviously has been a very controversial candidate. i think yesterday kind of encams lated why boft these candidates are relatively very unpopular with voters, in a nutshell. because you had the decision from the fbi which really got to the heart of questions about clinton's trust worthiness, her
credibility with voters, and then you had donald trump standing up and essentially praising saddam hughes heuses, kind of going off message and just kind of reinforcing doubts about whether he is a wildcard, whether he will-- is able to stay on message and do what party leaders in the republican party would like to see him do. and so a fascinating encapslation of voters' doubts on both sides of the aisle about these candidates. >> we have what seems like an ongoing cultural experiment going on in this country, the power of celebrity ver us the power to use your phrase a couple of moments ago, an onrunning three decade political franchise, this seems to me to be unchartered territory. >> it really is. and it's teutly fascinating. i mean the typical battleground for a presidential election is the middle. and in both cases, this hillary clinton has doubled down on being an establishment candidate with all of the advantages,
certainly, organizationally, moneywise, that that brings. it's also her comfort zone, that is not the middle in this unorthodox year. and donald trump has shown no sign that he is going to be drawn to the sort of what he would here to think is the political mushy middle where you you're a conventional candidate. if anything he has doubled down on being an unconventional candidate. and he's from the get-go, profited by the element of surprise. everyone is political friends and adversaries, us, everyone, have to wait and see what he is going to say or do next. >> bob, i will have you give us the last word because we are coming up about a week and a half away from the republican national convention. a crucial moment, i don't think it is anything other than obvious to say that for donald trump. and the some what anxious party around him. what are you anticipating from that convention? >> i think you are going to be
struck by how many republicans are not there. i mean compared to the democratic convention which will be pop lated by a president, fompler president, vice president, senators and governors, that democratic convention will be a display of party unity in cleveland, i'm already hearing reports from conservative activists that they plan some kind of protest. they know they can't get any delegate change really, as this point, this late stage to get trump away from the nomination. but they want to protest on the floor and outside of the q center, in cleveland. >> and do you fore see a convention that reinforces this power of celebrity where trump is sort of the dominating factor every single night and he really doesn't step aside to let other republican voice hold center stage? >> trump has told me and others in the past that he had once thought about speaking every night at the convention. and he may actually make some appearances according to his advisors, each night at the convention. but this will be all about
trump. this will not be so much about the party. >> excellent, robert costa of "the washington post," ann garrett also of "the washington post," colleen mccain nelson, "the wall street journal," thank you so very much. >> now for something completely different. no, this is not a segment of monty python, but it's an interesting topic, one that is here to stay, that could change the workplace, our culture, our sense of humanity, and our relationship not only to one another but to machines. topic, artificial intelligence and what it may or may not bring. joining me from chapel hill, associate professor at the university of north carolina zeynep tufeki, she covers the social impacts of technology. here with me in new york, lucy suchman, a professor of anthropology and science and technology at lancaster university. and nathan yal popper is a business reporter who also covers technology for "the new york times," an author of book "on bitcoin called digital gold.
lucy, let me start with you. artificial intelligence has now become a big topic for a couple of reasons. one, perceived advances in artificial intelligence, and their potential consequences and effects on the workplace and society at large. are those advances real and you can separate out where they are occurring and why they matter most. >> that is a great question. i mean if we go back to the middle of last century, 1950st which was when the phrase artificial intelligence was coined and the field started out, i think what we have seen recently, so that is what, 50 plus year, almost 70 years now, i think what we have seen recently that really changed things is the add vebt of very, very large data sets. and very fast computers. and a lot of networking. so those are really significant developments that they've made an enormous difference and i'm sure we will talk about those. >> just pour the data in, and
then what happens? >> well, if you have a lot of computing power, then you can do most of analysis of those data that would be impossible for humans to do. and i'm sure we'll talk more about that. but the direction that have i been following most closely is more the humanoid robot, so the project of really trying to create machines in the image and in the person of humans. and that is a project i would argue that has gone much more slowly. run into all sorts of deficits which i think are really dis -- really revealing of what we are as humans and what machines are and their capabilities are and limits are. so i think we fleed to sort of make a distinction between those. >> i think you would agree that if you are going to offer a script to hollywood, you make a script about scary robots taking over the world and crushing not only our spirits but our humanity right along with it and
taking over the world. but perhaps a more prosaic topic but one more important is data, and its impact on artificial intelligence and what they render or tell us about the world that we live in. >> it seems like the more immediate threat at this point is much more boring robots, essentially. your desktop that can do-- do ten percent more of your job this year and maybe 20% of your job next year. and that's a more slow-moving and less visible threat. but i think it's something that is maybe a more immediate point that we need to be grappling with. and i think when we talk about what we have seen that's new, you know, when we-- automation and the idea that automation, that the car is going to replace the horses that came before it and the bugie makers, we've been through this many times where you talk about new technology, replacing people. and generally what has happened
is that those people have found new jobs. and i think some of what i have been most interested in the last 15 or 20 years is some indication that there is some large number of people who are being displaced, maybe too quickly to find new work. and so i think that's one of the really interesting problems right now, is the fact that we've been through this cycle a lot of times but maybe this cycle does turn out differently. and you know, the workforce participation does go down and you start to see real displacement of jobs. >> zeynep, technological advancements that are benign over the course of 60 years and maybe beneficial over 08 can be terrifying and enormously disruptive in the first 20. if we take that as a premise, where do you think we are in this cycle with artificial intelligence? >> i think we just passed the new inflection point.
up to now, most prefer programming was done by writing code which is giving computers really detailed exact painstakingly detailed, if fact, instructions. what has recently happened is two fields have merged. machine learning, a field in which machines can learn by themselves, has met with the large data sets that lucy was mentioning that we now generate in our digital lives. and this is a new development because what is happening now is really different. we're setting these learning algorithms, sometimes called deep learning, we're setting them lose on the data and we're saying things like tell us who will be a better person to hire. tell us what news items should be recommended. and then they just collect this data. and then pick winners and losers. and the trick here is they're pretty good prob-- at picking winners and losers but we no longer understand the basis on
which they do. this i think it is the first really major step towards not just artificial intelligence but artificial general intelligence, that is learning to learn beyond our capacity to understand. and that's-- that is both exhilarating as a person but also scary. because we don't control these new things the way we did our old programs which had other problems. >> so zeynep, if i understand you, what you just said is we understand the question we are asking but we don't know how the answer is being rendered, is that right? >> correct. so think of the scenario. you have a hiring algorithm going through applicants data, including social data and you train that hiring algorithm on previous high performs, say go, my hiring ago gorit im, fin more high-performing people. and it doesment and you look at your performance data and it is looking great, it is doing betser than human hires, but here is what you might not know. maybe, and you don't know this, it say black box to you, maybe it's weeding out everybody with higher likelihood of depression,
not people who are depressed now, but just who have a higher likelihood, social media data can reveal such things, even if you have never talked about depression, that is their power. maybe it's weeding out women more likely to become pregnant in the next few years. maybe all these weird shady things, but we don't even understand that it is doing it. all you know is, hey look, my performance numbers have gone up, it is like-- it's less like coding something, it's more like raising a kid, training a dog, and now we're training these intelligent systems. and we're letting loose on all sorts of decision-making. and what are they doing? you ask the prommers, they don't know. you ask the people who oversee the system, they are like, i don't know. and that is really an important twist. >> the politicians really don't know. they don't have any idea. >> they don't know. >> lucy, on the plus side, i have read that this deep learning has created artificially intelligent
radiology-- radioediol guests with a better track level than humans but on the negative you have rendering suggesting some race is more likely to be a resid vis than other race, walk us through the complexities and real dangers. >> i would like to go back to the whole notion of learning here. because i think one of the things i've been most concerned about in relation to ai is the kiepped of anthro ddz pomorphizing language being used. could you look at it as trying to make things more understandable but i think it really mystifies and obscures what is going on. so the algorithms that zeynep was describing which are very powerful but we could hear from her description that they are basically reproducing the things that they're operating with are parameters that have been introduced by us. so and many of those are stereotypes. so stereotypes about what women
are going to be doing at certain ages of their lives. stereotypes about who's most likely to be vulnerable to depression and now we have predicted policing where we have stereotypes about who is going to be committing crimes based on statistical analyses. so in those ways, we're actually, rather than the machines operating on their own, they are actually reproducing and intensifying the stereotypes that, the assumptions, the free suppositions that we have basically defined into them. and i think the more we could actually rather than talk about machine learning, because learning involves all the things we think about with human learning, if we could actually talk more, in a way more technically but i think those can still be done quite intel i believably about how the systems actually work, and what kinds of parameters, again what kinds of
profiled assumptions are built into them, i think we would have a much better sense both of how they work and why they, you know, are very effective in some areas and what we should be concerned about. >> what kind of things could be done to address this now? >> well, that's a good question. and it's a hard question. and kate crawford who writers about these questions in a very compelling way, points out how difficult it is because we don't have a lot of access to these algorithms. either they are being developed by private companies and they treat that as proprietary nfertionz or they're being developed by governments and they see these as security issues. so i mean i disoo think it is the case that to some extent nobody fully understands how these are operating am but certainly there are people who understand a lot about this bias that are being built in. but we don't have much access to that. either as, now i analysts and
critics or as the people who are subject to being profiled thr are data sets widing out people at the front end of everything they applied for and you don't know why or what the data set is, is there not a civil rights implication here s there not a basic employment rights? implications. >> there actual absolutely is. let me play the optimist for a moment and note that along with all the other data we have, is the outcomes from the algorithm ims. we recently saw this in the case of this sort of sentencing algorithm imthat the states are using and able to go through the numbers and take a look at how it was treating black defendants and white defendants and how-- how unequally the black defendants were being treated because of this algorithm. we have found that out. it is not something.
we found that out i think a lot sooner than we might have in the old criminal justice system where you had data scattered and courthouses. where you don't even know what the imbalances and injustices are. i mean the flip side of that, i think it is a situation we have seen in the last few weeks is with self-driving cars. i think some of this, it dubt come back to understanding exactly from day one how it is going to work. but to take it slowly, essentially. and what you saw something very interesting to play out in the south driving car world is that tesla moved very quickly with it's technology because they wanted to be in the market with this. and you have seen google with the same technology moving much
more slowly, to understand both how the car works and how the human responds to that car. because you just don't know, and google has been much more slow to on to the highways than has tesla. and particularring it lowly-- slowly and wamping how it works, i think is probably going to end up being one of the best answers here do you sense that within this world, this largely private world of artificial intelligence and the dram ingvar-- dramatic increase that this is slowing down, it seems to me based on what i have been reading if it is accelerating it is going up fast. it is important to understand one thing, the sentencing algorithms, look how the people got sentences or pa roll and we
can check it against parameters that we know like race and gender there is way to audit these to catch to understand. but with the new machine learning systems, evened programmers may not understand it that is why design, that is the power of doing something that nobody programmed into it. so in the examples i give, there is no variable label depression anywhere to figure out this where corporations that i know, that are going as fast as they k because it has upsides. it could be less gender bias, for example, if you control for it because it's something you can check for. what worries me is that the very technical nature of what is exactly going on means that stuff that you don't anticipate like gender and race that you
can check, is really no way at the moment, technically, of even knowing what the system is selecting on, let alone go back and check its bias or what else it may be doing or where its predictive power comes from, that is the problem. on the one hand as you guys say, it's moving very tans. it is cost efficient in some ways. it's a better human-- it's better than human-- in somer ways and yet it comes with all these huge downsides and worries and potential bias that it picks up from the data or figures out on its own that we don't want it to figure out, that i don't think we're really paying a lot of attention to. and we should slow down and say wait, we should really use black boxes like this, unless we figure out how to audit them am how to make them interpret, how to check them and how to really think about what it means to put these things in decision making
positions. for-- there was an oxford study that suggested that the replacement percentage through robotics an artificial intelligent creatures could be as high as 47%. economist said it could be as low as 10% but that is still a lot if you are talking about an enormous workforce like we have in the u.s where we are heading in robotics and employment replacement. >> right, i wouldn't in anyway want to downplay the problem of job displace am. i think that is a really crucial problem and that is the preface to say that tilely i think the ability of robots to replace humans is highly constrained. robots work best when the world has been arranged in the way that they ned it to be arranged in order to be effective thasm is why assembly lines work so well. that is why amazon's war housework robots work so well because the entire world that it
is working in is designed in relation to the robot. but as soon as we get to open-ended domains and interestingly one of the areas most resistant to being robot sized is housework. you know, which we would think of as typically, historically as being manual labor. but actually it turns out that it is a very open-ended indeterminate environment you are working with with incredibly large range of skills which is more the case when it comes to things like care because we now talk about elder care robots about which i'm extremely sceptical. but i think, so i think we need to again differentiate much more carefully between the areas in which robots are incredibly powerful and look at what kind of job displacement is---- we need to be think being there. and differentiate that from other kinds of basically any kind of a work that involves engaging with an unpredictable, even in small kind of every day
ways environment. and there i think actually it will be-- there is very little progress that has been made. >> some of this wizardly can now be done by robots. >> it certainly has been well compensated. and i think wall street because of the pressure on it since the financial crisises had been pushed to look for efficiency anywhere it can and it is an information-based industry it is about fundamentally about processing information. and that is something that lends it self to automation, to machine learning to use the phrase, a lot more quickly, i think, than maybe even housework. but what you have seen on wall street already, i mean you look inside a firm like goldman sacks where you would think that everybody there is already a skilled worker. you know, these are obviously people all college graduates,
especially, but what is fascinating to see is how much of that work is at the end of the day still pretty much work. and so you might have two employees and you might replace one of them. you still do need people there to make hard decisions, to deal with the difficult situations. but if you cut half of the employees, that is still a lot of employees. if you cut out, you know, two people, the rout work in two people's jobs and then you only need one of them that is still a big shift. and so that is the kind of sort of slow-moving change that i think is maybe a bigger threat than worrying about the robot, the self-driving car, the weapon, automated weapon, it is the kind of thing that is going to play out so much in everybody's lives much more quickly and obviously, journalism, in my profession, we've seen this already. and it is sort of, it's the slow
creep, it's not an on-off switch. it's the slow creep of being able, realizing the next step that you need to automate and realizing how to sort of isolate that and make that into something that can be automated, even if the whole job is never automated. >> i will give you the last word. if you look at the entire human history, paid labor is some what recent invention, about 300 years. before that it was largely con d the opt mission-- opt missesr. say look, with the advances of artificial intelligence, will you remove people from the drujery of work for pay and allow them to open up all sorts of creative vistas that they can't even imagine, will be beneficial for society that is by far the most optimistic appraisal of a future that art fix intelligence is a main driver of. what is your sense of both the optimist presentation and either the pest mist or sceptists, orthos sceptical about the
real-life employment effects of artificial intelligence. >> so i will say that the optimist scenario is certainly plausible and potentially doable, if there say political and economic will to redistribute the benefits we get from automation, that is-- i will just put that out there. but let's look at the way the world really works at the moment. self-driving cars sound great in many ways, they're safer probably but driving is employment of last resort to hundreds of millions of men around the world. so what happens when that changes? if you look at the u.s. service industries a good chunk of it is basically natural language prosession. somebody looking at you over the phone or through a window and puts what you just said into the computer. natural language processing, you read the technical papers every year is now like this new revolution news cycle is coming. so those jobs are on the chopping block. and if you look at things like
elder care which worry moo too because there is the issue is we have vulnerable populations like elderly people or kids with disabilities, there are expensive to take care of. to be honest. but we don't want to spend that money and i fear that they are going to beware housing socialable screens that have some sort of ai chat bot paying humans to do that kind of job and do it well, and preserve the dignity of the whole situation. so i am actually quite pessimism because of how fast the labor market affects our coming, and how to put a t politely how infectionive and clogged up our political system seems to be, even the simplest problems aren't being soferled. and we've got this avalanche of employment affects coming down really fast. so i have this urgency, to ends on a positive note, the
optimistic mist scenario is possible if we say we are going to do it this way as a world, yes, why not. >> so you tell me there is a chance. >> i would like to hope so. >> university of north carolina, thank you so much. nathanael popper of "the new york times," lugsy suchman, fascinating conversation. i'm indebted to you all, thank you. >> for more about this program and earlier episodes visit us online at pbs.org and charlie rose.com. captioning sponsored by rose communications captioned by media access group at wgbh access.wgbh.org
>> announcer: the following kqed production was produced in high definition. ♪ >> must have soup. >> the pancake is to die for! [ laughter ] >> it was a gut bomb, but i liked it. in private moments about the food i had. >> i didn't like it. >> you didn't like it? oh, okay. >> dining here makes me feel rich. >> and what about dessert? pecan pie? sweet-potato pie?