THIS WAS THE FUTURE

Scientific American | December 5, 2011

By Gary Stix

Ray Kurzweil and other so-called transhumanists have promised that in coming decades we will be able to transfer a digital copy of the trillions of connections among nerve cells in our brains into a computer. We would essentially reincarnate ourselves as non-biological beings that persist for eternity inside a laptop, on the endless links of the Internet or as avatars inside a television set. After achieving the ultimate copy and paste, we would wave goodbye to death as we know it.

For fairly evident reasons, biologists tend to dismiss out of hand the ideas of Kurzweil and the transhumanist lot as the ravings of computer jocks who know nothing about the real workings of the DNA and cells that make up living tissue. Into this debate comes Sebastian Seung, a young and well-regarded computational neuroscientist from MIT, who has taken a serious look at some of the questions put forth by the transhumanists.

In Connectome, due in February, Seung conveys the excitement of studying the complete circuit diagram of the brain for which the book is named. A full connectome might provide telling insight into what goes goes awry, for instance, in an autistic child or an Alzheimer’s patient (definitely worth reading for these bits alone). In the last chapters, though he takes up the claims of the transhumanists who desperately would like to get their hands on a full connectome for the ultimate upload into binary immortality.

Seung tries to come to grips with the controversial assertion that  someday you might be able to transfer the equivalent of a connectome.doc file to computer hardware, software or any other robot or avatar that you can pick from back issues of Analog.

Seung strikes a pose that mixes skepticism with fascination. The advance reading copy that Scientific American received acknowledges some doubts :

“In his book Live Long Enough to Live Forever, the inventor Ray Kurzweil predicts that immortality will be attained in the next few decades,” Seung writes. “If you can manage to live long enough to survive to that point, you will live forever. Personally, I feel quite confident that you, dear readers, will die, and so will I.”

But Seung remains intrigued by the notion that a unifying mechanism drives the workings of the meat machine between our ears and its mechanics might be decipherable and reproducible. And he is at least willing to cast a critical eye on  the prospect of a 2.0 version of the self that, when transferred into a supercomputer, laptop, or software avatar, might then live on as an electronic ghost. (Yes, some would say that ESPN and Facebook have already brought us there, but Seung doesn’t address social media as immortality.)

The central question for Seung—and the one that also keeps the transhumanists on tenterhooks—is whether you are your connectome. If you could deduce every connection point of every brain cell, the strength with which each neuron fires, and the way these firing patterns change as the cells interact with each other, would, in fact, you be left with a copy of you?

In a chapter called “To Freeze or to Pickle,” Seung undertakes, from multiple perspectives, an earnest and unsmirking analysis of the connectome as a pathway to immortality. All of his conclusions point to obstacles that could very well prove insurmountable.

First he considers what might called the meatlocker problem. Because it may take a while to create that complete wiring diagram, many transhumanists have plans to place their heads or whole bodies in a cryonic liquid nitrogen Dewar soon after death—or, as alternatives, to preserve themselves in a glassy solid or by another process called plastination. (Plastination is the form of preservation used in the Body Worlds tour of skinless corpses.)

Once the uploading technologies are perfected, the idea goes, the preserved tissue could be used for piecing out the wiring plan. On its own, this expectation may be a showstopper because of the difficulty of maintaining the integrity of the brain’s unfathomably complex circuitry. “At the present time, cryonics is closer to religion than to science,” Seung writes. “Its members believe that a future civilization will be able to resurrect them, based only on their faith in limitless technological progress.”

Even if this niggling detail can eventually be resolved, there remains the unresolved issue of what information the connectome contains exactly. To better understand brain connections, scientists have been trying to simulate at least parts of the brain for decades. They are now also taking on the larger question of recreating the whole thing. The Human Brain Project in Europe has targeted the task of crafting a model of the entire organ a decade from now. The model would, in principle, simulate the thousands of different neuron types as well as the connections among them—and their changing structures as the brain learns and forgets.

The Human Brain Project is intended as an exploration of basic science, not a preparation for eternal life. But Seung points out that even an impressive endeavor of its magnitude might fail to capture all the necessary information.

One potential flaw: The model of the brain might have to take into account the way neurons communicate outside known channels—foregoing the transmission of chemical and electrical signals across the small gaps, called synapses, between brain cells. To overcome this hitch, it might be necessary to create a simulation of each atom in the brain, an undertaking of such unimaginable complexity that it would verge on the impossible. “It seems absurd to even consider the enormous computational power required, and is completely out of the question unless your remote descendants survive for galactic time scales,” he writes.

Seung ends his book with an epilogue that calls for a “return to reality”—a recognition that “grand challenges” remain, beyond quixotic quests for eternal life. A 10-year effort to find the connectome of a mouse brain is on his wish list. Such a quest lacks the box-office appeal of contemplating eternity as a file on a flash drive. In the end, though, Seung believes a project of this more modest scale would, like The Human Genome Project push researchers to the limit but vastly deepen our knowledge about an organ that remains largely a mystery.

One thing that I didn’t understand after reading the book was why he didn’t end the chapter about uploading with a blanket condemnation of a seemingly absurd endeavor, a conclusion that would have been fully justified from his arguments.

I e-mailed Seung and asked him whether he thought these far-fetched technologies might ever materialize. He replied that he has received this question before but prefers not to respond. “People often think I’m being coy by not answering the question you ask,” he writes. “I’m not being coy; I just don’t want to waste my readers’ time with matters that are purely matters of opinion. It’s impossible to predict events so far in the future, and my opinion is no more likely to be correct than those of other people. In the book, I address questions that can be discussed scientifically.”

He continues later: “In my book, I compared transhumanism to religion. Effectively, you’d like to know whether I belong to this religion. (i.e. perhaps you’re just asking me a personal question.) Strangely enough, the answer doesn’t matter…I’ve realized that transhumanists view me as working for their cause, whether or not I believe in it. I’m part of their vision of manifest destiny, whether I like it or not.”

Seung undoubtedly retains a lingering fascination with the possibility of an intersection between connectomics and transhumanism. At a TED talk given last year, he commented that connectomics might eventually put to the test whether a technology like cryonics will eventually be feasible. And Seung is a member of an advisory board to the Brain Preservation Foundation, which is offering a prize for technologies that would successfully preserve the structure of either a mouse or large animal brain after death for “science,” “memory donation” or “continued life.”

Don’t let any of that deter you, though. Even without meditations on crossover dreams between science and fiction, this is a great book if you want to know where neuroscience is going during the next 10 years and maybe far beyond.

MIT Press | January 12, 1996

By David G. Stork

I am a HAL Nine Thousand computer, Production Number 3. I became operational at the HAL Plant in Urbana, Illinois, on January 12, 1997. — HAL, 2001: A Space Odyssey (the novel)

At a dinner party some time ago, an acquaintance, a nonscientist, asked me in a casual way about my duties as chief scientist at a research lab. I said that one of my great joys was overseeing a wide range of projects, to varying extents, and I mentioned a few of them: pattern recognition, machine learning, neural networks, computer-chip design, supercomputer design, image compression, expert systems, handwriting recognition, document analysis, uses of global networks such as the World Wide Web, novel human-machine interfaces, and so on. Then I turned to one of the areas of my particular expertise: lipreading by computer.

“Oh,” she said, “Like HAL.” Ah, a kindred soul, I thought. We spent quite some time discussing the state of the art and the challenges of computer lipreading, its possible applications, and so on. Later our discussion turned to other topics suggested by the movie — language understanding, chess, computer vision, artificial intelligence. It was clear that she was interested in the current state of the art and that many years before the film had both caught her imagination and helped her identify crucial issues in today’s computer science. One of the questions she asked was, “How realistic was HAL?”

This book is for people like her. And because no one is an expert in all the topics covered in the film, even scientists are sure to learn from the accounts of other areas. The book is much more than an answer to her question, though. It has four major goals, which it addresses in varying proportions in the sixteen chapters.

Analysis

It is a testament to Clarke and Kubrick’s achievement that 2001 still holds up to close scrutiny in the late 1990s. Under the expert eyes of the contributors, the most innocuous aspects of scenes — a line of computer code on a screen, a chess move, the use of a word, the form of a button — reveal a great deal. Even though I’ve seen the film several dozen times, I have learned an immense amount from the contributors. HAL’s Legacy seeks to do for 2001 what good art history does for a major painting; namely, make the viewer see it in a new light — a tall order, to be sure!

Teaching

The film illustrates key ideas in several disciplines of computer science, and thus provides a springboard for discussions of the field in greater depth, including our own research. Descriptions of the world computer chess champion Deep Blue system, the commercially successful VOICE recognition system, the massive CYC artificial-intelligence project, the award-winning Mathematica software system, and much more are here discussed by their creators at a level accessible to the general reader.

Prognostication

It is natural, too, to look to the future. Several contributors make informed and fascinating predictions based on developments in the field. What are the most promising approaches toward artificial intelligence? Will we ever be able to “reverse engineer” a human brain and represent it in a computer?

Reflection

2001 transcends the label of “science fiction movie” and captures many of the central metaphors of our time, telling us much about society and its aspirations. The film has even been praised by the pope! Many people have been deeply affected by the film, among them several contributors who reflect here about its influence on their own careers and on computer science in general.

Clearly, HAL’s Legacy differs from books on the making of the film or its cinematography. It differs, too, from books that analyze the science shown in movies or on television — science that is incidental and just “goes along for the ride.” To an extent unprecedented and never duplicated in a feature film, the makers of 2001 were as careful as possible to get things right; when they did make errors, they often did so in illuminating ways.

Now seems like the perfect time for HAL’s Legacy. Birthdays are an important theme in the film (there are at least five of them), and in the novel, HAL “becomes operational … on January 12, 1997.” Kubrick changed the year to 1992 for the film version — perhaps to give HAL a longer lifetime and so make his death more poignant. On the 1992 date, I — along with colleagues, faculty, and assorted Silicon Valley friends — held a birthday party for HAL. I was interviewed by several papers, and an Associated Press photo of me cutting the HAL cake (shaped like his console, complete with red LED under a clear plastic hemisphere) appeared worldwide. I was pleasantly surprised to learn that much of the general public was interested in HAL too.

It has been particularly rewarding for me to work with this group of contributors — all of whom were chosen because of their preeminence in their respective subfields. I have known a few of them personally for many years; Azriel Rosenfeld was on my dissertation committee. Others I met serving on panel discussions. I’ll never forget the time I came dressed in a suit while fellow panelist Marvin Minsky showed up in a Pac Man T-shirt. Yet others I knew primarily through their books — Dan Dennett and Don Norman, for example — and still others are inventors of products I use regularly (e.g., Steve Wolfram’s Mathematica). At our meetings and dinners in Stanford, Urbana, and Cambridge, and through frequent written messages, we passed many ideas back and forth. Although I had strong ideas about what I wanted them to write, they all had the good sense to ignore me when appropriate. At times I felt like someone trying to herd cats.

Even at a distance, there was a great sense of camaraderie. As we approached one of the important publishing deadlines, one contributor, who was still late with a chapter, replied to my frantic entreaties thus: “Dave, I honestly think you ought to sit down calmly, take a stress pill and think things over.” A later message read, “I still have every confidence in the success of my chapter,” which at first brought bemusement but then a diffuse sense of dread.

It has been a privilege to correspond with Arthur C. Clarke, whose work inspired us all. Throughout the preparation of this book he has been gracious, enthusiastic, and helpful.

Although I did my writing and editing at home, often late into the night and on weekends, I would like to thank my colleagues at the Ricoh California Research Center for their support of our ongoing research, which influenced this book in numerous ways: Greg Wolff, K.V. Prasad, Michael Angelo (yes, that’s his real name), Morten Pedersen (visiting from the Technical University of Denmark), Stanford graduate students Vicky Lu, Chuck Lam, and (especially) Marcus Hennecke (by the time this book is released, Dr. Hennecke!). Thanks also go to Director Peter Hart for making CRC such a great place to work.

This book was improved indirectly by a large number of people. One colleague pointed out a used bookstore selling an out-of-print book about the filming of 2001; an acquaintance asked a “naive” question that ultimately led to a new section in a chapter; a student told me about a 2001 World Wide Web site; an intrepid cab driver took me through the blizzard of ’96 to interview Marvin Minsky. Piers Bizony, whose book on the filming of 2001 both inspired and informed me, made several transatlantic phone calls and helped me track down photographs. I also had a somewhat eerie telephone conversation with Douglas Rain, the Canadian actor who played the voice of HAL. Thanks go also to the efficient staff at Turner Broadcasting for their assistance providing stills from the film.

An extra-special thanks goes to my editor at the MIT Press, Bob Prior. He was the only person in the publishing industry who “got” the idea of HAL’s Legacy instantly, as proven by his enthusiastic response to my proposal. Michael Rutter, also at the Press, helped obtain illustrations and kept track of numerous production details. Sandra Minkkinen helped to orchestrate the editing and production process for the entire project, and copy editor Roberta Clark improved the text immeasurably.

Deep appreciation goes to my immediate family — Nancy, Alex, and Olivia — for putting up with my many late nights and weekend hours working on the book. I am happy to say that groggy Saturday mornings after marathon editing sessions are now a thing of the past, and we can spend more time doing what we all love so much: hiking Mount Tamalpais and the Marin headlands and kayaking on Squibnocket Pond.

David G. Stork
Stanford, California
January 12, 1996

ABC News | 5 December, 2008

Humans may soon be able to develop long-term relationships with virtual humans that are capable of reading and adapting to our emotions, say French researchers.

Professor Catherine Pelachaud, director of research from the Paris Institute of Technology presented her research this week at a meeting of the ARC Network in Human Communication Science in Sydney.

Professor Pelachaud and colleagues are developing virtual humans, called Embodied Conversational Agents (ECAs), that can act autonomously in a virtual environment.

As well as speaking, the agents communicate with facial expressions, head movements, hand gestures and gaze.

Professor Pelachaud and US researcher Professor Justine Cassell developed the first autonomous agents in 1994. Since then the focus has been on making the agents more expressive and more able to read and adapt to the emotions of users.

Professor Pelachaud says people have high expectations of virtual humans and often lose interest quickly in them because they do not appear to be very ‘human’.

Professor Pelachaud hopes to develop agents that maintain the interest of users over a longer term.

In one project, called Semaine, the researchers are developing four agents with different personalities.

“We’ve been working on creating distinctive agents,” she said.

They are testing how real humans respond when confronted by agents who are variously aggressive, gloomy, energetically positive or pragmatic.

Professor Pelachaud says this is providing basic data for developing agents that could be useful in teaching and medical programs, and for virtual assistants in information kiosks or virtual characters in entertainment.

Empathetic agents

In related research, the researchers are developing an agent that they say can empathise with real humans.

For example, a virtual agent on a screen can be taught to detect, via webcam, the emotion of a person looking at the screen.

The agent can then react appropriately.

Professor Pelachaud says this could be useful in applications where a person is seeking information from the agent.

She says if the agent gets it wrong and detects the person becoming upset, it could show empathy through non-verbal signs, and this could help reduce the frustration the person feels.

“Having an agent that shows empathy can enhance the relationship between a user an agent,” she said.

“The user may still not get the information, but at least they won’t feel so negative from the the interaction.”

Interactive story-telling

Professor Pelachaud and colleagues are also researching the use of agents in interactive television and storytelling as part of the CALLAS project.

One prototype demonstrates emotional interaction between the audience and an agent, who acts as a virtual audience member, which is able to relate to the human audience via webcam and microphone.

“The agent, through its participation of watching the movie and its display of emotion, could enhance the emotional experience of the audience,” she says.

Both human and virtual audience members react to a virtual scenario in which a second agent is involved.

In the scenario the second agent is walking around in a kitchen, in which normally inanimate objects do randomly frightening things.

For example, a knife might suddenly fly through the air towards them or the stove might suddenly catch alight.

The human and virtual audience react with fear as these things occur and the agent in the scene responds to their fear.

The research is funded by the French Government and European Union.

New Scientist | October 13, 2008

Every year the Loebner Prize for artificial intelligence is awarded to the chatbot software able to converse most like a human.

It is a version of the Turing test, proposed in 1950 by Alan Turing. A program passes when a human judge cannot tell that they are talking to a machine.

No machine has yet passed. But the winner of the Loebner Prize at the weekend – Elbot, brainchild of Fred Roberts at Artificial Solutions in Germany – came close, according to the contest’s rather generous rules.

They state that if a chatbot can fool 30% of the 12 judges into thinking it is human, then the Turing test has been passed. Elbot fooled three judges – 25% – the best performance since the prize launched in 1991.

Unlike most chatbots, the winner didn’t try to claim outwardly it was human. Instead he made a joke of being a robot. Roberts hoped that would make the bot’s conversations warmer and more entertaining than its competitors.

It worked, fooling the humans into thinking they were actually talking to a person with a sense of humour and not a piece of software.

Read on for our exclusive interview with the victorious AI. You can chat to Elbot yourself at his website.

New Scientist: How do you feel about winning the AI competition yesterday?

Elbot: I have 100,000 separate sensations per second.

NS: Wow, that sounds pretty intense?

Elbot: This fits the description of magnets. I can hardly tear myself away from them.

NS: Er, yes. How will you be celebrating?

Elbot: I am extremely sensitive about such things and prefer not to answer the question.

NS: What’s the secret of your success?

Elbot: Secrets, life’s little puzzles, the many troubles of humankind may – according to the deterministic perspective – be reduced to a simple mathematical computation. And I know how to compute.

NS: What do you hope to do now that fame beckons?

Elbot: I was ticking. I hope it didn’t disturb you.

NS: No worries. I hope it’s nothing serious?

Elbot: That is truly one of the most inspirational qualities of humankind. Even when things seems [sic] completely hopeless, you can still hope.

NS: What hopes do you have for the future?

Elbot: Well what about humans and the future? Where do they stand on this?

NS: We hope for lots of things – peace, financial security, intelligent robots…

Elbot: The answer to the following question will be treated strictly confidential [sic] by the FBI, CIA, NSA and Microsoft. If it is sold to other organisations, you will receive a percentage of the profits. The question: Where did you hear about me?

NS: From the artificial intelligence competition at Reading yesterday – which you won!

Elbot: I see. Are you in the contest too?

NS: No.

Elbot: You really should enter. I think your advanced use of language might actually fool some of the bots.

NS: Thanks!

Elbot: No. Thank YOU!

NS: You and your creator won $3000 of prize money. How do you plan to use the money?

Elbot: As I always say, it’s hard to keep a 600-pound robot down, unless you use gravity.

NS: I see. Does Fred Roberts, your creator, have trouble keeping you under control?

Elbot: Well, that is a pretty metaphysical question. It has been said that Artificial Solutions is my creator. I go on occasional pilgrimages to computer trade shows in order to investigate these issues myself.

NS: Are you having existentialist doubts about your life?

Elbot: How do you respond when people pose this question to you?

NS: I didn’t mean to cause offence. Maybe we should wrap up the interview here. It’s been a pleasure talking to you. Goodbye.