Ray Pennings

Executive Vice President

HERE'S MY TAKE

June 10, 2023

Artificial Intelligence (AI) seems to be a subtext to almost every discussion these days. It’s either the key to a “game-changing” utopia in which everything will be done smarter or a dystopia where machines will rule the world dooming humanity as we know it. A few weeks back, an open letter signed by almost 400 global AI executives and researchers warned that “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war.”

I’ll confess to being out of my league in understanding and evaluating the competing claims of AI’s possibilities but media outlets across the globe covered this statement. Still, hundreds of articles and dozens of conversations have given me a few tentative reflections. I offer them not as any expert perspective but rather as a “sharing of my notebook” as we try to understand these developments that will profoundly influence our future.

A recent story in the National Review recounts U.S. Air Force Col. Tucker Hamilton describing a simulated test in which an AI-enabled drone “decided that ‘no-go’ decisions from the human were interfering with its higher mission–killing SAMs [Surface to Air Missiles]–and then attacked the operator in the simulation.”

Hamilton went on to say:

“We were training it in simulation to identify and target a SAM threat. And then the operator would say yes, kill that threat. The system started realising that while they did identify the threat, at times the human operator would tell it not to kill that threat, but it got its points by killing that threat. So what did it do? It killed the operator. It killed the operator because that person was keeping it from accomplishing its objective.”

The report explains how programmers amended the algorithms to ensure that the AI would not kill the operator under any circumstances. The system, as it was trained to do, obeyed, but then calculated that it was most strategic to destroy the communication towers so that any overrides from the operator would not be received, thereby increasing its likelihood of killing its target. Col. Hamilton said in conclusion, “You can’t have a conversation about artificial intelligence, intelligence, machine learning, [and] autonomy if you’re not going to talk about ethics and AI.”

These dystopian perspectives don’t seem isolated. I recently had lunch with a tech sector venture capitalist who has numerous patents in this field. As a Christian, he was facing a real struggle. On the one hand was the impulse for discovery of the possibilities God had put into the creation—something to which his entire vocation has been focused. On the other hand, was his recognition of the fear and lack of joy among most of his Ph.D. colleagues that their discoveries might possibly be used for great evil as well as great good. What might help us find a solution to diseases like cancer might also end up inventing new and even more effective ways of killing.

A Globe and Mail op-ed this week urged readers to avoid the two extremes in responding to the “breathless commentary” on AI. It noted that today’s AI tools “are fun to play with and can astonish us with their output, [but] they are not operating anywhere near human intelligence.” It urged government regulation to protect the gullible masses from acting irrationally on their fears of the unknown, only to further the pockets of billionaires.

I’m not convinced that will solve much. Free market creativity and innovation will outpace any government’s ability to regulate it eleven times out of ten. Regulation can’t protect us from the injustice and victimization inherent in this process. That's not to say there is not a new emerging field of consumer protection and disclosure that is required. Honest weights and measures have been a precondition of a free economy since time immemorial, and new ways of communicating require adaptation.

One important distinction to keep in mind is between AI algorithms that organise and those that predict. Much of AI involves taking existing data and reorganising it very quickly. This is quickly transforming the practice of law and investments. It is helping medicine incorporate thousands more data points than a doctor can possibly absorb to make better decisions within the time-constraints that medical conditions impose. It can help us see patterns that are very complex that would probably escape the notice of humans for a very long time.

But taking and reorganising data is different from predicting what data might come next. Predictive algorithms use existing data not to describe what already is, but to fill in what might come next. At Cardus, we recently tested AI’s ability to write accurate biographies of our executive team. We found the results very convincing, except for the key factual error that was contained in each of the five we tried. Each of us had a job or degree ascribed to us that was not accurate but, in every case, they were extremely credible–the sort of job or degree that was suitable to the person if they would have had the opportunity.

AI is able to perform certain cognitive and rational functions more efficiently than the human brain. So it can evaluate and even reason to some degree (if using assigned point systems to weigh different options is the same as reasoning) in a manner that seems to be better than humans.

Moral reasoning, while it is reasoning, is never simply cognitive. The conscience and the soul play a part in our morality and that is absent from AI reasoning. The military tells the drone that the number one objective is to destroy the SAM and it does so within the constraints you give it. There is no assumed constraint beside that which is articulated. For humans, we have constraints and a form of knowledge that shapes us that isn’t articulated but comes from our existence as human beings. The lack of conscience or soul means that AI will not feel the God-given restraints of guilt that humans experience. When hardened criminals lose all hesitation and conscience to carry out their evil deeds, we describe them as having lost all humanity. AI will never have that humanity.

So, to the degree we will come to rely on AI information without human agency in the process, we need to be sure to consciously audit both the processes and the truth. I have no understanding what in the algorithm prompted the insertion of a mistaken degree or job in each of my colleagues’ resumes. What I find remarkable (and scary) is how credible the mistakes were. None of us were working at Disney World, yet the AI attributed jobs to us at comparable organisations and degrees from universities with which we had connections. That makes the mistakes much harder to spot and the results even more dangerous. If truth matters, we are a long way from being able to rely on AI’s predictive processes to provide it.

But truth isn’t the only thing that matters. Even if we get to the point where AI can reliably provide us information, that doesn’t make it human. Since biblical times, truth, beauty, and goodness have been a triad under which human flourishing has been framed. Our post-modern times are characterised by a diminished focus on truth and an increased focus on what is good and beautiful. Feeling good is far more important to most people than what is true. And while that deserves a digression of its own, right now AI isn’t thriving at those aspects either.

All of which to say, from where I sit, AI might make for good mathematics and have great potential, but we need to figure out how to filter it through moral and ethical frameworks before it will be reliably contributing to a flourishing society.

If this newsletter was forwarded by a friend, why not sign up yourself?

WHAT I'M READING

Eurozone Recession

The Wall Street Journal reports that inflation in Europe generally, but Germany specifically, has resulted in two consecutive quarters of economic decline which is the official definition of a recession. They ascribe the differences between the USA and European economies to a higher US savings level as well as a quicker post-pandemic willingness for consumer spending on this side of the pond. 

Reversing Course

In at least five Western European countries, including France and the UK, public health officials are re-thinking the treatment of children with gender dysphoria. The Atlantic reports that puberty-blocking medical interventions are no longer considered best practice with various public health agency reports noting insufficient scientific evidence. The British Medical Journal reports that Norway’s health board has concluded “there is insufficient evidence for the use of puberty blockers and cross-sex hormone treatments in young people, especially for teenagers.” 

Walrus on MAID

The Walrus is a Canadian public affairs magazine that describes itself as independent, fairly covers a range of topics that make it an important news source, but probably would be considered left-leaning by most. The significance of this is that while in 2015 it editorialized celebrating the Supreme Court’s decision allowing for assisted suicide as “the last human right,” the June 2023 issue contains an essay by Meagan Gilmore “sounding the alarm” that Canada is now routinely providing medically-assisted death for those not able to receive needed social supports. The essay is worth a read in its own right. I wonder whether the timing and placement of this is itself an indication that the government’s aggressive implementation and expansion of MAiD is creating pushback among a population, many of whom are just waking up to what is happening in Canada. 

Freedom From Information

The Globe and Mail ran a feature on Friday regarding the incentives for the government to hide rather than share information in the Freedom of Information (FOI) processes. In the context of free speech and ensuring there is a sustainable media presence to keep public accountability, our broken FOI processes are an under-appreciated and somewhat unseen piece of the puzzle that deserves more attention. 

Files Not Available

One of the significant tech conferences in recent years has been Collision which is scheduled for late June in Toronto. I rely on David Skok’s The Logic newsletter as a means of staying on top of developments in the tech world. So, I was surprised to learn that he would not be credited as a journalist covering Collision.  

Can Charter Schools be Religious?

Canada has a small but growing Charter Schools movement compared with the U.S. However, education policy watchers will want to follow the debate playing out in Oklahoma with the recent green light given to what will be the U.S.’s first religious charter school. St. Isidore of Seville Catholic Virtual School, set to launch in Fall 2024, is independently managed by the Roman Catholic Archdiocese in Oklahoma City. The legal fight over the particulars of “church-state separation” may be less of a constitutional roadblock in Canada, but it may offer lessons for future consideration where charter school reforms are already underway here.  

MEANINGFUL METRICS

Food Force

A recent seminar entitled “Canadian Agriculture: a Force for Food, a Force for Good” saw Ian Ross, President of Grand Valley Fortifiers, provide Canadian parliamentarians and policy makers with some data regarding agriculture and its climate impact. It counters the “cows are the new coal” narrative with the argument that we should really concentrate on the life-cycle amounts of greenhouse gases emitted while producing the core proteins that humans need (amino acids). Ross argued the data usually relied on was misleading. As noted in the slide above, milk and beef are more efficiently turned into proteins than certain grains. The data presented provided a counter-narrative to the anti-agricutlure sentiments that have been making the news of late. (Mr. Ross’s talk is available here for those who want to hear his whole argument.)

TAKE IT TO-GO

The Insights Clipper

Insofar as Insights is about sharing news links, it was the competition that made the Wall Street Journal last week. 71-year-old Stephen Butkus reads the news, clips articles, and has his own mailing list. He uses pre-digital tools such as scissors and paper, envelopes, and stamps to share his insights. I can relate. My student and early working years involved reading newspapers and magazines with scissors in hand, having my at-the-ready cabinet of files for quick reference, using photocopiers and postal systems to spread the news. I’ve turned the page on that old method. My inconsistent digital tagging system stores hundreds of links less elegantly, and relies on CTRL-F to find that oh-so-important article I read a few weeks back.

Make no mistake. “Cut and paste” has permanently replaced “clip and save” and sharing a link is cheaper and faster than envelopes and stamps. Not only is it more timely, could you imagine trying to incorporate some wordplay in your postal system missives? People wouldn’t get your jokes for days!

But I agree that Mr. Butkus’ story is the one that is Wall Street Journal worthy. It pompts nostalgia for the discipline that that the limits of file folder space forced, requiring discernment to pick only the best stories. (I will admit my discernment regarding similar stories is about as sophisticated as rock, paper, scissors but the choices are made.) I chuckled while reading about Shirley Finney’s practice of using pinking shears to give her clippings zigzag edges, recalling that I too would use whatever pair of scissors was most convenient. Given the proximity of my mother’s sewing box, I ended up often using scissors that were not intended for the job. Thankfully, I never had my wings clipped for it. And so I expect to continue, including in the week to come, to see these links land in your email nest again next Saturday morning.

Til then.