Can AI defeat death?

Danny, the much-loved son of one of the people I love most in this world, died this summer. Yesterday would have been his nineteenth birthday.

No words can make sense of Danny’s death, but there are far more powerful words inspired by his life. Some are Danny’s own: his poetry, his witty off-the-cuff remarks, his messages to friends and family, the wonderful novel he wrote. Others come from the people he loved and who loved him: after he died, the outpouring of appreciations and reminiscences made Danny feel more alive, more vibrant, more important and closer. The conversations we had with him and about him continue. He inspires us and makes us laugh and think and feel anew, and in new ways. He may have died, but he will never feel dead to us.

And yet he has gone. For ever. And my dear friend his mother, who has been there for me at the worst as well as the best times, is left bleak and alone beyond my comforting.

Most religions offer a hedge against death: cleave unto us and you will enter eternal life, be resurrected, start a new existence in a better or at least a different place. Perhaps those promises come true, perhaps they are more a way of soothing the rawness of death and bringing people together. As Philip Larkin wrote, we shall each find out for ourselves one day. But in the meantime technology offers a new possibility: that we can live on in digital form, or swap our carbon bodies for silicon and steel.

That we can defeat death.

For obvious reasons, I’ve been thinking about that.

Some of AI’s promise of immortality has already been delivered: many of Danny’s words and thoughts live on in digital form. A generation ago, all we would have had left of him would be our scattered memories, perhaps some letters and photographs, a manuscript or two, a handful of personal possessions and clothes. Now we have his social media posts, his messages and artworks, his creative writings casual and deep, his life in photos and videos, stored in the cloud for us all to access. At the darkest hours, we can click a link and once more hear his voice and his laugh, read his words as they were written, see the way he moved and watch him looking back at us, full of curious mischief and irrepressible energy.

But that does not feel like enough. The Danny on the internet is done. He cannot tell us anything new, or answer our questions or hear what we want to say to him. He is nothing but our memory of him.

We are left wanting more.                                 

A few months ago I watched – ok, having checked, I watched it back in 2013 – an episode of Black Mirror: Be Right Back, where a woman uses technology to create a robot simulation of her dead boyfriend, drawing his personality from an training dataset of his social media posts, texts and other electronic records.

Today that technology sort-of exists. While carbon-freezing Han Solo is nothing but a Hollywood myth (and a Lego kit for gruesome seven-to-twelve year olds), it is not a million miles away from cryonics – preserving the body at ultra-low temperatures, in the hope that future scientists will have cracked resuscitation. Companies like bioquark are working to reverse death by reanimating brains via an injection of stem cells and amino acids into the spinal cord. Dmitri Itzkov’s 2045 project and an LA-based outfit called Humai are having a go at building robot avatars you can upload your digital self into. Nothing beyond frozen bodies has emerged from these projects, but early days and all that.

Meanwhile researchers at MIT’s Media Lab and Ryerson University are working on a more modest digital-only augmented eternity: a chatbot trained on our digital outpourings to embody our personality in new, live conversations.

We have always been able talk to the dead, but now they can answer back.

Does augmented eternity work? Reactions have been predictably mixed. Some think it’s a great, and obvious, next step for humanity. Others worry that our online selves are only a fraction of our full human identity, and that identity is not reducible to pattern-reproduction of past utterances, so at best we will end up interacting with a caricature. Definitely the early chatbots seem pretty primitive - @DeepDrumpf isn’t anything like as vivid as @RealDonaldTrump. But the technology and large-enough personal data sets to train a neural network are both developing fast. Siri was pretty clunky when she launched back in 2011, and Microsoft Research’s MyLifeBits project estimates the average person’s lifetime data would take up less than one terabyte of storage space, so there’s room in eternal life for all of us.

But I find myself asking a deeper question: could a chatbot ever be human, or could it only mimic a real person? Could silicon and steel become as conscious as flesh and blood?

Nobody knows. If you’re convinced by Mysterianism (the only philosophical term inspired by a 1957 Japanese sci-fi film and/or a 1960’s Michigan-based Mexican garage rock band), you buy into the theory that humans are just too stupid to understand something as complex as consciousness. Perhaps consciousness can take many forms: look underwater and you will notice octopus and squid demonstrating what looks very like conscious behaviour, despite an entirely different design of nervous system from our mammalian brains. If you follow Dr Guilio Tononi and friends’ Integrated Information Theory, you have to accept that anything can become conscious given the ability to connect and organise information (Google Contacts, what are you thinking about?). Then there’s good old quantum mechanics. Max Planck saw all matter as derived from consciousness, and Einstein said, in a condolence letter for the death of a friend:

“For those of us who believe in physics, the distinction between past, present and future is only a stubbornly persistent illusion.”

I find myself wondering, in a parallel universe, might Danny still be alive?

But that’s not how quantum theory – never mind experienced-reality – works. Yes, what we think of as matter is likely a function of, and definitely shaped by, our conscious observation; but the constant in the calculations – energy – is nothing but a conserved quantity, a function of dependent variables based on the assumption that fundamental laws of physics are unchanging.

It is a mathematical truth, not something you can reach out and touch.

And that is what we want with the dead: not just for them to still be alive but for us to be able to touch them and be touched by them, to bury our faces in their hair and sob out how we miss them, how much we love them, how sorry we are and how desperately we want them back.

Rudyard Kipling, of all people, got this. When he inaugurated the World War I monument in his village, a monument that bore his only son’s name, he began his speech to the assembled bereaved families with the words:

"We all know, grief cannot be cheated…"

Because that is what we are trying to do with all our theories and chatbots and cryogenics. We are not trying to defeat death, we are trying to cheat grief. Cheat the lockstep of love and loss that makes us fully, deeply human.

I am left with two thoughts, two memories.

The first takes me back thirty years to a desk in the library of the college where I first met Danny’s mother, where I sat deciphering a poem written down almost three thousand years ago: Homer’s Iliad. When Hector, the closest thing to the poem’s hero, is about to be killed by the rage-maddened, half-divine, doomed-to-die-young Achilles, he says something that moves me as much today as it did when I first stumbled through the original Greek:

"My doom has come upon me…but let me first do some great thing that shall be told among men hereafter."

That is what we all want: for our lives to matter, to be worth remembering, to inspire the future. For people to remember us when we are gone.

The second memory is more recent, an interview I read, not long after I got married, with Mariane Pearl, whose journalist husband Daniel was tortured and executed by Al Qaida. Mariane talked about being inspired by how he appeared in the photos taken by his captors.

"In one, he is giving the camera the finger. In another, despite the gun at his head, he is smiling. Danny, he was so brave. At the moment of facing the real thing, he said fuck off. You know? So I thought, I just can't invalidate his courage. I have to stand up to those guys no matter what: Danny did it in the face of death, I do it in the face of life…I consider happiness as a constantly moving state. Happiness has to do with other people. Happiness has to do with overcoming obstacles. You fight for them and even if the fight is really bitter, your triumph is happiness."

I’ll end with two words of another Danny, the Danny I knew: the name he chose for his Facebook account:

"Glorious Victorious"

Yes.

Footnote: Danny died as a direct result of the opioid epidemic. Donations to Shatterproof in his name can help transform the broken addiction treatment system: www.shatterproof.org/memorial/daniel-foley