A recent issue of the New York Times Sunday Magazine featured two articles on artificial intelligence. One story described the use of a deep neural network in determining when to switch from curative care to palliative care. This decision, to switch from healing to comfort, is agonizing and emotional, for patients and families that must consent, for those who provide care and realize that there will be no cure. The timing of the decision is critical, for there is a sweet spot, a window between three and twelve months before death, when making the right decision makes a world of difference, can make for a good death.
Enter Anand Avati, a graduate student at Stanford. His team started with data on 200,000 deceased patients. They entered all of the data for the first 160,000, including medical records of diagnoses, procedures, hospitalizations, and treatments, as well as the actual date of death, then used the remaining 40,000 patients to test the accuracy of their algorithm. The results were remarkable. Not only did the algorithm successfully predict those who would die within the nine-month palliative care window with 90% accuracy, but it also was 95% accurate in identifying those who would live longer than twelve months, in other words, those for whom it was definitely too early to switch from curative to palliative care.
This might have done little for poor Lazarus, given that the gospel makes it look like God intends for Lazarus to die and for his sisters and friends to grieve just so Jesus can pull off this penultimate sign. “This illness does not lead to death.†Really, Jesus? Great at healing, but maybe not a great diagnostician.
The same issue of the Times Sunday magazine contains a feature article on the use of an algorithm in child protective services, focusing on the Allegheny Family Screening Tool created for Pittsburgh and the surrounding county. Rather than a human screener having to make a snap decision based on gut and the records of one agency when a case of suspected child abuse is reported, the tool uses four years of data and a hundred criteria from eight different databases to assign a risk score to each report.
While Avati’s work is about a good death, the work in Pittsburgh is about the chance of a life, any life at all. In 2015 alone, 1,670 children in the United States died from abuse and neglect. Before the algorithm, 48 percent of low-risk families were incorrectly screened into the system, while 27 percent of high-risk families were screened out, an error rate that is not only inefficient, it is deadly. During one recent four year period, the child protection agency received 18 calls where a child would later die or be gravely injured. Eight of those, or 44%, had been reported but screened out. No one is willing to live with that error rate. We clearly prefer that our children die in school.
The child protection algorithm in Pittsburgh seems to be working as a predictive tool, is producing good results, though there has also been criticism. It is great to build a system to eliminate human bias, but what if the data that is fed into the deep neural network is already biased? Can the machine filter for the effects of institutional and systemic racism? That would seem necessary, but also seems like an impossibility. And concerns about bias are important, as systems like this are increasingly being used in the justice system, where neural networks are helping decide who gets bail. There is that old adage “garbage in, garbage out.†The fear here is that bias goes in and bias comes out.
Part of what complicates the discussion even further is that even those who design and build neural networks don’t know how the networks create the algorithms to reach a particular decision, because they are learning machines, changing with new input, just like us. We have created machines almost as mysterious as our own brains.
We were always taught in math class to show our work, but these massive number-crunching networks don’t do that, and if they did, it would involve way more data than we could ever take in ourselves, the very reason we have turned over this sort of work to computers. The Economist reports a growing interest in regulating these deep neural networks, requiring that they “show their work†or at least articulate the algorithms, especially in areas that are already heavily regulated, like banking. It isn’t entirely clear if this goal is achievable.
Then there is the growing chorus of big tech gurus who are warning not just that technology has allowed Russian agents and bots to undermined the fairness of our elections and divide our country, that the integrity of the ballot box is not longer guaranteed, but are also warning that AI itself, deep neural networks, might become a threat to human kind, not just taking our jobs, but waging actual war against us.
What I can’t figure out is how they are just figuring this out. I was taught about the threat of deep neural networks almost 27 years ago thanks to Terminator 2. Hasta la vista, baby.
We all dread sequels, but sometimes the sequel is better than the original. Terminator 2 was way better than the first film in that series that seems to never end, and the second film made way more money than the first, six times as much even adjusting for inflation. The same is true for the re-booted Mummy franchise. The Mummy Returns earned $25 million more than the first film on its opening weekend, though they were only two years apart and both released the first weekend in May.
I’m sure someone is working on an algorithm to figure out what movies are going to turn a profit, though Netflix spent many years and millions of dollars trying to figure out movie recommendations with questionable results, and AI attempts at writing fiction are, at least to date, hilariously incoherent. Check out AI written Harry Potter stories if you don’t believe me.
Both Terminator 2 and The Mummy Returns added a kid, so while neither was a “family film,†both were films about family. Both had the feel of a video game: racing against time to unravel clues, reach checkpoints, and complete missions or you will die and with you, all of humankind.
Because the Army of Anubis, made up of jackal warriors from ancient Egyptian religion, is not what we want to see running down our streets, nor is an army of cyborg terminators.
If Hollywood got a hold of this part of John’s gospel, it would probably be as big of a flop as Noah: Action Hero, but you wouldn’t really have to do much to make it exciting, for all of the elements are there. There is real danger for Jesus and his followers. In fact, when Jesus finally gets around to suggesting a trip to Bethany, the disciples try to change his mind, reminding him that he was almost killed the last time he was in Judea near the capitol. It is worth noting that the courageous disciple who is not only willing to go to Judea but even offers to die with Jesus is Thomas the twin, given such a rotten reputation by history. So the man is a skeptic. Aren’t we all?
The scene at the tomb is pretty gruesome, worthy of the mummy, the Boris Karloff original or the reboot. We are told that Lazarus has been dead for four days. That number is important, for Hebrews in that age believed that the spirit hovered around the body for three days before going to Sheol, so the fourth day is way too late. When Jesus says to open the tomb, Martha warns about the stench of decay, but Jesus will not be put off.
The thinking of the authors and the communities that produced our disparate gospels, that produced the living and evolving tradition that we have inherited, is like a sealed box, like a neural network. We can only guess at the calculations that went into forming this story this way, this calling of a dead man, wrapped in strips of cloth like the fictional mummy, from the tomb.
The calling of what seemed lifeless back to life.
Even before Lazarus is called from the tomb, lurching into the light, we have this entire exchange about belief in life after death. There is a certain irony to it if you know the broader cultural context. Belief in any form of life after death was fairly new in the Hebrew religious trajectory, and was not universally accepted. For example, the Sadducees, the sect that controlled the Temple, believed that dead is dead. They only accepted the written Torah. There was a movement that believed in a living and evolving faith, a body of evolving belief they called the Oral Law, one movement that believed in resurrection. It was the ultimate bogeymen for the early Jesus movement, the Pharisees. When Jesus quizzes Martha about her belief, asks if her brother might yet rise from the dead, she repeats Pharisaic doctrine.
She affirms a resurrection on the “Last Day,†the mass restoration of life at that moment when God has made the world right again, for resurrection was physical and communal, none of this individualized heaven and fluffy white clouds. While belief in resurrection would become a less prominent feature as Pharisaic Judaism morphed into Rabbinic Judaism, it would become central to Christian belief, both the bodily resurrection of Jesus, the founding event of the faith, and the bodily resurrection of believers, though again, the Eastern and Western churches would split in a significant way. Like the Pharisees, the Eastern church still views resurrection as communal, much like the image in 1st Thessalonians 4, where trumpets sound in the air, and the dead in Christ are the first to be restored. Everyone is resurrected at once.
And down the centuries, us, with so many questions. We mostly discount the idea of bodily resurrection, for how can a body be resurrected that no longer exists. If every atom in my body that once belonged to another body were removed, there would be nothing left, for like the madman, I can proclaim that I am Legion, this flesh made up of new sunlight and old dirt, part cow, part pine tree, and part people who have gone before.
But who knows what a power that speaks creation into existence can do?
It is hard not knowing, but I do not know and will not know. This I do know: I don’t have to wait until I pass through the veil of death, or run into the brick wall of death, depending on how I end. I don’t need to hear horns blow and see the saints in the sky, or watch as the anti-Christ takes power, though about that…
I don’t need to invent a resurrection of the dead, for I have seen folks who were dead come back to life. Not like Lazarus, to be fair, no mummy dearest crawling from the tomb, but living people who were all but pronounced, assigned a time of death.
It is part of a pastor’s job to weep with those who weep and to sit with those who wait beside the bed.
And so I have been there again and again when doctors have said “We have done all we can do. The damage is just too great.†And I have welcomed some of those would-be corpses back into church, living and breathing people.
And I have seen drunks get sober and addicts get clean.
I have seen those drowning in grief crawl onto the shore of life, wet but alive.
It always takes a call, sometimes from the divine, but usually from one of us.
Lazarus, come out of the tomb.
This is belief. This is call, our call. It is our call to call others out of living death and into whole life, and to trust that when we are prematurely curled up in tombs of our own making, that they will weep and call us back out into life.
We do that individually, with addicts and those trapped in sinkholes of depression.
We do that when we name and affirm those whose lives are devalued, when we say Black Lives Matter and hang a rainbow flag and tell women that they are not meat meant to serve the whims of men.
We do that when we remind our church and our community and our nation that we are better than this, for we are surely trapped in a living death of division and violence, and we need to call ourselves out of the hell we have learned to tolerate.
What would Jesus do? Jesus wept. Jesus called. Jesus loved.
Come out of the tomb, Lazarus.
People who don’t believe don’t bother to call people out of living death.
I hope that those who are still dying can live until they die and die well when the time comes, because we are finite. But until that moment, we must speak one another back to life as surely as God is speaking each of us right now.
I hope that neural networks and underpaid and overworked caseworkers can save more children.
I hope that the day will come when the hypocritical rhetoric of thoughts and prayers will be turned into common sense action, when we will once again be shocked by carnage in our communities, when phrases like active shooter, mass casualty shooting, and school shooting will no longer be a part of our daily speech.
Our nation has been upgraded and rebooted before. We are already in a sequel, United States: The Federal Republic 2.5. We are overdue for new life, new creativity, a new nation. We are overdue for an upgrade. We can be the coders who insure that everyone is included in the new version.
Our faith has been upgraded and re-booted again and again, sequel after sequel, some really bad if we are honest, some timeless and classic. We are beta-testing a new version of our faith, and we are the neural network that is figuring it out, and let me assure you, the inner workings are just as mysterious as the deepest neural net, the most artificial of intelligences, and we are still encountering bugs in the program. But we have this power source that has lasted for two millennia, this story of a man who challenged us to see things in all new ways, to experience life in all new ways, to engage life in all new ways.
There is more life and more light and there is hope, even in the most hopeless of times. When despair wraps its claws around your heart, resist.
I know that he will rise again.
Indeed he will, Martha.
Believe. Then call them back into life. May we be called back into life.
Amen.