The Age of Em

Paradiso and Inferno in Robin Hanson’s ‘The Age of Em’

Robin Hanson’s future scenario in “The Age of Em: Work, Love and Life when Robots Rule the Earth” reminds me of Dante. On the one hand, many people will transcend (current concepts of) humanity and “transhumanize” – a word invented by Dante in Paradiso, Canto 1 – to become uploaded souls running on high performance computing circuitry. On the other hand, they will live in red-hot metal cities that create strong hot winds to disperse the excess heat generated by billions of uploads computing their way to continued existence. The infernal city of Dis, described by Dante in Inferno, Canto 8, comes to mind.

Though recently published, Robin’s book has been much talked about. “The Age of Em” has been many years in the making, and Robin has been so kind to send me many drafts, so I have been able to watch the book as it was being written and send comments. I think all futurists, transhumanists and citizens interested in possible futures radically different from today’s reality should read “The Age of Em.”

Robin’s future em world is derived from our world with one – and only one – big change: the arrival of operational and cheap mind uploading technology, sometime in this century. Robin’s methodology is to take our world as it, with all the facts and trends that we can see in technology, society, politics, and economics, and add mind uploading technology to scan living people and copy them to “ems” – software emulations running on suitable computing hardware.

Robin is refreshingly dismissive of metaphysical problems related to identity and skips the endless (and often boring) discussion of whether the copy is really the original, for which this reader thanks him wholeheartedly. Robin just adds ems to the world and tries to see what happens, guided by common sense and solid battle-tested assumptions in social science and economics. There ain’t no such thing as a free lunch, they say, and that applies to ems as well. Somebody must pay for the running cost of an em… or else.

The outcome: a world full of ems who ferociously compete for computing resources in red-hot cities like Dis. The best ems live hundreds or perhaps millions of parallel lives as independent copies, and most ems essentially work all the time to pay for the hardware and energy needed to run them. Some lucky – or unlucky? – ems live in permanent robotic bodies (probably more cockroach-like than human-like) to do physical work in the physical world, but em office workers live as pure software in virtual reality habitats. I imagine endless three-dimensional arrays of virtual cubicles where the poor sad ems work their virtual butts off for the right to exist. Dis, indeed. Most ems can’t even communicate efficiently with the Donald Trumps of their society, who run on much faster hardware.

What about humans (that is, old-fashioned biological humans running on obsolescing flesh, blood, and three meals a day)? Robin doesn’t even mention them much. The assumptions is that they are still somewhere on the planet, far from the em cities, but perhaps they don’t get three meals a day. Unless, that is, they were wise enough to buy insurance or invest in the em economy at early stages (in which case they could be rich beyond dreams). Some humans could make a marginal living in the shrinking human economy, with jobs that require personal contact with fellow humans (I guess counselors and shrinks will do well).

The Age of Em

But let’s come to the Paradise part now. Ems are transhuman cyber angels in virtual cities of pure, pristine computation, and we should try to see things from their perspective to realize that their life can be good. Forget that they get very limited leisure in objective time: they can choose to stretch their leisure periods at will in their minds. If work is boring, a smart em can split into as many copies as he can afford in the morning, send all copies but one to work, let that one copy spend the whole day in a spectacular virtual resort with plenty of awesome scenery, friends, and virtual sex, and save only that one copy at the end of the day. So from his point of view, the smart em remembers lots of leisure and only little routine work.

“Even if most ems work hard most of the time, and will end or retire soon, most remember much recent leisure and long histories of succeeding against the odds,” says Robin.

“To most ems, it seems good to be an em.”

We shouldn’t think of the functional virtual realities of 9-to-5 (sorry, 9-to-8:45 the morning after) work time or the spectacular virtual realities of leisure time as too different from the world that we know. According to Robin, the first ems (and the book only deals with the first ems – see below) will retain much of our psychology and preferences. In particular, sex will continue to important, and the lack of a reproductive function of sex could let ems explore a wide space of sexual attitudes and practices. A bored em could take an “open-source lover” who specializes in great sex.

“As long as someone else paid for their time, open-source lovers might allow a wide variety of situations and tweaks to be tried in search of a good spark.”

Of course, ems must save for retirement… or else. But even if final erasure looms ahead, an interesting features of the em culture is a relaxed attitude toward death: if other copies of you will stay around, you don’t need to fear your individual “local” death, at least not as much as we do. I think this cultural trait could be very usefully imported from the future em world to the real current world.

I like “The Age of Em” a lot. If you accept Robin’s premises and assumptions, his radical, intriguing, and perhaps scary scenario follows naturally. Some readers will love the book, some will hate it, but nobody will remain indifferent.

However, I disagree on one key assumption. Robin thinks that mind uploading is likely to be developed much before sentient Artificial Intelligence (AI). I think the two are likely to develop at comparable paces with strong feedback loops, with advances in one stimulating advances in the other (or roadblocks in one creating roadblocks in the other) and reach operational maturity at more or less the same time near the end of the century, give or take a couple of decades. But if I had to bet money, I would bet on AI coming first. Robin’s scenario is believable if uploading comes sooner than sentient AI, but I think this assumption is wrong. In my favorite scenario, human uploads and AIs co-evolve.

It’s interesting to compare Robin’s assumption to the opposite assumption that Nick Bostrom makes in “Superintelligence: Paths, Dangers, Strategies” (2014), a book that has scared the hell out of a lot of smart people with the prospect that the coming superintelligent machine overlords will exterminate us like vermin (or, even worse, exterminate us casually as a byproduct of their strange pursuits). Nick thinks that the AI approach (engineering a purely computational machine intelligence), though not the only one, is especially likely to result in superintelligence radically smarter than us. He mentions mind uploading as another possible path to machine intelligence, but doesn’t consider it in depth.

It’s important to note that Robin doesn’t think AI won’t work, but only that it will come after uploading. Based on extrapolations from current trends in science and technology, Robin thinks uploading could work in about one century (on that I tend to agree), and AI could only work much after that, perhaps in a few centuries (on that I tend to disagree).

But let’s accept Robin’s premise. Then, and this is an important point, there will be a developed em society without AI for some time. Now, the em economy develops very fast and pulls society, culture and technology in its wake at high speed, doubling at faster and faster rates:

“And within a year or two of this new doubling rate, the economy in such a new era might have doubled another 10 times, and thus could plausibly be ready to change yet again to a new era, perhaps even one that doubles in hours.”

Therefore, Robin’s essentially conservative analysis applies only to the very first Age of Em, which could last for only a few objective years, though those few years will seem like centuries from the subjective perspective of fast ems. Other Ages of Em, essentially unpredictable, will follow the first.

The Age of Em – Robin Hanson in Second Life, June 26

I hope I haven’t misinterpreted Robin’s thinking too much. But don’t worry:  Robin will present the book at the next Turing Church meeting in Second Life on Sunday, June 26. The presentation will be followed by a discussion, and everyone will be able to ask questions.


The meeting will take place on Sunday, June 26, at 8:30am Pacific time (11:30am Eastern time, 5:30pm European time, 4:30pm GMT). Please check the time if you are in another timezone.

[Click here to teleport to Stonehenge @Terasem] or teleport directly to Terasem, 180/180/31.

Age of Em, SEcond Life

  • David Román

    I can’t make it, being in the Asian time zone (the time will be 1130 Sunday night, my time, and I work early Monday), but I hope somebody makes this question for me: if you think of the kind of tech advances that would lead to Ems, and that are now driving virtual reality, wouldn’t they be counterproductive for Transhumanism as a whole and for the Turing Church in particular? I see a clear risk of seeing mankind plugged to the Matrix, making up whatever it needs in software paradises, with no need to maintain technological progress or explore the actual, uncomfortable, distant universe.

  • Giulio Prisco

    Hi David, too bad you can’t make it. This is a common problem with online events, you just can’t find a time that is good for people in all timezones. I will try to record the talk and discussion on video.

    In the book Robin makes it very clear that computation is physical. Ems “live in software paradises” as far as their subjective experiences are concerned, but their software paradises and their own minds really run on physical computers that need physical space, energy, cooling and maintenance (and somebody has to pay for all that). Therefore, ems do need to maintain technological progress and a strong presence in the real universe.

    I see that some people joined your meetup group in Singapore. I just joined, I hope I’ll be able to attend some meetings remotely.

    By the way, this post was republished in IEET, Google News, and some VR magazines are picking it up. I hope there will be a good audience and discussion next Sunday.