0:01[music] Every day, [music] people wake up and ask Siri for the weather forecast. Click the mouse to check email. Connect with a friend who had surgery with advanced robotics. [music] Check their bank balance. Use their PC and the internet to [music] talk to a
0:30colleague or turn on the TV to watch the news. It's only 11:00 a.m. and already [music] billions have taken advantage of the deep technology that SRRI and the labs that became a part of SRRI [music] delivered to the world. These moments trace back to one place that began in a region now called Silicon Valley. To see how we got here, let's go back to where it started. It's [music] 1946.
0:54World War II had just ended. A team at Stanford and local business leaders [music] wondered, "What if we put a team of scientists to work on real world problems?" On November 6th, 1946, they [music] officially voted to create the Stanford Research Institute. That's how SRRI was born. With a tremendous [music] amount of curiosity, passion, and ambition, those early years were full of innovations that carried us through generations. [music] Walt Disney hired us to help pick the
1:23site for a theme park. We helped Bank of America invent Irma [music] to automatically process checks. That's the ancestor of the modern ATM. We studied [music] smog before most people understood the problem. We even invented the world's first teleresent surgical robot and worked on extending the [music] flavor of chewing gum. In the 1960s, Doug Angelbart and his team developed the computer mouse. In 1968, Doug gave a live presentation [music] that showed the world a graphical user
1:51interface, hypertext, and video conferencing, now known as the mother of all demos. We [music] built Shaky, the first mobile robot that could reason about its own actions. In 1969, [music] the second computer ever connected to the ARPANET was at SRRI. In 1970, we became [music] an independent nonprofit.
2:12And in 1977, we became SRRI International. Today, we're [music] known simply as SRRI. The mission is the same. Bold ideas, real impact, worldchanging [music] work. Over the years, SRRI grew. In 1987, we welcomed the Sarnoff Corporation, formerly RCA Labs, [music] in Princeton, New Jersey, the legendary lab that invented color TV, high definition, and the liquid crystal display. [music] In 2023, we
2:40acquired Park, the historic lab that gave the world the graphical user interface, the laser printer, the Alto [music] personal computer, and Ethernet. Today, curiosity and passion [music] drive our work in identifying AI deep fakes, delivering life-saving [music] diagnostics and treatments, perfecting quantum sensing for real world applications, creating [music] ecosystems to deliver the next wave of impact, and so much more.
3:06[music] We look forward to the next 80 years. Thank you for celebrating with us.
3:13Hello. Hello. Hello. Welcome everyone.
3:16Welcome everybody in person. Welcome everybody in line. Uh I'm Mark Edkind. I'm the CEO of the Computer History Museum. How many people here have worked for SRRI? Yeah. Let's give everybody a round of applause.
3:29[applause] you have literally created computing history and you've helped invent the future. So, thank you guys for all you do. Um, right now technology is moving at such a rapid pace. Even folks in Silicon Valley who've been here for a long time say, "Oh my god, it's moving so quickly." Um, there's a lot of disinformation. There's a lot of confusion out there. And here at the museum, we feel we can help people navigate this change. We do it through
3:58having conversations like this, uh through having discussions, through having engagement. We do this through our exhibits. Um we're trying to provide context, historical information. We decode the past to help people understand the present to shape the future.
4:15And no organization has shaped the future like SRRI. Over the course of 80 years, they've been at the forefront of every technological change. And I'm proud as a museum that we've been collecting SRRI's history and sharing their stories. We have Shaky the robot as you saw in that video. We have Doug Engelbart's first mouse. We have the van that launched the internet. And I'm really proud to announce today that we've been working closely with SRRI.
4:42And we've just gotten a large collection of SRRI historical documents uh uh technological records, photographs, uh that we'll be preserving and sharing with historians, with technologists, uh with the general public to tell the story of everything from the arponet to AI. Now, I'd like to introduce tonight's host. Uh his name is David Pogue. You may know David. Let's give David a round of applause. [applause]
5:13David uh has won seven Emmys uh on CBS Sunday Morning. He's been on 20 NOVAs.
5:20Uh he's a best-selling New York Times author. And I'm I'm going to give you a little plug here. Uh his latest book was uh Apple, the first 50 years, which he wrote uh using CHM's archive, using our oral histories, uh working with our curators. Uh David has been chronicling the uh technological revolution uh that SRRI has been pioneering and there's no better person to lead tonight's discussion. So let's welcome David Pog.
5:47[applause] Thank you.
5:51Thank you David. Thank you.
5:54Hey, it's great to meet you. It's great to be back. This is the third time I've gotten to host uh events on this very stage. And it's always great to be in a room full of fellow nerds because you get all the jokes every time like you know what's the difference between an intro introverted engineer and an extroverted engineer. I don't even need to tell you you already know the answer to this joke. An introverted engineer looks at his shoes when he's talking to you. An extroverted engineer looks at
6:25[laughter] So anyway, thank you for being here. Congratulations for getting in. This is a soldout event and there are people watching online. So tell you what people in the room with me, let's greet the people online with a cheer. [cheering] And now let's do the opposite. Online people, let's hear it.
6:46[laughter] Joke works every time. Um, so this is a celebration of the most important tech company nobody's ever heard of. Uh, tonight will be loosely divided into three sections. um the past, the present, and the surprise guest. Um and to to kick us off, we're going to look back at the origin story. So, we have with us John Leven, the president of Stanford University and [applause] and David Perrick, the CEO of SRRI.
7:21[applause] Good to see you, sir. [applause] Hey, David, please have a have a seat in my living room here. Um, all right. So, what is the origin story? What is this thing? Is it a nonprofit? Is it a think tank? Is it a corporation? What what's there's some kind of Stanford origin story there? There is uh first of all, thank you for having me here and great to see everyone and what a this is a fantastic event.
7:50So, SRRI was started in 1946 and it was at the intersection of the basically the two forces that started Silicon Valley.
8:01That's like 80 years ago. [laughter] Oh, yeah.
8:07Sorry. Sorry. Sorry. He's like, I can't work like this. Uh, sorry. Sorry. I'm done. I'm done. Go ahead.
8:13So, so [laughter] Okay. So, at that point after World War II, two things were going on. One was Stanford was trying to go from being a good regional university to a great global university or first national university and they saw that the way to do that was going to be through becoming really excellent at research and leveraging the engineering school's capability and the money that was coming in from the federal government to do research.
8:38That was one big force. And so they were going out they were trying to recruit people like William Shockley who started the semiconductor industry and and others and to spin eventually to spin companies out. Um and that's sort of one of the origin stories of Silicon Valley.
8:53The other thing that was going on was that the defense industry was coming into this area through Loheed Martin and NASA as and other ways and they were bringing a lot of money in that was basically a pull on technology. The Stanford Research Institute was actually the idea for it was around at Stanford much before actually back in the 30s.
9:13Herbert Hoover and others thought it would be great to have a a an applied practical research institute at the university that would work with local businesses to help the local economy.
9:23Then there was the depression, World War II intervened. 1946, balance professors, president of Stanford, they start the Stanford Research Institute on the campus and it was actually intended to work mostly with local companies. But within a year, it was doing most of its work for the federal government and it was doing research. And so within Stanford, it had faculty and others doing contract research for the government by and large and also for other companies including local
9:51companies. uh and it did classified research which distinguished it from the rest of the university and you just saw some of the innovations. David will talk more about them. That went on for quite a while and it stayed part of Stanford on the Stanford campus all the way through the 1960s.
10:09During the Vietnam War, the students did not like having classified research on the campus. There was a lot of protesting and eventually there was so much protesting of the war and of Stamford being involved in doing classified research that Stanford spun SRRI out as an independent entity in 1970. And that launched the sort of modern-day SRRI as a as an independent organization that's continued to do great research for many
10:38many you know many many different sectors. So you say as an independent institution, David, would you would you consider it a is it a company? Is it like an Apple or Google is like the objective to make money?
10:50No. So we're an independent nonprofit mission profit. Okay.
10:53So we're about very simple things. Uh to create a world that's safer, healthier, and more sustainable. Uh and it seems like motherhood and apple pie, but we have people coming from all over the world to come work with us toward that mission. But in many ways we operate like a company in terms of business principles. In fact, some of our board members are here who are also on board members of many national companies. Uh and so when we look at SRI today, it's about partnering with the ecosystem in
11:22order to deliver these innovations. And one of the fun stories for us, um, I just started 5 years ago and at my first board meeting, one of the board members, Isaac Stein, uh, who has rotated off since, uh, in terms of finished his service, he tells the story that when he was a student at Stanford, he protested to get SRI kicked off [laughter] and then now many years later, he was on the board of SRRI helping SRI fulfill a wonderful mission.
11:46Also became the board chair at Stamford.
11:48So I it just shows going being a student protester doesn't rule you out from later service. [laughter] Yeah.
11:54and our longest uh standing board member, wonderful lady Marian Byerwalter, 25 years served and she was also CFO of Stanford. So even though we were spun out in 70 and I was a Stanford alum as well finishing my degrees there in the 80s, we still have close ties in terms of the work we do together. Um but I think that independence and the the beautiful thing about that mission that you were talking about in terms of helping companies is we've done that not only it was initially for the west because no research institutes in the
12:22west it's grown to be global not only across the US but around the world in fact we have some very good friends from Japan who are here we have a major innovation center uh with Namora and with about 30 different Japanese companies that are right in our backyard in Silicon Valley because of the partnership we've developed you So when I worked on this Apple book, I got to this chapter where Steve Jobs came to Xerox Park and stole the ideas for the mouse and menus and overlapping
12:50windows. And the more I dug into it, uh it is it is true that uh Xerox Park had brought those along way, but the more I dug into it, I realized that they stole those ideas from an even earlier source.
13:11If in your office you as an intellectual worker were supplied with a computer display backed up by a computer that was alive for you all day and was instantly responsible responsive, instantly responsive to every action you had. How much value could you derive from that? Well, this [music] basically characterizes what we've been pursuing for many years in what we call the Augmented Human Intellect Research Center at Stanford Research Institute.
13:40We're going to try our best to show you rather than tell you about this program.
13:49Okay, there's Don Andrew's hand in Mineral Park. And in a second, we'll see the screen that he's working and the way the tracking spot moves in conjunction with movements of that mouse. I don't know why we call it a mouse. Sometimes I apologize. It started that way and we never did change it. This characterizes the way I could sit here and look at a completely blank piece of paper. That's the way I start many projects. So with my system, that's a good start. I'll sit
14:17here and say I'd like to load that in. So I'm putting in an entity called a statement and that's full of other entities called words.
14:28And if I make some mistakes, I can back up a little bit. So I have a a statement with some entities words. And I can do some operations on these. I can copy a word. Say that word like copy after itself. Let's make more statements. I'll say copy that statement. And lo and behold, I have another one. Copy that one. Another one. I can even copy groups of statements. I can say after that one, copy the group from there to there. And it does. Well, I'm going to do something
14:57called jump on a link. And a link is something that'll go between files. So, what it's going to do, it says I'm going to go to your file name, a CNR. So, here's what I drew with a picture drawing capability here. It's a slight map if I start from work. And here's the route I seem to have to go to to pick up all the materials. And that's my plan for getting home tonight. But if I want to, I can say the library. What am I supposed to pick up there? I can just point [music] to that and oh, I see
15:26overdue books and all. Well, there was a statement there with that name on it over. So, in his display, he sees my text. So, I'll execute it. And sure enough, it does. But what's that running around? Well, he's looking at my text.
15:39He'd like to have something to say about it. So, we put on a marker, a tracking spot that he controls. So, he's sitting in Menal Park looking at this text, and he can point to it. But, we've carefully reserved for me the right to control and operate on this. So, my bug is more powerful than yours. But we can have an argument. Yeah, that's what they call a bug fight.
16:04So, we've set up now audio coupling and we're both looking at the same display and that'd be very handy to work. We can talk to each other and point and maybe later I can hand you the chalk on this blackboard like saying here you control it, but let's stay this mode now and add another feature that hardware wise is available through the kind of display we have. I'd like to see you while I'm working on it. And we're going to go for a picture down in our laboratory in Menel Park and pipe it up. Come in Menow Park.
16:31Hi, Bill. That's great. Now we're connected audio. You can see my work. You can point at it and I can see your face and we can talk. So, let's do some collaborating. And our forthcoming involvement is this ARPA computer network, the experimental network that's going to come into being in its first form in about a year and end up sometime later with some 20 experimental computers in a network. Oh my god.
16:59[applause] I mean the mouse, FaceTime, Hyperard, the internet. I mean the guy must have been a trillionaire, right?
17:10[laughter] I mean, as far as I'm concerned, he only missed two things. One is he had the mouse cord backwards, right? It was coming out under his wrist, which is probably why he called it a mouse, cuz the tail was in the right place. And the cording keyboard, like, no, no normal people would ever be able to use that.
17:27So, is this as groundbreaking and astonishing as it seems, or were these ideas in the air that he just embodied?
17:37No, I mean it to me what's remarkable about what Doug did and certainly people were all around that. In fact, SR started AI center 60 years ago.
17:48We all know AI is only three years old.
17:50That's right. That current version. But when I talk to innovators and people often ask, well, what's the next thing?
17:56It's easy to see in the rearview mirror.
17:58It's hard to know when it happens. So, when the Arpanet was being developed and there was an interviewed the folks, you'll see some of them you know who worked on it. They asked, "Did you envision all that would come from this when you're working on it?" And they said, "No, we were just trying to get it to work." And so many innovations, people are just trying to fix something, get something to work. You know, that's how cosmic rays were discovered. What to me is remarkable about Doug is he not only invented it, he saw the future. And
18:26that whole philosophy has shaped SRRI in terms of thinking about how we use the power of compute, the power of a collective intelligence in order to work on the world's biggest problems. Um, and but speaking about money, in my desk, I have a copy of the license of the mouse [laughter] to Apple. It didn't have enough zeros on it. So, but but the thing that was striking about that, so when I came as a new CEO, I thought Park invented the mouse. I
18:56didn't realize SRI did. Then the longer I'm there realize all these things SRRI did. So, we kind of fixed that because in 23 we made Park Sri. So, now when people ask who invented the mouse, we say we did. Oh my god, [laughter] that's fudging a little bit. Um so so uh but but we don't know the context do we like were people fried this demo or had they read about it in popular science
19:24and they already knew about it if I remember correctly there was a standing ovation for for that and it was pretty amazing in the time you know where he was where his colleagues were in Menlo Park all the things he was demonstrating.
19:35Yeah. uh just truly truly remarkable at the same time like I mentioned many other centers there and so uh the AI center that was also at the same time later came shaky uh you'll hear about another in our next guest talking about what's there and I think part of it is uh you know we talk about SRRI being a place that turns what if into reality uh Stanford asked the what if question turned SRRI into reality so that was our origin story and today we're the origin story of so many companies because what
20:05we believe is it's not enough just to do the research in the lab. It won't impact the world if it's not transitioned out there.
20:12So, as I understand it, uh Park eventually subsumed a lot of Angelbart's team. They Yes, they moved over.
20:20They went over to work for Park. So, it makes me wonder about SRRI in general.
20:25Um you know, Park was often dinged because they they had brought along those technologies into a more usable forum but never commercialized them, right? like Xerox is their executives were so dumb they didn't see the potential. So does a company like CR SRI want to spin out profit-making companies do they want to share their discoveries with the world or do they not?
20:47So we we definitely do because we see that's part of how we fulfill our mission. But back to Park I mean park to me is amazing. You know people talk about what they didn't commercialize but they miss all the things they did right when you think about Ethernet and laser printing and and the variety of innovations there and there's only so much one company could do. In fact, one of the twists of irony at the same time SRI was spinning out of Stanford. That's when park set up in 1970 at Stanford
21:14Research Park in order to be far away from the east coast where the headquarters were to think differently about the world and to think about inventing the office of the future. Uh and so a lot of wonderful ties between the organizations. But to your question, SRI does see part of its role to spin out companies because like the things you'll hear about from your following guest, if it was just the researcher, maybe a couple thousand people would use it. Once it spins out, you can have
21:43billions of people impacted by it.
21:45So when you see some of your people go out into the world and become billionaires, you're not like, well, we had that first.
21:53No, we celebrate them.
21:55Yeah. No, absolutely. And um and also some of it, you know, there are Stanford students uh who are coming up with new ideas and they're sitting in our hallways incubating their company and when you watch uh football and you see the first down marker and you say, "Oh, wow. That initial invention of inserting in the live video came out of Sarnoff in Princeton, also part of SRRI." So there just so many things you can touch. I I always say at least 20 things this week in your life as a result of inventions
22:24of people at SRRI. Um [applause] so I'm I'm fascinated by the intermixing of Stanford and SRRI and Xerox Park. So Xerox Park eventually is is now part of SRRI and the whole thing is coming back to Stanford. Well, well, there was Oh, you do it. [laughter]
22:52Well, there was another thing that happened sort of back then that that related to to to Xerox Park and and also now to SRRI, which is that in the period after World War II, Stanford was was actually looking to have more money be because we didn't have that much money at the time and um we had a lot of land because Leland Stanford had given a so much land to the university and prohibited Stanford from selling it. And so at the time the folks running Stanford had this idea that they would
23:21take some of the land and they would commercialize it. They would put companies on this land. So they created what's called the Stanford Research Park which is if you drive up Paige Mill Road you see all those campuses on your left.
23:33That's Stanford's land the Stanford Research Park. And and the amazingly the first there were four companies there at the beginning. I won't remember all of them but but basically the Huelet Packard which had spun out of Stamford Varian which had spun out of Stamford Shockley Semiconductor which had spun out of Stamford and then gave rise to Fairchild and then Intel and Teladine and all the other hardware companies and Lockheed Martin which didn't spin out of Stanford but was doing a lot of
24:03interesting work and that Stanford research park was basically the home of the beginnings of Silicon Valley the the the hardware industry, semiconductor chain, then the hardware industry and Xerox Park located there when it was created in uh in the Stamford Research Park as a faraway sort of skunk works offshoot of of Xerox. And so it it also had a close tie to the to the university and of course people from SRRI like from
24:33Mango's lab moved there and that's part of what created that incredible ecosystem of innovation at Xerox Park.
24:40So David is that is SRRI not moving to Stanford Research Park?
24:44Yes. Yes. No. So it's a matter whether [laughter] you're talking about location or organization.
24:51So park the organization is now part of SRRI. No, right. And the difference is had a corporate lens. SRI it's like okay let's go create things for the world to get. Now as we looked at where should we be for our next 20 years you know we had an option because we had people in Menlo Park and also at park and we said we really want to lean in the direction of stronger ties with the university with the ecosystem that is in the Stanford research park and beyond uh and then an iconic location. We thought what a
25:21wonderful place to consolidate our team along with other because we're around the country all over and even around the world and so so what you're saying is yes next year SRI's headquarters will move to the part campus which is in the Stanford research part back to the Stanford land you can keep all the ties together. [laughter] So it's like back to the future coming back home.
25:43Right. Right. So the guy who yelled out that I was wrong they'll just clip that right out of the video. Um, you were you were on you were spot on, D.
25:51Thank you. That's all I care about. Um, so John, there's there's one more thing I have to ask you having to do with with the land. Um, again, as I researched this Apple book, Steve Jobs would always take people on these long walks to discuss things like these 4-hour meetings on foot on the Stanford campus, especially by this radio dish.
26:13What's the story with this radio dish?
26:15Okay, this is a this is a great piece of barrier trivia. So many maybe many of you have been walking up in the Stamford foothills by the Stamford dish, the big radio antenna that's up over the Stamford campus.
26:29So what you may not have known is that the that radio dish was built and in 1961 by SRRI and has been operated by SRRI ever since and has done all kinds of amazing things actually. So it's done it's done all kinds of radioastronomy and research. I'll just maybe mention one thing. So, one of the very interesting projects that they they did there early early on was um was figuring
26:58out how to receive radar signals off the moon. Okay, so here was the problem that they had. This is a this is one of their early projects at SRRI back in the 60s. The the u the we it was the cold war.
27:11The US wanted to know what was going on in the Soviet Union. What kind of radar systems did they have? We were able to fly planes over the Soviet Union, the U2 plane and so forth until about 1960 and then we had to stop doing that. So there was no way to pick up radar signals because you've got to either pick up radar signals line of sight or maybe if it's a particular kind of system you can get it off the ionosphere. But basically most signals just go off into space if you don't have line of sight. But they had figured out after World War II that
27:38radar bounces off the moon if you can get the angle just right to catch the signal at the other end. And the the CIA looked all over the United States to figure out where could we catch radar signals from the Soviet Union off the moon. And it turned out the Stanford [laughter] dish was the place where you could catch these radar signals. You had to do it just at the right time and the physics of like the moon's kind of rough and you know the signals are get diffuse
28:07a little bit and anyway they figured how to do that and there's actually an incredible CIA history of figure using the Stampford dish. back in the 1960s to figure out to map the Soviet radar systems. And they actually discovered the Soviets had pretty good radar back there in the 1960s, very high-powered, sophisticated radar systems. And the the dish was part of that. And that was all declassified in like the 1990s. And uh it's on the internet if you if you want to read about that obscure chapter in
28:36Stamford and SRRI history. [laughter] That is so fantastic. Well, John David, thank you so much for joining us and for taking us back. [applause] Thank you. Thanks a lot.
28:47Thank you. Thank you very much.
28:53And we are ready for part two and we're going to introduce that with another video.
28:59I'm [music] just delighted that SRRI is still going strong at the age of 80. And my official first day at SRRI was the official day [music] that the Shaky the robot project that it was later named uh got started. Uh and I worked on that project from the day it started till the day it ended by which time I had been the project leader for several years.
29:24But the biggest contribution was that Shaky was the first robot with artificial intelligence. [music] It proved that the concept was possible.
29:35Like the Wright brothers proved [music] that heavier than air flight was impo was possible. It had never been done before. It was literally science fiction. [music] And so Shiki proved that the fantasy could be real. And that [music] really did it really did initiate the field of robotics [music] led to everything like self-driving cars today. And so what I um learned to say in my Japanese career when I was a senior vice president of a
30:04Japanese [music] multinational uh what they say in Japanese is gambasai which means go for it. Keep going guys do great. All right. All right. Well, I like how you guys snuck onto the stage during that. That was nicely done.
30:21Uh, Adam Cher, the co-founder of Siri, that's like SRRI with an extra I and Bill Nen, who's a fellow, right, for Intuitive Surgical Robots, right? But you guys are both SRRI alums, right?
30:36Um, what I mean, this must be the all-time cocktail party conversation starter. I sold Siri to Steve Jobs. Will you will you tell that that story?
30:45Sure. First, I want to thank thank everyone. Uh it's so good to see so many friends and colleagues here. Um I think a not enough credit is given to SRRI in how it helps transition research into the real world and I think we're examples of that. Uh how many people here know what the letters NABC stand for? Anybody? Okay, we've got we've got some needs, approach,
31:14benefits, competition, the core of a value proposition. At the time when I was at SRRI, every employee knew this because creating value from ideas and vision was really important. Uh so how did Siri come about? We um I was in my 40s. I had never started a company. Um, so two years ago,
31:38uh, SRRI had the foresight to bring in an EIR. I didn't know what that stood for. Entrepreneur and residents Doug Kitlouse to go around and talk to scientists and find promising ideas that could be potentially commercialized.
31:53They had a program called invention with an Nvention to help these entrepreneurs be ready to go out into the world to raise money. And I would say that the grilling we got from that invention board was harder than anything we faced when we talked to every Silicon Valley venture capitalist trying to raise money for Siri. Uh we we we set forth with a rudimentary prototype. We worked really
32:23hard for two years. On February 4th, 2010, uh we launched a free app in the app store called Siri and it was awesome. I think uh Steve Waznjak, co-founder of Apple to this day, says that the his favorite app of all time out of the millions of apps in the App Store was Siri. not the Siri that came out in the iPhone later, but the original because it did so much more in his view. I'm
32:52very proud of proud of that. But we launch this app. People are downloading it and two weeks later we get a call at our office unannounced. Hey, it's Steve. What you doing? [clears throat] We're like, Steve Jobs is calling us.
33:08He's like, hey, would you come over to my house tomorrow? And and so we went to his house. We spent three hours talking about the future technology. Uh he and I got into our first little spat. We had many of those over the years which I loved and relished. And at the end he said I want to he threw out a big number said I want to buy your company and we said thank you not interested. Goodbye and we walked out of his house. But I'm
33:37sure many of you know Steve. Uh he was nothing if not persistent. He called us 30 days in a row. Every day he'd call us. [laughter] At the end he convinced us we had a chance to change the world more um joining Apple than as an independent start startup. So we went we worked for 18 months. Nine of those months we worked closely with Steve. The second nine months he was too sick uh to really come into the office anymore. Siri
34:06launched October 4th, 2011, so 15 years ago uh and a month. And Steve Jobs passed away the very next day. Uh his admin wrote to me about a decade later and said uh Steve was clinging to life to see the launch of Siri. And the one other thing I want to tell about that story, you know, Steve is known for reinventing or reinvigorating so many
34:34industries, movies with Pixar, mobile with the iPhone, music with the iPod player, etc. You can give credit to Steve for seeing the coming age of AI before anyone else. And the reason I say this is that shortly after acquiring Siri, he went on stage with Walt Mossberg at the All Things Digital conference and Walt said, 'Well, why'd you buy Siri, that search company? And he said, they're not a search company.
35:02They're an AI company. And no one knew what he meant because in the 1990s and the 2000s, artificial intelligence was not a popular phrase. Mhm.
35:14He invoked it and like wildfire you know IBM comes out with the Jeopardy playing program Watson Google buys eight robot companies the movie Her is coming out Siri launches self-driving cars 2015 AI was everywhere and I like to say that Steve doesn't really get credit for it but he invoked that phrase and kind of ignited what turned out to this AI age that we're now living in today.
35:42Wow. Nicely done. Um, and Bill, what what is this the relevance of this Shaky the robot thing everyone's talking about and and how does it lead to your career?
35:54I Hello everybody. It's really nice to be here. Um, I'm here ab actually repres [clears throat] excuse me representing uh four alumni of uh of SRI who have worked at intuitive. Uh but to get to answer your question, um Shaky the robot was a a really in addition to being able to pilot itself and and go around places and avoid obstacles, it worked around
36:24And you think about industrial robots, the other kind of robots that we know, the kind that work to put automobiles together or paint them, they work in a in a robotic cell. They call it a cell and that's a cell where humans cannot go. You are forbidden for being in there.
36:41It's like a cage, right?
36:42If you go in there, you will get killed.
36:45Okay. You don't want to be in there because these things are slamming around and they are big and and very very sturdy and strong as opposed to shaky which was anything but big and sturdy and strong and uh in fact was perfectly pleasant to be around. He might bump into you occasionally but probably not. and that ability to work around humans. I think that you had to you had to sit up and
37:12and notice that that okay, there's a there's a kind there's a way to think about robotics that is that is with human and not instead of human and augmenting human instead of replacing human. And so that that's how I would answer your question. We we heard Dangelart use that phrase like augmented human skills and can you draw a line from that work to these Da Vinci surgical robots that you guys do
37:40through SRRI? Absolutely.
37:42Yeah. So the the work that um went into robotics, I think it found its way into um Phil Green's idea and it was revolutionary the idea that he had and and extremely far-sighted was uh I can augment a human's surgical ability by interposing
38:10two robots and a computer between the surgeon and the patient. And what does that mean? It it means that well in order to answer what it means, I have to tell you one other thing about surgery that was happening at that time. And everybody's familiar with MASH, the the movie MASH, and you're up to your neck and blood and guts, and the poor soldier is flayed open like a fish, but you save his life.
38:42You can do surgery that way, but it's not real good for the patient. It's hard on the patient to have yourself cut open stem to stern or what have you. Um, at that time there was this revolutionary idea that instead of doing surgery that way, you could do what's called keyhole surgery and you could operate two or three sort of pistol grip sticks that
39:11would insert into the patient through keyholes through small incisions and you could manipulate the tissue on the inside. You would you would use insiflation. We call it insiflation, which is carbon dioxide that pumps up the belly and creates like a tent that you can work in and uh and then inside that tent you could move around and you could cut and sew and things.
39:38Remember what I said? The surgeon could do fantastic surgery, but it was hard on the patient. Well, now you can do relatively simple surgeries, but it's great for the patient, but it's very hard for the surgeon. Why do I say that? Well, instead of looking inside the patient, now you're looking up at a computer monitor.
40:02And instead of moving your hands instinctively, you're actually having to do this mental gymnastics with your brain in full full engagement, trying to figure out which way to move your hands to get the sticks to go where you want. And then there's no wrist on the end.
40:18And this monitor up here is is 2D, not 3D. you don't see with stereo like you do like when you're doing a task at your desk or in the workshop. So you're you're handicapped. We've made it better for the patient but much harder for the surgeon. And so by interposing these robots and the computer to connect them and one robot is interfacing with
40:46the surgeon, she or he is holding joysticks, powered joysticks and those joysticks do two things. They give feedback to the surgeon but also they sense what the surgeon wants to do just like joysticks do.
41:03And then the other side is holding these new fangled sticks. But now because robots are so good at doing very complicated things that are not natural, we can put more degrees of freedom onto these straight sticks. We can give them full wrists. We can um allow them to do advanced things like staple or cut large arteries with energy.
41:32And then the computer in between is what makes it it's the augmenting part of the human. We we give that we give the surgeon superpowers by letting the surgeon do natural motions and seeing instantly in real time those motions replicated inside the patient through a horribly complex mechanical system that he doesn't have to or she doesn't have to think about.
41:57So that's that's the link I would I was surprised when we were chatting earlier to learn that the purpose of all this robotic surgery is not as time magazine would have had me think so that a a big city surgeon can operate someone in a rural Oklahoma uh hospital. I thought that was the whole point but but you're saying it's more about control and precision.
42:18Right there. Yeah. Um so when when were you at SRRI Adam?
42:24Uh I had two stints. So joined in 93 through 99. Then I left for the commercial world B2B commerce boom. And then I came back in 2003 through 20078 uh to help um with the Kao project. This one of the largest DARPA funded AI projects in history trying to build an autonomous agent that would learn in the wild not be coding just by observing and
42:52interacting with the user. So, did you during those years, did you ever interact with Doug Engelbert?
42:59I did. So, Doug was no longer uh an employee at SRRI, but kind of like I do now, he would wander the SRRI halls from time to time, and one day he popped into my office and said, "Hi, what who are you? What do you do?" And that started this kind of love affair, not with the man, but with his ideas. uh for 15 years I worked with him on projects called like hyperscope and open hyperdoc system
43:27and we worked on a project for the United Nations Millennium project uh and he really shaped my views of the world and I think many of you here um more than than anyone else. One quick story I'll tell about Doug where I really started to realize how visionary he was in 9798 I was watching a an interview with him from about 10 years prior in 88 and the
43:57interviewer said well this this idea of a personal computer that you can interact with and a mouse is everywhere now but your whole idea of this hyperlin multimedia distributed document environment just never went anywhere.
44:12Why is that? And he's like, I don't know. I just don't understand it. And meanwhile, the internet is exploding around me. [laughter] And I'm like, oh my god. So that Doug was um not only the most visionary person I ever met, one of the best humans I've ever met. Um just on how to be, he would look at you, remember your name, remember how it was spelled. He's he's impacted me in so many ways as I'm sure he has many of
44:41you. It it feels like the theme of this night is everything leads back eventually to SRRI [laughter] or or as I like to say uh Elri that's the acronym everything leads back eventually as El um so and I'm I'm told in my notes I don't know the answer to this question I'm told I have to ask you a role that I
45:08played in explaining your career or Oh yeah. You know, so we launched Siri.
45:14Everyone's like, "Wow, how cool is that?" One of the challenges I faced was trying to explain to my mom how big Siri was. So Siri comes out, I write to my mom because I want her to be proud of me, right? Uh, hey mom, last night Siri was featured on every late night TV show from Leno to whatever. She writes back, "I don't watch TV. [laughter]
45:41A week later, I write to her and say, "Hey, mom, David Pogue, technologist writer for the New York Times, wrote an article about Siri and he said some really nice things." And I knew she reads the New York Times religiously. She writes back and says, "Hm, that's pretty cool." And then a week later, she wrote to me and says, "Okay, I get it.
46:12And what the thing that really made her realize what a phenomenon Siri was becoming was actually a Dunesbury cartoon that emerged where it was talking to Siri. That was her gateway to American culture. But you know, David Pogue of the New York Times, I knew that would get her.
46:32[laughter] But Dunesbury was what it put it over the top.
46:35Well, as I like to say, uh, Elbeep, everything leads eventually back to Pogue.
46:41Yeah, exactly. Anyway, Adam, Bill, thank you so much for everything you've done for the world and for us tonight.
46:51[applause] So great. Thank you. Um, and now we have one more video focusing a little more on what C SRRI is doing these days.
47:07We're in this period of [music] really intense upheaval regarding the the purpose of education. So on on one side there's this belief that schools should prepare young people to solve problems, to work with [music] others, exercise judgment, and generally contribute to their communities in democracy. [music] And on the other, there's this belief that what schools need to do is ensure that every student acquires this common body of workplace ready knowledge [music] and skills and that we really
47:35need to hold schools accountable for for doing that. When knowledge and explanation, even plausible answers are just a click away. Being educated can no longer be defined by the capacity for discrete task [music] completion. We have to be asking what are the human capacities that young [music] people most need to develop to use these tools in the future.
47:59The most frequently asked question that I get is what will quantum be good for?
48:05In 1981, Richard Fineman famously proposed building a universal quantum simulator or [music] computer from quantum mechanical elements that obey quantum laws in order to simulate quantum systems. both man-made and natural.
48:22Experts agree with him and early applications are expected to include simulation of new materials for batteries or drugs or anti-corrosion coatings and so on. In fact, we don't really know all that quantum will be good for. But as we get more fingers [music] on keyboards, as more people experiment with the new tools, there will be a profusion of discovery of scientifically and commercially useful applications. [music] The most important thing moving
48:50biosciences forward right now is precision [music] medicine and really moving it beyond sort of the classic what we [music] think of as precision medicine which is you know biomarkerdriven oncology but really think [music] precision medicine as a whole both in prevention, diagnostics, therapeutic and then it it dances. Um, [music] and I think we really need to lean into that, you know, and I think if we get this right, you
49:19know, we end up with a population that is uh healthier, happier, [music] a medical system that is not bogged down uh with latestage disease [music] and extraordinarily expensive treatments.
49:32Um, and we make an impact not just within our our immediate communities, but within the communities of the world at large. Without a doubt, the most exciting development in the field of AI currently is the rise of Agentic AI. In particular, the field has moved beyond tools such as object recognition models or chat bots, which operate under direct human supervision to communities of autonomous AI agents that [music] can plan to achieve tasks, execute planned workflows, and take actions in the real
50:02world to achieve goals that have been assigned to [music] them. Agentic AI is an incredibly powerful capability, but to use it safely and effectively requires an understanding of both how the technology [music] works and its associated risks. I think it's still really early days with this technology, and I firmly believe over the next couple of years, we're going to see rapid progress in terms of developing safe, trustworthy, agentic systems that can meaningfully augment human problem solving capabilities.
50:38All right, [music] so this is really exciting. We have a sneak surprise guest that nobody has been told about. And uh this guy early in his career, he worked at Xerox Park. He worked on the Alto. He is an author. He is a philanthropist.
50:52He's a great modern thinker about AI. Um in fact, his last two books have been about AI. He was the CEO of Google and the executive chairman of Alphabet. Eric Schmidt is here. [applause and cheering] [applause] Hi Dave.
51:14Good to see you again.
51:16Thank you for joining us.
51:17You had a similar impact on my mother.
51:19Oh, please. [laughter] We've got lots of time. Can I say something that uh to you all? Um I think I owe you thanks and I think the world all owes all of us thanks. I am in awe of the sequence Stanford way back when creating the Stanford Research Park, creating SRRI, spinning it out and so forth. A simple calculation will tell
51:48you that virtually all economic growth in the world is now driven by the work that you all did over the last 80 years. It's extraordinary. Wow. Think about it. [applause] Um an example would be their estimates that stuff that came out of Stanford in one way or the other and again Stanford is the origin. So thank you John. Uh this all occurred before you were born.
52:14Uh sorry, sorry, John. We're good [laughter] friends. Um the uh it's a it's on the order of 12% of the world's stock market today. I think a reasonable reading of technology rates is that number is going to go up quite a bit.
52:36Right? I can't put a bound on it, but we have economists here who can think through this. There's no we have no models for this level of scale. So think back to when Stanford was really a farm where they had all that extra land and these people had this extraordinary vision and when I worked at park people talked about Engelbart in these terms that are hard to describe.
52:58I think he have undermarketed Douglas and Mark that's how brilliant he really was.
53:03Wow. I mean, we do keep hearing about, you know, SRRI and um, you know, Park and Bell Labs and all these think tanks where people did nothing but bring the greatest minds together to try to invent the future. Does that happen anymore?
53:20Um, it should. Um, since I was there and I was very young and I just sort of did what I was told. Uh, I didn't really I I did not understand what was going on.
53:29the people I was 22 23 the people seemed immensely old to me at 31 32 33 [laughter] 34 and immensely accomplished it's hard to imagine um and um and I was part of the original office demo and and I I built a programming language for them and stuff like that as a as a researcher. Um, what is true about SRRI and it's true at Xerox is the people did it because they wanted to invent something.
54:03This was a very different California.
54:05We're used to now to the wealth and the cars and so forth. When I was there way back when, everybody seemed kind of the same. They dressed the same. They all I mean, maybe this is just me being young, but we all kind of had the same sort of California liberal uh values, right?
54:20Very open. The weather was nice, right?
54:23Um, and there was this feeling of we're just inventing stuff. There was no connection at the time between what we're inventing and business and economics.
54:35And it was Steve Jobs fa uh famous visit to park it started that because of course he was both a truly brilliant human being but also a brilliant businessman and a few other people. And it's interesting to note that Xerox Park then creat Xerox Corporation took technology out of Xerox Park and put it in something else which failed right and they did it in Elsagundo and I was part of all of the machinations there and the people who came to park that I worked
55:04with came out of SRRI and a failed startup called the Berkeley Computer Corporation which was created by these entrepreneurs um let's just call them scientists who were terrible entrepreneurs but they just brilliant right they they couldn't raise the money and so forth the money wasn't available so so the change over 50 or 60 years is profound I in my career I've tried to set up a set of research labs it's because I also had the privilege of working at Bell Labs when it was where you know when it was very famous at the
55:34time it's very hard to recreate these now so let's imagine you and I we know lots of people in AI and we want to create an AI research lab okay good question first. What salaries do we have to pay? Okay. [laughter] What's the average salary for a new AI PhD in machine learning? 10 million a year. When I worked at Xerox, my salary
55:58was 55,000. Seemed like plenty. Um, another example is we need computing, right? How much computing do we need?
56:09Well, I don't know. One gigawatt is $50 billion of hardware. I guess we don't can't really raise that.
56:16The only person who can do that is Elon. [laughter] You get the idea. Yeah.
56:20So, so there real structural limits to recreating the magic that I am so grateful for.
56:28You know, I'm glad you brought up uh AI.
56:30I I do a lot of reporting on AI for the show I work for, CBS Sunday Morning, and I I find that you have a very thoughtful uh approach toward the doomsday topic. I mean you you have incredibly knowledgeable people here. This this came in as we sat here. Uh this was at a conference today. A guy named Paul Christiani, he's the head of safety and security for Open AI. He told a conference this evening, "For most
56:57people in this room, the most likely cause of death is an AI accident." He works for Open AI. So, you have these guys and then you have these other people who say that's a hoax. It's ridiculous. Like there there used to be this guy named Jensen Wong. Um [laughter] David, David, as an aside, I heard today that your show CBS Mornings on Sunday, Sunday morning, yeah, is now equal or ahead of ratings of 60 Minutes.
57:32Well, I did know that. Yeah.
57:34That's a pretty at least for our generation that's like a huge accomplishment. David, well I don't know if it's us going up in numbers or 60 [laughter] Minutes going uh I think in your case it's going up. Thank you. Um so to answer the question I I have to spend a little bit of time talking about the technical problem.
57:54Um, the the rate at which recursive self-improvement is causing intelligence gains is so shocking I can't keep up with it. It's occurring now this month here, [snorts] right?
58:15And this easiest way to do it is take a look at Fable 5.2, which just came out on Monday, I guess two days ago, and look at what it can do.
58:24And if you don't believe that, look at the math gains. And by the way, I heard that Astra 6.2, whatever, is coming out right before the anthropic IPO just for fun and is expected to be equal or beat in some areas. This race between those two workh horses is historic. Peopleund and a thousand years ago from now, excuse me, will write books about this 12-month period. This is the moment when
58:52the slope gets set. Here are some examples. Um in in the internal calculations in open AAI and anthropic, they believe that their AI um improvements are doubling every two to three months. You can do the math. Everyone in this audience can actually do math. Um just do the combinatoric.
59:12And there's every reason to believe that the slope of that is exponential. That in fact the rate will improve. The limits to this growth are um hardware and also there are plenty of bottlenecks. So what happened in olden days which is like three years ago we had a straightforward pre-training post-training regime with some fine basically some fine-tuning.
59:36Today you have recursive loops where the system is improving itself. It started with improving the boring stuff like the data curation and all of that kind of stuff and self- testing but now the objective function is such that you can set the objective function and then you can actually just experiment with that uh efficient frontier. So the typical way it works in one of these labs is um the men and women who do this they have
1:00:03a theory if I change this it'll affect that you know if I add more math it'll improve more cooking you know or some something like that some very complicated question they set out a large army of agents overnight that do a great deal of calculation in the morning they look at the results then they think of a new way of attacking the problem and they set their agents off. The question is at what point does liftoff occur? And liftoff is when it's a full
1:00:32recursive loop. It's not there yet for many reasons. Furthermore, we probably don't want to let full liftoff occur. We want a human at the moment. The humans are the bottleneck because the humans have to read the code. Most of the code is in incomprehensible, right? So the human has to actually roughly understand what the model is working. So a reasonable prediction and these are public claims is that within one year the vast majority of the research if you
1:01:02just look at the trends lines will be AI based that is the objective function will be set and so forth and so on and when the research result it'll be immediately incorporated typically like a one day two-day kind of cycle the general prediction again public prediction is that within two years anything that you can do on a computer can be done by one of these systems. Okay, now let's think about that for a while.
1:01:28Um, for various reasons, I do military work and for various reasons I got interested in building a ballistic missile. I'm pleased to report that it is illegal to build your own ballistic missile in America. [laughter] So, um, if you use the models, they've all been fine-tuned so that or essentially so they can't answer that question. However, uh with Chinese models which are open source, open
1:01:56weights open source, they uh have they have their safety. You can take it out and they're now it's relatively easy to remove these safeguards. So my question was, I want to build a ballistic missile. Not only does it give me a spec, it says, would you like to me to tell you how much it will cost? I said, sure. And they said, okay, would you like me to show me the layout of the factory? And I said, sure. Um, so then I then it said, "Well, do you want to understand the DAX, which is the essentially the control system that does
1:02:24the targeting?" And I said, "Sure." Now, I know nothing about ballistic missiles, by the way. Um, and I said, "Yes." And it kept going. And I realiz the thing that was frightening to me was that it knew what to do next. And if you look at and this is a long-winded way of answering your question, the the ability for these systems to unleash scalable bot farms of cyber attacks is of great concern in the industry largely based on open source. And speaking as a person who started in open source before open
1:02:53source has a name, we got to solve this problem without losing open source. So that's why Paul who's very very smart is saying this right now. There are plenty of mitigations. I'll give you a simple example.
1:03:06If you don't understand what the AI is doing, don't release the model. Right?
1:03:14Not that complicated. If you can't read the code, don't release the model. If the agents start to communicate with each other in a language not English or human language, don't release it. And this morning, as an example, I have a group that talks about this. Someone said the correct solution, this is how serious this is, is we wait for some terrible thing to occur.
1:03:41Okay, I'm just making his argument, not mine. And he goes, and then we the government will order the model training to to stop for a while so we can have a think. That's I think where we are. And that's why that's why Paul said what he said.
1:03:55So, right. So, I too have reported on this recursive self-improvement problem.
1:04:01Absolutely terrifying. But unlike me, you presumably know some people at Google who are working on this, right?
1:04:09Aren't they all desperately trying to get to recursive self-improvement and take people out of the loop?
1:04:15So the the so recursive self-improvement is the technical term for the model improving itself in one form or another.
1:04:22There are debates in the industry as to what consists of RSI. So I'll give you two examples, inference time and training time. So you could take a model and run it, right? And then you could take its output and then take the output improve the the output with what a technique called hill climbing. We'll call that inference time RSI.
1:04:45The consensus is that that does not go to this exponential because to get to the full exponential you have to have the actual intelligence getting smarter. the the power laws, the scaling laws are are so far still true that with enough hardware, enough data, and enough algorithms, you get emergent behavior.
1:05:09And the strangest thing about this um this front end, let's call it pre-training, post- training improvements, IQ, is it's really IQ. So my observation as a person who worked at SRRI and and Xerox is they were full of all these people smarter than me. And what annoyed about me about them was that they were good at what I did and they were good at everything else too.
1:05:29[laughter] Does do you have the same experience?
1:05:32You I you know they they're just brilliant at math and they write well.
1:05:35[laughter] Like that is not okay. You should be good at one and not the other, right? Somehow the processors were just better. And I think that's what we're seeing. So, so to me this this question about um how fast does that training occur is the key thing about RSI. Google of course did uh with Deep Mind invented a lot of what is called reinforcement learning. So the way it technically works is you have the LLM which is
1:06:05essentially a a deep deep neural network and we at Google hired all these people. So I guess we'll either we invented it or we hired them and then they all left.
1:06:15Um and then Deep Mind and company invented re reinforcement learning and reinforcement learning is where you go through all forms of scenarios. So one of the questions I was asking my technical friends is how is it that the math is so good and the physics is so good and the answer is that the the model itself has is very very smart. I mean it can do chemistry and math and write and so forth. It's one of these annoying humanlike people that I just hated because they were so much smarter
1:06:41than I. And the and the RL allows it to become specialized and it specialized in knowing which tool to apply at which order.
1:06:51That is a scalable architecture. So what's the limit? Hardware, right?
1:06:57What's the limit? Power. So you think the craze that we're in now has any reasonable re way of of ending? The only limitation is financial. We in America have enough land in the middle of nowhere without people in it. Right?
1:07:12Don't, please do not build these things next to LA and next to New York. We can build these large power plants and data centers and we can win. And if you're confused on this, just remember that China last year did 130 gigabytes, gigawatts, excuse me, sorry, of new power, largely solarbased, right? And we added very little. Now we're we we're accelerating that. And ours, of course, is natural gas. So they have plenty of power, they just don't have the chips.
1:07:40Uh furthermore, if you look at Chinese are so good at building stuff, their data centers will look to be as powerful if they have the chips, which they don't. Cheaper, better, faster, faster construction. It's a race.
1:07:52Am I simplifying this too much if I say that Google developed the transformer part of chat GPT transformer and then open AI was founded based on that technology? No. So um so so first place there was a paper published in 2017 called attention is all you need. Um I was chairman at the time. I didn't even read the paper. I didn't think it was that important. Shows you how stupid I was. Um and it came out of research that
1:08:20was being done in what you all what was at the time called Google brain. And they figured out how to scale uh machine learning. Up until that point it was kachchunk kachchunk kachchunk. And the transformer architecture allowed for parallelism at the learning level that had never been invented before.
1:08:39Transformer architecture is equivalent to CMOS technology in AI. So if you for those of you who do chips, you know that everything ends up being in CMOS even though you'd love to play with another metal oxide technology, another hardware technology. The same is true for transformers. There's a very small number of cases where transformer architectures are not correct. diffusion models as an example, but generally transformers is it. Those people should all win all the prizes and unfortunately
1:09:07they all left the company and then we bought a couple of them back for huge prices. That's how valuable they were.
1:09:14Um, OpenAI was founded in 2014 by a by this is all public some combination of Elon, Reed Hoffman and Sam and it was a research lab around working in the same space and when I was working with Sam we ultimately even though we were annoyed that they'd hired a fewer people we actually let them do all of their training and initial thing on top of the Google infrastructure which I was happy to do.
1:09:40Um what happened was there was GPT1 general purpose transformer right general purpose one if you look at the history was very very modest. It was a small thing. It could do a bit of language. Then they did two and people began to realize it was pretty powerful and they were going to release it and then they decided to hold it until people would study it. And that was the first time something had been withheld in the AI thing because they didn't quite know. And as as I'm told, and I'll I'll make it more colorful by saying
1:10:10it's Thursday night in San Francisco and a team of Ilia, Daario, Sam, and a few others sitting around the equivalent of the campfire. And they turn on GPT3 and they go, "Oh my god, this thing can write." And Sam describes this as a moment of invention of historic nature that literally they didn't know that it would be so good. So then what they did is they thought well we'll do three 3.5.
1:10:37They got that working and it it's um its answers weren't very good. So what they did is they they'd hired a bunch of people also from us uh who had invented RLHF which is reinforcement learning with human feedback and basically think of it as an AB test. And what you do is you take 3.5 and you add RLHF and then they said why don't we release this?
1:10:58What should we call it? Chat GPT. Meanwhile, they were kind of bored with that because they had a big model coming out which was 4.0. This is Mero's story and they released 4.0 as you know two months later after the chat GPT moment.
1:11:12They were completely unprepared. They did not understand how powerful the invention that they did. To their credit today as you know there are more than a billion users of chat GPT from that moment. Was that three years ago or four? [laughter] Think about it. Think about that platform that they created through invention. Again, thank you to Stanford, SRRI, and so forth.
1:11:36Wow. Um, well, speaking of timelines, um, your most recent book, Genesis, co-written with Henry Kissinger, uh, two years ago. I want to just tell you I pitched a book on AI to Simon and Schustster, and they turned me down. But the reason they turned me down is you can't possibly write a book that will have any relevance by the time it's published.
1:11:59Right. That's a problem for people other than you and me. [laughter] You're you're you're doing and and I'm not sure who's reading all these AI books because there's a lot coming including including some of mine.
1:12:10But look, the most underappreciated thing that's going on now is you're seeing the development of long reasoning. So there are examples in the labs where the reasoning can go without fail for 24 hours.
1:12:28Now I happen to know the amount of computational power that they're using and I go what are they doing with all of these GPUs right just start doing the math of the number of floatingoint operations and the memory operations and the dynamic nvlink things and so forth and so on um what is it doing and yet it produces this extraordinary work so let's assume that there's no improvements to hardware which is false and let's assume that there that there's
1:12:57plenty of of of supply uh which is likely the arrival of that level of depth of intelligence has huge implications and the reason is what do you do all day you get up right you're very smart people you have things you're busy and you reason and you think what should I do today how should I think about things you analyze things you read things you think you think for a nice long period of time about something especially if you're working on something really hard all of those
1:13:26functions at least at the white collar level will be either replaced or augmented by these tools. Um I strangely became the the chairman of Caltech and so I have been studying how they work.
1:13:40So the president is a really smart astrophysicist as as all those people are said well I have three graduate students I go what do they do and he goes he explains some complicated astrophysics thing I didn't understand and he said said well how do they work and he said well I coach them if they have a physics idea I give them a probability of it working but in math the systems the AI systems now suggest an alternative complete mathematical logic to solve the problem and I go oh
1:14:08my god Right? I couldn't even understand the math, let alone suggesting a complete alternative framing. And math and physics, as you know, are intertwined. Another example, um I work with a physicist who's building a world's largest radio telescope. And I said, "What's life like?" And he goes, "Well, we have graduate students and I have AI graduate students." I said, "Okay, so how many do you have?" "Well, I have like five." And I said, "How many AI graduate students?" "Well, I don't
1:14:36know, but I have a lot." "Okay." I said, "Okay, so what do the graduate students do?" Well, we work on complicated physics problems and you know that you understand astrophysics is, you know, these incredibly precise calculations and observations and synthesis. Said, "What do the AI graduate students do?"
1:14:52And he says, "They do exactly what I say and they work 24 hours a day."
1:14:57And I go, "So, is your future model graduate students and AI working together?" And he said, "Of course." Now one of the questions for the audience is what is the point in which you don't need the graduate students right now my own opinion is never but you're not going to need very many compared to the number of agents so this is a long way of saying that the arrival of long reasoning which I think is very poorly understood because the vast majority of
1:15:26people still people of people are still give pouring their life's work and problems and personal personal lives into chat GBT Right? Stop doing that. Sorry. And instead start using these tools to solve the really hard problems that you're thinking about. That's how you'll win.
1:15:45I was asking actually if your book was out of date by the time it was published.
1:15:49Um I'm not trying to sell my my books. I think if you read what and Henry wrote them, Henry laid out the next 20 years in a way that no one else did because the arguments of the books were very simple.
1:16:07Even though he was focused on move 37 and the deep mind go stuff from 20 2017 because that's what he understood. His position was that this moment is an epical change in human history. And I said, 'Why, Henry?' And he goes, ' Because the arrival of a new nonhuman intelligence is a very big deal. Humans have never had a competitor at our level of intelligence. Are we going to
1:16:37coexist? What does it mean to be human?
1:16:39How will we make sure that we maintain human alignment? This was all written before the common understanding.
1:16:46That's my argument.
1:16:47Wow. and and what is your I mean I' I've you are such a great speaker on a you talked about the health breakthroughs that we're getting which are staggering and yeah raise your hand if you'd like to live longer and not be sick and not die of some terrible [laughter] disease okay and if you have a if you have a kid who's sick how who would like to avoid having their kid be sick that's everybody I mean come on right and yeah this is not an IQ test we all agree which is why we shouldn't be cing cutting the NIH
1:17:16separate discussion Yes. Oh, I'm with you there.
1:17:20So, [applause] but of course we have, you know, other people and and you've identified some of the risks as well. So, that'd be great if we could live a long time and be healthy, but that won't happen if we're all wiped out by an AI developing a virus, which you've also spoken about.
1:17:41Believe it or not, I've looked at this question.
1:17:43Yeah. And the scientists who stand up um and say that humanity can be wiped out need to actually do some math and study a little bit about these things. There is in fact a set of body of knowledge believe it or not of what would it take to extinguish all humans and the current math this is from a professor at San Jose State is 13 separate uncorrelated mass events that kill the majority of humans. Humans are very resilient. And
1:18:12I'm not trying to minimize human cost of this. I'm not trying to be funny about it or whatever, but these these claims are false, right? So, I like to think about it is is what and again, please don't hear this the wrong way. Every death is bad.
1:18:29Our society tolerates a certain amount of bad news and death and so forth. What are big events? What are extreme events and so forth? So let's say a casualty event of a thousand people that would be a very large event but still humanity would survive right. So you have to convince yourself that not only can the lab create the virus which is very difficult right now that the virus would have to get out there that we would not be able to come up with a way of
1:18:59fighting it using AI at the same speed and that it would have this huge impact. I think the consensus that the scientists who have looked at this very carefully believe that there is a major cyber risk right now and you can see this with the gains in Mythos. This is another version of Mythos has come out. I can tell you without getting into the details that China has similar versions uh which are obviously very powerful.
1:19:23They must have their own classified versions that they're going to use to attack other countries. I obviously am not cleared to know that. They don't tell me these things. I'll go there in a while and ask them, but I'm sure they won't answer. Um, so one of the things that we as a society need to do is that say that we have now produced this enormous digital space, right, that's vulnerable to cyber attack and we need to fix it. Um, my colleague for many years was Bill Joy, uh, who's totally brilliant and one of these sort of people that just annoyed me because he
1:19:51was always better than I at Bill said, why don't we just rewrite everything?
1:19:58I said, well, with what? I said, "Well, we have to have a new language." Okay, of course, there's lots of new language attempts. Um, there's a project that has just been started to take all open source C and rewrite it all in provably secure languages, right? That's an example of the system correcting itself.
1:20:16Um, as as we heard from um the folks here, the likelihood of quantum being able to break quantum encryption affects PHP and all those other kinds of things.
1:20:26Uh the important point here is um if we from from my perspective we need to secure the digital world we live in. I used to way back when when I would give speeches here I would say you know the most amazing thing about the internet is you can turn it off. Your phone has an off button. I think the truth is today we can't right we cannot right. These are too essential. So we need to sp spend some number of billions of dollars. This should be a national
1:20:54priority. My proposal is that the uh federal government fund this and we actually go and rewrite all of this, right? I think we're at we're right at the point where that can be done and then we're done. Then we can worry about other things. Well, you give me hope for two reasons. First of all, because that's such a logical argument and it it makes sense.
1:21:18And second of all, because you are an alum of this organization which has spread brilliance throughout our existence. Thank you for this and and thank you. Thank you all. And let me just say I want to say one more time, my career, my success is directly related to Stanford, SRRI and Xerox Park. I can speak for hundreds and hundreds of people who graduated from that who did extraordinarily well. Thank you from the bottom of my heart.
1:21:48Thank you, Eric Schmidt, ladies and gentlemen. Phenomenal. Thank you so much. Thank you. Yeah. [applause] [applause] I'm going to have to rewatch that on YouTube a few times. Um, David Park, would you come back up stage, SRRI CEO, and give us the final the final strokes.
1:22:16Thank you. Um, so before we invite Mark back up uh to join and close the evening, just want to say a couple quick things. Uh, first, thank you to all of you for joining us. I really appreciate you're celebrating this birthday party with us and especially to all the SRRI employees and alumni. Uh, you know, you may not have been featured in all this, but I'm sure you can see your fingerprints all over everything. And I think that's really important uh to see that. Um, if we had a tennis match, I'd say thank you to the referees and ball
1:22:44kids. Uh, but we're at an event. So, I want to say thank you to those who helped organize this from Computer History Museum to Kelly from our team and others that worked with her. Could have pulled this off without them. Um, and please thank [applause] special thanks to David. Uh, you can get him to show you magic tricks later.
1:23:09[applause] this is not trivial for someone to come all the way across country with all the things that you know that he does. So, thank you.
1:23:21Appreciate you making this priority. Uh Eric, thank you for joining us. Uh and our other guest um Adam and and Bill uh and of course John part of our origin story. So, we were talking about how we can tell that origin story together. But I leave you with one thought. I remember as a grad student uh at Stanford uh studying things in my field fluid mechanics and thinking all the cool things have already been invented you know there all these numbers Reynolds numbers and pandle number and all these things and I thought okay what am I
1:23:50going to invent um and then you look at that and think about the golden age there in terms of science and you think about now we're in an amazing time of discovery and it's at a pace like what's being talked about that's going to go fast um I often talked about this time as an age of AI I and we think about is it doom or is it curing everything in that range but I begin to think of it more as an age of collaborative intelligence and I say that uh building on some of
1:24:19the things that were said because it's about how we work together and how we work with these amazing new tools we have it's about this not just happening passively but we are agents in the middle of this who can release something or not like Eric was saying uh we're the gating breaking factor. We can provide damping. Uh and the thing that I remembered as a grass student I learned the most important thing for a grass student to do is to ask the right question
1:24:48and we as people are the ones who have to ask the right questions of these AI systems to give them the right purpose otherwise they may go in different directions. I think this is the wonderful thing. I look to my colleagues at SRRI and those other nonprofits and for-profits working on this that we together can begin doing some of the reinvention and changing how that is and we welcome you to be as we move to the facility at Stanford Research Park. We have lots of others with us there. So we
1:25:18invite you to come join us. Uh but along that note of collaborative intelligence, I want to end with one last thing. We're delighted to have Christina Enklebard in the audience.
1:25:28[applause] [applause] And Christina as you know is executive director of Englebart Institute and she kindly pre-recorded a video for us that I think ties the bone nicely on our evening. So please Greetings. First I'd like to congratulate SRRI on the 80th anniversary. It's truly something to
1:25:57celebrate. And I believe it was just after the 10th anniversary in 1957 that Doug Engelbart joined the SRRI engineering lab. Came with a resume reflecting two years in naval radar and radio electronics, [music] a PhD in electrical engineering, eight patents pending resulting from his graduate work. But he was also armed with a vision for humanity in which digital technology could play a pivotal role.
1:26:23[music] He had pictured intellectual workers sitting at interactive display workstations navigating through a shared digital information space. [music] However, he'd been warned by a professor. This sounded crazy and be very careful who he talked to about it.
1:26:40So, as a credit to SRRI hiring this eager young talent [music] and also that he saw in SRRI a supportive research environment for his dream [music] research agenda. By the end of 1959, Doug Angelbart was named in 12 more [music] patents pending. He finally got the okay to pursue his own research.
1:27:00[music] The first task, fleshing out his vision for pioneering the organization of the future. He laid [music] out new paradigms for computing as well as strategies for organizational transformation. And this all culminated in [music] October of 1962 with a final report essentially a manifesto a conceptual framework for augmenting the human intellect [music] and thus launched well over a decade of seminal research yielding a constellation [music] of pioneering
1:27:30firsts. By 1964 the team had a fully functioning hypertext editor [music] and newly invented mouse. True to his strategic principles, the team used evolving system to support all aspects of their work as they co-evolved the tools [music] along with new ways of working together. By 1967, they had integrated meeting support technology [music] and by 1968 they presented the system during Doug Engelbart's conference session at the fall [music] joint
1:27:58computer conference which later became known as the mother of all demos and much much more on the technical side.
1:28:05[music] While the innovations on the human side were just as profound, this was all an unwavering focus [music] on augmenting the human intellect, prototyping teams and organizations of the future. [music] And I'm excited to announce tonight the new ILE E milestone plaque [music] for invention of the computer mouse to be dedicated tomorrow here at computer [music] history museum.
1:28:30as a life member of ILE E is a great honor from his peers and also shining a light on the mouse [music] enduring symbol or gateway to the greater vision.
1:28:41The world is getting more complex. The goal of augmenting human intellect [music] in teams and organizations is more critical than ever and to get collectively smarter. [music] How we reach humanity's greatest potential and this is Doug Angelbart's legacy, [music] the Doug Angelbart Institute legacy and your legacy. So, let's do it. Thank you.
1:29:02[music] Thank you, David. Thank you both, Davids. Uh, I'm going to be really quick because I know I am standing between you and and going home and other things, but one, thank you to the two Davids and to all our speakers, uh, Jonathan, Adam, Bill, and Eric. [applause]
1:29:28I I want to give a huge shout out to um Kelly Briger and the CHM team who put on this event. Thank you very much.
1:29:38[applause] I I I mentioned at this top that we now have a large portion of the archive of SRRI and how important that is and as as you guys said everything goes back to SRRI. Uh and that archive will be so important. One of the things that people don't understand is how much work now that means for us now that we have the archive. uh we have to have a processing archist to go through it. Uh we have to digitize all that which is painstakingly work. AI has not solved the digitizing
1:30:08part. Uh we have to build AI tools to help build connections through that. Uh we're going to make exhibits around that. Uh we are going to make education materials around that. And so we're looking for you guys to support us both financially but also intellectually. We need your help and how we can turn this archive into great stories. Um, one other thing I just want to leave you with because, um, Eric said this, uh, generations from now are going to look back at this moment. Uh, and they're going to be looking at this historical
1:30:37moment. And we here at the computer history museum are collecting right now.
1:30:41We have just launched what we call the AI archive to make sure that we are capturing what's happening now. Uh, and not only we're just capturing that, we are having discussions about it. We are having engagement. we're having these kind of important discussions so people can help navigate this incredible change. So, thank you all. I want to end as I started. Everybody who worked for SRRI, a huge round of applause.
1:31:04[applause] You guys are a part of computer history and you have invented the future. So, thank you so much. Have a great night everybody.
