Book Review: Virtually Human by Martine Rothblatt


“Virtually Human explores what the not-too-distant future will look like when cyberconsciousness—simulation of the human brain via software and computer technology—becomes part of our daily lives.” by Martine Rothblatt Ph.D., MBA, J.D.

Martine is a lawyer, entrepreneur, and medical ethicist. In 1990 she founded and served as Chairman and CEO of Sirius Satellite Radio (now Sirius XM). When her daughter was diagnosed with a rare disease, Martine left Sirius to search for a cure which she found. She founded United Therapeutics in 1996 and has since served as Chairman and CEO. Martine is also a leading legal advocate for the human rights and has led the IBA in presenting the UN with a draft treaty on the genome and one of the founders of the Tersem movement as well.

In this book Rothblatt suggests that we’ll soon be surrounded by mindclones, digital copies of ourselves that are conscious or what Rothblatt calls “cyberconscious” digital entities. Your mindclone will be exactly like you, indistinguishable, but running on an advanced computer and software system which does not quite yet exist today. But Rothblatt argues it isn’t that far away.

For example consider Bina48, the “world’s most sentient robot”, was commissioned by Rothblatt and created by Hanson Robotics. Bina48 is a real working mindclone of Martine’s wife Bina Rothblatt that can engage in conversation, answer questions, and controls a robotic head. But Bina48 is a pretty early stage prototype obviously.

Rothblatt not only suggests the potential for the appearance of mindclones, but also examines  the possibility of these entities becoming very numerous and even treated as legal persons in society. Mindclone and human/mindclone sex is covered as well in case you were interested. The book argues for a sort of “digital immortality” through mindclones that many will find attractive but it will also be controversial.

Is your mindclone really you?

To counter a lengthy list of objections to creating mindclones Rothblatt takes a “kitchen sink” approach including mentions of various possible solutions. It’s an engineering approach that suggests a series of relatively easy obstacles to overcome. For example the “Brain Activity Map” project gets mentioned as do Kurzweil’s explorations of machine creativity. But while we can see solutions to specific problems, i.e. making machine minds creative, the suggestion that this leads immediately to an integrated solution may not be quite accurate. From a technical perspective the book doesn’t really outline a specific approach to constructing mindclones but merely mentions various areas of research that might be used to produce “mindware”. While the lay reader obviously couldn’t be expected to grasp a more detailed argument, specialists are going to find themselves looking for more meat here.

Even Rothblatt doesn’t go quite this far, instead suggesting that our notion of self is itself extendable and malleable and today already includes some of our digital footprint via social media services like Facebook, Twitter, and Instagram. Rothbaltt suggests that your mindclone and your present self are really part of a future larger “you” encompassing all of the above. Perhaps most controversially, Rothblatt argues that if you are active on Twitter or Facebook, share photos through Instagram, or are just blogging regularly, you’re already creating a “mindfile” that can be used to reconstruct your digital homonculous, your perfect mindclone.

This isn’t just about the potential for creating a simple Markov model of your social media postings, we know how to do that, Rothblatt is suggesting constructing mindclones that are really indistinguishable from the original person. While these public displays may be revealing, people do not commonly record or publish their innermost thoughts and feelings. At best a mindclone created this way would be the public relations version of you, the you that you want to project into the world socially. But we all know that in truth the social person is not the same as the true inner person we are. It seems to me that a real mindclone would have to recreate this inner experience of being myself and social media does not include that sort of information at all.

Recent exaggerated claims about the chatbot Eugene Goostman aside, existing AI software still has a good ways to go to pass a Turing Test. And Rothblatt is suggesting here that your mindclone will be indistinguishable from you even by those that know you best, your loved ones and lovers, which is much more challenging. In contrast Turing specifically ruled out such “expert” judges from his test instead suggesting that average people or randomly selected non-experts should be used. I’m going to openly state I’m skeptical about this idea of using social media to produce a mindclone at this level of fidelity and I still wasn’t convinced after reading the book.

However, a more detailed capture of human experience via neural interface, audio and video recording, etc. might be a different matter entirely.

Rothblatt’s main arguments aren’t tied to the notion of reconstructing a mindclone from social media, but a critical view of this point does suggest that success might be somewhat further off into the future than the timeline which the book lays out. Rothbaltt interestingly discusses the theory that the mind is about making predictions, an idea I have personally explored through the development of AI software for making predictions about the future. The notion is further backed up by recent neuroscience and I think it warrants a much deeper exploration than was given here. Perhaps that is another book.

The social, economic, and personal implications of viable mindclones would be enormous. A mindclone could fill in for you at work and would be capable of doing any intellectual task that you can do. They might do them faster too. But mindclones would also be separate conscious entities.

Would mindclones be treated as slaves to the original person or given rights and freedoms to act independently? Could mindclones get married if their real world analogues weren’t? Can mindclones be murdered? Rothblatt discusses these sorts of implications in detail and argues convincingly I think that mindclones would eventually be granted rights as “real” persons. This is where Rothblatt the lawyer and ethicist rather than Rothblatt the inventor and entrepreneur shines through.

Although the book’s subtitle promises coverage of both “the promise” and “the perils”, this isn’t a book about risks. You won’t, for example, see anything here about use of mindclones to imitate someone without their permission. A perfect mindclone could be used to create materials for blackmail,  for identity theft, to access bank accounts, and perhaps for crimes we haven’t yet imagined. A convincing copy of a world leader could be used to start (or end) a war. In fact these sorts of ideas are not even new; false digital voices were reportedly used against the Taliban in Afghanistan for example. Darkside uses are not mentioned here but we should at the very least examine them.

In summary this is an important book that h+ readers and transhumanists generally are going to want to read. This book makes a nice companion to Kurzweil’s books and will be understandable by most lay science readers. Academic readers will be happy to find an extensive notes section, but sadly the eBook version page numbers do not correspond to the pages listed in the notes. Even where I felt the book went off track, e.g. suggesting my Facebook feed is a workable mindfile, it does so in an interesting way that will prompt further detailed examinations of the technical challenges. If your social media postings aren’t enough to create a viable mindfile, what level of detail exactly would be enough? Interesting question.

I also enjoyed the Ralph Steadman illustrations as I am  a huge Steadman fan. The book is available starting today from Amazon, get it.