David Segal writes a feature in the June 2 edition of The New York Times about Dmitry Itskov's dream that we all live, peacefully, in machine avatars: "a digital copy of each mind in a nonbiological carrier." The funding, research programs and Global Future conferences borne of this effort, dubbed The 2045 Initiative, paint an optimistic view of an immortal human future only three decades away. The prognostications of this line of thought, backed up by the likes of Roger Penrose and Ray Kurzweil, are at once both absurdly far-fetched and maddeningly timid.
On the far-fetched side of the ledger, proponents suggest that machine avatars are less than a lifetime away; Martine Rothblatt suggests that avatars with downloads of our consciousness may seem absurd now, but it's no different than the medical view of liver transplants in 1960. This is an awkward comparison. The first liver transplant was the culmination of well more than 100 years of incremental advances in medical research and practice- from artery and vein transplants to skin grafts, doctors comphrended the plumbing and mechanisms behind transplantation thoroughly, and they understood the problem of organ rejection for decades as donor matching was refined and drugs were invented. The development of conscious avatars has no such rich historical antecedent.
So what does it take to make these avatars real? We face two significant hurdles: body and brain. One challenge is the invention of a machine that approaches the sinuous dexterity of a human; the second is that we need to embody a person's actual consciousness into avatar software. On the machine side, I have written about mechanical innovation and how it simply fails to follow Moore's law regarding the speed of computers. Motors, sensors and batteries are hard to improve, and the transformative improvements we do witness happen in fits and starts with little forward predictability. Practical, robotic forms that mimic human-level physical capabilities could easily be twenty years away or two hundred years away. As for downloading consciousness, we fail to appreciate how much harder it is to build a new silicon-based mind than map the signals in an existing brain. We have witnessed major progress in processing brain signals, and this in turn leads to prosthetics that can have several controllable motors instead of just on¬e --all triggered cleverly either at the brain itself or in large muscle groups near the robo-prosthetic's attachment point. But the human body has more than five hundred muscles and ten trillion nerve cells. Just interfacing to all of this will be a monumental undertaking. In Brainspotting I suggest the only way to eventually succeed may be by inventing whole new species of nanorobots. But copying consciousness means solving all the above challenges, and then solving one massively larger one: building a complete, working brain. That is and then a miracle happens territory, since we do not have a clue how to forge a path toward that goal. All said, we will have robotic, dexterous pseudo-human vessels far, far earlier than we will have downloaded an individual human's mind to insert in those vessels.
On the timid side of the ledger, futurists make the mistake of hanging on too dearly to the present. Given the ability to download consciousness and build machines with human musculature, do they really believe that we will all simply replace our soft tissue with silicon and our brain with a neural net simulation in software? This vanilla replacement trope misses the point of disruptive innovation. Imagine, for a moment, sipping tea with some mechanical engineers and carrier pigeon trainers in the year 1900. The engineers explain how they are working on a whole new way of having people communicate: they're going to build mechanical pigeons that don't need food and don't live on roofs. People will have thousands of mechanical pigeons in every town, and everyone will be able to send and receive messages daily, if not hourly. Our cellphones don't eat worms either, but that's about the end of any similarity our new age of communication has to the naïve prediction one might have posited then. Now consider today's predictions more critically: Itskov hopes avatars will end world hunger, because we all become machines that need no food. He imagines a world where we mass-produce a low-cost avatar for everyone. Charities now devoted to feeding and clothing the poor become avatar fabricators and distributors. Really? If avatars become electronically feasible machines, there will be a whole trajectory of human-controlled avatars way before anyone has a clue how to download consciousness. This means telepresence on steroids, as I describe in CEO of Me, Inc. The wealthiest will have the very coolest avatars, and nothing will stop such telepresence from halting at human-level capability. Welcome to commodity Supermen-bots, so long as you have a great brain-machine interface and a whole lot of capital. What about the poor? If we really did discover how to download consciousness, what's cheaper? Download a human to the Cloud, or build and maintain a real machine avatar for each soul? You can see where this goes, and it is not a pretty picture. Suffice it to say human identity, consciousness, authenticity and control would witness massive boundary shifts.
In the last chapter of Robot Futures I suggest that the antidote to today's robo-hyperbole is to round out our crop of visionary engineers. Give them courses in sociology, ethics and human factors. Today's robotic inventions are no longer trapped in laboratories- they are influencing our real future. Do we really want the track of technology development to be set by the railway barons searching for immortality? Let's get the inventors to think more clearly about what the whole future might become.