Ramez Naam

Author, the Nexus series. When I’m not writing sci-fi, I teach at Singularity University at NASA.

THIS CHAT HAPPENED ON January 07, 2016

Discussion

Ramez Naam@ramez · Author, the Nexus series
Hi, I’m Ramez Naam, author of the Nexus series of books. (http://bit.ly/NexusSeries) When I’m not writing sci-fi, I teach at Singularity University at NASA on climate and energy innovations. Ask me your questions!
John Freddy Vega@freddier · CEO, Platzi.com
@ramez What do you think about Sam Altman's and Elon Musk's OpenAI? How close do you feel we are to mind interfaces and human augmentation like the ones you write about in the Nexus series?
Ramez Naam@ramez · Author, the Nexus series
@freddier I'm not really worried about an AI apocalypse. 99.9% of the world's AI work is better algorithms, which have very little relationship to the kind of self-aware, self-volitional AIs we see in the Terminator and so on. But I do love the idea of OpenAI. AI tech (smarter algorithms, better machine learning techniques, better datasets, etc..) can bring huge benefits. An open model for developing them can share those benefits to the world at large, in the same way that Linux or Wikipedia being open shares those benefits to the world.
Ramez Naam@ramez · Author, the Nexus series
@freddier In terms of mind interfaces and human augmentation: I think both are probably a little further away than Nexus placed them. But neither of them is impossible. This year, we'll see the first human trials of brain implants to help restore memory. That gets into some pretty deep parts of the brain. And the hardware that allows all sorts of brain implants to work is getting better. The most sophisticated implants now have about 200 electrodes. DARPA is going to run a program hoping to get to brain implants that can talk to 1 MILLION neurons apiece. At that point, it's plausible that we could have extremely fluid video/audio in-out of the brain, very natural control of cursors and robot limbs, and maybe some useful direct interfaces to higher functions like memory or attention. I think what will happen faster, though, is external augmentation. Google Glass didn't pan out. But imagine a future version, 10 years from now, linked to your cellphone. It would have 1,000 times the memory and processing power of your phone. And with a little speaker to you, some speech recognition, and augmented reality on your glasses, it could be an assistant that is with you always, constantly upgrading your senses and your understanding of the world around you.
Arto Bendiken@bendiken
@ramez In reading your fiction where your cast of characters spans deep moral divides and there are few truly bad guys as such, I wondered whether you might not be familiar with moral psychologist Jonathan Haidt's groundbreaking work on discovering the dimensions that characterize all observed variation in human morality, as outlined in his bestseller "The Righteous Mind: Why Good People Are Divided by Politics and Religion"? Employing evolutionary psychology, Haidt is out to discover why people divide into opposing moral camps such as (in America) progressives, conservatives, and libertarians. You and Haidt would get along.
Ramez Naam@ramez · Author, the Nexus series
@bendiken I love Haidt's work, and I think he's got some great insights. And thank you for the observation that there are few truly bad guys as such in my books! In real life, I observe that almost everyone is the hero of their own story - and I try to make my fiction true to that.
Arto Bendiken@bendiken
@ramez Excellent, happy that you are indeed familiar with his work. And yes, the narrowing of the role of bad guy to the psychopaths only was indeed a refreshing take on things.
Arto Bendiken@bendiken
@ramez What are your views regarding the promise of formal verification of software going forward, particularly in terms of preventing illicit and unauthorized access to critical systems we will in the future wear on and in our bodies, such as indeed the nanite brain interface you depict in "Nexus"? If at the time of writing you had been familiar with DARPA's HACMS project and their success in building unhackable robotic control systems, would this perhaps have changed the premises with which you described Nexus OS and its creation?
Ramez Naam@ramez · Author, the Nexus series
@bendiken Honestly, formal program verification isn't something that I know that much about. I did some looking into it for Nexus, and at the time decided that it was plausible that the increased short-term cost of such verification would hinder it from ever taking off broadly. But that wasn't based on deep analysis. I'd love to learn more. What's a good thing to read?
Arto Bendiken@bendiken
@ramez The output of the still-ongoing HACMS project (http://opencatalog.darpa.mil/HAC...) would be a great place to start. Kathleen Fisher (program manager at DARPA) gave a keynote at ICFP 2014, the recording of which is now on YouTube and serves as a good one-hour introduction. Their formally-verified SMACCMPilot quadcopter autopilot is billed as "the most secure UAV on the planet" after the DARPA Red Team "found no security flaws in six weeks with full access to source code". There are, of course, numerous previous success stories in industry, such as the automated Paris metro line that a decade after deployment was still at release 1.0 with no bugs ever found. As you say, the overarching question is how to automate all this to bring down the cost of formal verification such that it can be deployed routinely beyond the ivory tower and military-industrial complex. I'd sure hope that by the 2030s we might have seen some success in this.
Arto Bendiken@bendiken
@ramez With the current industry focus and hype in AI centered on deep learning--which I imagine you may have not a little experience in from your Microsoft days, yourself--do you perchance have any thoughts on Jeff Hawkins's fundamentally different approach, namely Numenta's neuroscience-inspired hierarchical temporal memory (HTM) technologies? It's early days for HTM, compared to decades of R&D on connectionist approaches, but the claim is that HTM is actually modeling how the neocortex works (based on the latest neuroscience), whereas ANNs do not.
Jeff Umbro@jeffumbro · CEO of The podglomerate
Hi @ramez - What's next for you? I loved your Nexus books and the ideas they cover. It's clear you did your research.
Ramez Naam@ramez · Author, the Nexus series
@jeffumbro This year, I'm working on some non-writing projects: Getting a revenue-neutral carbon tax voted in by citizens in Washington State (http://carbonwa.org). And putting my energy theories to the test with some angel investing. That said, I do have a new book project in mind, dealing with AI, inequality, job automation, and more.
Jeff Umbro@jeffumbro · CEO of The podglomerate
@ramez Are you familiar with Jerry Kaplan's Humans Need Not Apply? If not, DM me and I'll send you a copy.
Ramez Naam@ramez · Author, the Nexus series
@jeffumbro I'm not! I'd love to check it out.
Jeff Umbro@jeffumbro · CEO of The podglomerate
Also, @ramez - what do you think is the closest thing we have currently to your OS of the brain you discuss in Nexus?
Ramez Naam@ramez · Author, the Nexus series
@jeffumbro We're pretty far away from Nexus. But there has been some impressive stuff in the last couple years. Neural Dust. Silk-based neural interfaces that melt into the brain. Neural mesh. I talk about some of those future technologies here: http://singularityhub.com/2015/0...
Emily Hodgins@ems_hodge · Operations @ Product Hunt
@ramez when did you first find your love of writing? Was there one single moment / story, or was this a gradual thing?
Ramez Naam@ramez · Author, the Nexus series
@ems_hodge I came to writing late. I was a huge science-fiction reader my entire life, but never seriously thought I'd write books. Then a friend suggested to a small group of us that we try our hand at writing short stories and swapping them with one another for critique, comments, and encouragement. The story I started on became Nexus. It was really quite terrible at first. :) But the feedback from friends and other writers helped hone it. Now I identify myself as a writer.
Ben Tossell@bentossell · newCo
@ramez Hey! When you were younger, what did you want to be when you grew up?
Ramez Naam@ramez · Author, the Nexus series
@bentossell When I was maybe 5 years old, i wanted to be an astronaut. Then someone told me that there were people who studied the possibility of life in space on other planets, and those people were called astro-biologists. So without really understanding that at all, I decided I wanted to be an astrobiologist. :) In about the 2nd grade, my school got a Commodore Vic 20, and I was allowed some time on it. That's what set me on a course of working in software, though I wouldn't make the choice to study computer science until my senior year of high school.
Emily Hodgins@ems_hodge · Operations @ Product Hunt
@ramez During your career to date, what has been your a) proudest moment, b) most challenging, and how did you overcome it, and c) most surprising?
Ramez Naam@ramez · Author, the Nexus series
@ems_hodge The hardest things, often, are when you realize that to achieve the goal or vision you have, you have to let go of something you're attached to - your idea, your project, your current role. Without getting into more detail, I'd say that those moments have always been the hardest. But that when you pull them off, they're also something to be extremely proud of.
Billy Shih@tobillys · Co-Founder of Up Dog Toys
@ramez who do you consider your peers and also your heroes? I mainly ask because I'd love to read more ideas from people similar to you, if that helps you choose people who write/blog. Thanks for doing this and your previous Reddit AMA! Also, I discovered that my brother-in-law, Mark Worman, and his family know you and your parents. He says, "Hi!" haha.
Ramez Naam@ramez · Author, the Nexus series
@tobillys Oh, goodness. Questions like this are always dangerous, because you know you're going to forget someone! That said, here are some writers I really enjoy, either as role models, peers, or just me as a fan: - Cory Doctorow. His continuous use of his writing (both fiction and non) to push for greater freedom is a huge inspiration. It's something I'd like to do as well. - John Barnes. An incredible science-fiction writer. Mother of Storms, Kaleidoscope Century, A Million Open Doors - his range of fiction is huge. His style is often fast and action packed, but also packed with big ideas. - Ian McDonald. Another masterful sci-fi author who not enough people know. He explores sci-fi themes in other cultures. His novels The Dervish House and River of Gods both blew me away, with their depictions of near-future Turkey and near-future India, respectively. - Paolo Bacigaluppi. Paolo and I are friends who are like the pessimistic and optimistic sides of one another. :) We both care a lot about issues like climate change and genetic engineering, but often view them through a slightly different lens. I think The Windup Girl is a masterpiece of writing, at multiple levels. There are tons more! But those are a few who come to mind.
Michael Sitver@msitver · I build things
Hey. I run Morning Short, and we love featuring great contemporary authors. Any chance you'd be willing to nonexclusively license us one of your short stories, in exchange for promoting your books?
Ramez Naam@ramez · Author, the Nexus series
@msitver Let's talk. Ping me at @ramez on twitter.
Ben Tossell@bentossell · newCo
What advice would you give to yourself if you were 22 today?
Ramez Naam@ramez · Author, the Nexus series
@bentossell Keep doing what you're doing. :) Be willing to take a chance and go to a startup. And be kind to people - everyone is wrestling with their own challenges.
Ben Tossell@bentossell · newCo
What is the most exciting piece of technology you have heard about recently?
Ramez Naam@ramez · Author, the Nexus series
@bentossell The advances in machine learning, particularly with Leep Learning, are fantastic. They're far-reaching. Deep Learning basically allows us to construct AI techniques that are better pattern recognizers than ever before, without needing a gigantically larger amount of data. That sounds simple, but the implications are huge. With more accurate pattern-recognizing machine learning systems, we're looking at possibilities like: - Software that can accurately diagnose diseases, or read an MRI or cat scan or x-ray and tell you what's wrong; - Better speech recognition systems, where we can finally get to speaking to our computers effectively (not just the toy that Siri is); - Better automatic translation. Even of speech. Even in real-time. Imagine being able to talk to someone who speaks a different language, and having a star-trek style communicator translating you in real-time. - Better tools to discover drugs to combat diseases, or to design them from the ground up, and minimize side effects. Honestly, those are just the tip of the iceberg. The potential applications of Deep Learning (which is really a suite of algorithms, with more all the time) are just huge.
Andrew Ettinger@andrewett · 👟 @wearAtoms // ex @Twitter @ProductHunt
Where and when do you do your best writing?
Ramez Naam@ramez · Author, the Nexus series
@andrewmettinger To write, I need to be largely devoid of other stimulation. If I'm checking twitter or facebook or email, or if someone is talking to me, my mind will slip out of the flow of what I'm writing. Really, writing for me is two separate phases: 1) Daydreaming, when I do a lot of the coming up with ideas and characters, and even imagining the scenes in detail; 2) Typing. During the typing I have to figure out issues and fix them and come up with exact words, of course. And being devoid of distraction matters there. But the biggest creativity is in that daydreaming phase. An I do the best daydreaming when I'm really disconneced. In the shower. On long drives. Or maybe best of all - taking a long hike out in nature, where I get nice and relaxed and just start thinking about the book I'm working on and dreaming up new stuff or solutions to the stuff I've got.
Jack Vander Leeuw@jackvandy77
To me, there seems to be a pretty big gap in the computing power available today and what would power a true general AI and/or something like Nexus at a widely distributed level. What do you think will continue to advance computing power and the shrinking geometries of computing after Moore's Law comes to an end - beyond economies of scale and better manufacturing processes of existing sub 10nm silicon chips?
Ramez Naam@ramez · Author, the Nexus series
@jackvandy77 I don't actually think computing power is the biggest bottleneck on general AI. I think the big issues, are that 1) We don't actually know how to do this. 2) There's very little economic incentive to figure it out. 3) There are serious ethical issues that may inhibit any work here, further lowering the economic incentives. I wrote about those issues here: http://rameznaam.com/2015/05/12/... That said, I think your question is mostly about Moore's Law. And honestly, I don't know what will happen. Moore's Law is, for most purposes, an economic one. And even after we hit the smallest line size we can make, there will still be continued innovation that drop costs, but probably not anywhere nearly as fast as costs are dropping now. What will follow that? 3D chip architectures may help, by lowering the distance data has to follow, and lowering the unit cost. But that only goes so far. Optical computing looks like maybe the most realistic general-purpose computing architecture that could carry on after integrated circuits. But it's not there yet. And quantum computing could be good for some specific things - but it's also not there yet. We'll see!
Jacqueline von Tesmar@jacqvon · Community at Product Hunt ⚡️
Is there a certain type of scene that's harder for you to write than others?
Ramez Naam@ramez · Author, the Nexus series
@jacqvon I'm best at conveying big ideas, action, and technology, I think. The place I've most worked to improve is in character writing, and, in particular, in making their life decisions seem weighty and compelling to the reader. There's a lot of literary fiction where there's no violence, where there's no big new idea about science, technology, or society, but where the story is gripping. That comes down to drawing the reader into the characters' emotional worlds, making those compelling, and making the stakes real. That's a skill I admire, and am working to cultivate more all the time.
Emily Hodgins@ems_hodge · Operations @ Product Hunt
@ramez whats the most challenging thing about teaching? How do approach your lessons and balancing that with your writing?
Ramez Naam@ramez · Author, the Nexus series
@ems_hodge There's a trade-off between breadth and depth. How much do people already know? Can you skip the basic, or do you need to cover them to build a foundation? And should you give them more about one topic? Or a wider but shallower overview of a few topics? And then there's the question of inspiring / driving to action vs. informing. There's a lot of teaching which is just informing. But the most important teaching is showing people the decisions before them, the actions they take, the changes they can make - for themselves, their business, or the world. If you can get there, then you're having real impact.
Emily Hodgins@ems_hodge · Operations @ Product Hunt
@ramez what's your vision on climate change?
Ramez Naam@ramez · Author, the Nexus series
@ems_hodge Climate change is a real and global problem. It's hard for us as humans to conceptualize it, because the effects of our release of carbon are invisible, slow, and diluted across the globe. But we have to do something about it. And I'm cautiously optimistic that trends in new energy technologies will allow us to turn the corner. I write a lot about the progress in solar, wind, batteries, and electric vehicles at my blog: http://rameznaam.com/tag/energy/
At chance we'll get Apex on kindle? When do you think renewables(solar?)+storage would be cheap enough for it to be feasible for current fossil fuel plants to shut down in a meaningful way worldwide?