PASTCF President’s Column May 2023
Pondering this time between Easter and Pentecost, it occurred to me that something amazing happened during that interval, epitomized by the journey to Emmaus: the disciples were finally getting it! Whereas before Easter they were still admonishing Jesus not to talk about suffering and were squabbling who might be the first in Jesus’s (conventional earthly) kingdom, afterwards they gained a deeper understanding of his mission. That said, “getting it” is not a static state; we continuously need to repeat and update our search for the true meaning of the gospel in our time and place; for what it means to be a community of faith.
To a certain extent, the scientific process follows a similar pattern. We stumble about for a while, trying to understand what it all means, and then, all of a sudden, the pieces fall into place, and we get it. And also, in the realm of science and technology we need to be prepared to revise our understanding (although that is not so much tied to our time and place). I will highlight two examples of technologies that require a thorough (re)assessment.
At the Ecumenical Round Table gathering with our Lutheran friends (the Lutheran Alliance for Faith, Science, and Technology), at the end of April in Chicago, the PASTCF and LAFST boards started a serious conversation about Artificial Intelligence. Sure, we have been well aware of the great promises of AI: its ability to provide better and faster (medical and technical) diagnoses than most humans can; safe self-driving cars; intelligent efficient scheduling of traffic lights; and more. But we are also aware of sticky problems: much of the input to AI tools came from copyrighted material (how is that handled?); other input came from questionable sources (that doesn’t help); ChatGPT may be able to preach a great sermon, but where does its theology come from, and do we really prefer it over a human pastor? And there are deeper questions: we are aware that AI is embedded in countless systems in an unregulated fashion, so the opportunities for subtle undetectable manipulation by nefarious actors are near unlimited. This is an area of technology where we need to reassess our stance, to have a clearer view of what it all means and to question whether we really “got it” over the previous decades.
The weekend after our gathering there was another jolt highlighting the urgency of such a reassessment: the resignation of Geoffrey Hinton from the Google corporation. Hinton has been called the “godfather of AI” and expressed, after his resignation, serious misgivings about his godchild and the work he did and inspired. This has become public through the many interviews he gave. I seriously encourage anyone to Google “Geoffrey Hinton” and take a look at one or more of the articles that pop up. The most serious issue identified by Hinton is that AI is uncontrolled and becoming uncontrollable. Imagine asking your smart car to take you to your doctor’s appointment and the car decides that that is a bad idea (for whatever mysterious reasons) and instead delivers you to the police station, parking itself illegally. One obvious problem is that even those who designed the software can’t really trace back how it reaches its conclusions. The more scary part is that the software may figure out how to make itself smarter, in an uncontrolled fashion. Already five years ago, Sir Martin Rees, the Astronomer Royal, in his fascinating book “On the Future” (2018), paid extensive attention to the potential development paths of AI. Consciousness, and particularly self-consciousness, remains one of the great mysteries that are not understood. Consequently, Rees argues, it is quite conceivable that AI may develop self-awareness, allowing computers to develop into an electronic life form — that could coexist with, assist, or supplant human life. We need to tread carefully in developing and applying AI. An encouraging sign is that, while I was writing this, the US Senate held a hearing on the subject where there was a rare bipartisan consensus, not only on the urgency of developing rules on AI development and deployment, but also on the necessity to do so in an international context.
Another area of research and technology that warrants a careful assessment is the invasion of the human brain. During the last decades there have been remarkable advances in detecting, measuring, and interpreting electrical activity in animal and human brains, much of it employing non-invasive technology. It has allowed quadriplegics to speak and, recently, Gert Jan in Lausanne (who was paralyzed from the waist down in a bicycle accident) to walk again. The technology is, in large part, aided by AI: large amounts of brain wave signals are collected and analyzed by AI software to allow these systems to predict what the subject is thinking — which is then fed into a speech generator or into the spinal cord below the break. But there is also a dark side to this. The data collected for a particular purpose can also be used to dig deeper into a subject’s brain and thought processes. Electronic headbands are currently being used to determine truck drivers’ alertness, based on brain wave activity. This detects oncoming drowsiness much earlier than optical systems monitoring eye movements. The system generates lots of data, to be transferred to, and analyzed by, a company’s computer systems. But, if there are no safeguards in place, the data could also be mined to find out what the driver thinks of her boss or whom he voted for in 2022. It is easy to see where this could lead. We have freedom of association, freedom of movement, freedom of speech. What we do not have is explicitly guaranteed freedom of thought. A report by a team led by Dr Ahmed Shaheed , presented to the UN in 2021, offered a framework that includes the right not to reveal one’s thoughts, not to be penalized for one’s thoughts, and not to have one’s thoughts manipulated. That would be a good start.
