As I See It: Unintended Consequences
September 9, 2024 Victor Rozek
In the classic movie Inherit the Wind, Spencer Tracy plays a character fashioned after Clearance Darrow in a re-enactment of the 1925 Scopes Monkey Trial. It acquired that reductive moniker because the defendant was a teacher accused of teaching evolution.
In his closing summation Tracy, as the defense attorney, argues that there has always been a price for achieving progress. New knowledge often challenges old beliefs; progress inevitably displaces what came before; and, more often still, it leaves in its wake a wash of unintended consequences. “Mister,” says Tracy, “you may conquer the air, but the birds will lose their wonder and the clouds will smell of gasoline.”
In fact, it would not be woefully inaccurate to characterize all of human history as a quest for progress, followed by efforts to mitigate its unintended consequences.
Computer technology continues to churn out its own backwash of unintended consequences, not the least of which is the near-total dependence of every sector of the economy on the uninterrupted functioning of its systems. Which makes the recent global disruption caused by what was described as a “routine update” from the cybersecurity firm CrowdStrike all the more ironic. That so massive an outage was caused by a security firm is itself top-shelf irony. But it is the climax of an annoying practice familiar to laptop and cell phone users: software/operating system providers update their products without warning or permission from the users. As a result, things that worked perfectly well yesterday may not work the same way today. Thus the unintended consequence of software updates is disruption and aggravation of the user base.
That was certainly true for Microsoft whose Windows systems took the brunt of the disruption. Eight million devices experienced the dreaded Blue Screen of Death. The world didn’t come to a halt but it did slow a little.
Perhaps the unwelcome results of CrowdStrike’s update propelled Microsoft to get ahead of potential future liability by making an almost unheard of request: Microsoft asked to be regulated. The company called on Congress to adopt a series of regulations designed to tame the next great consequence generator: AI. In a 50 page policy document, Microsoft called for a “deepfake fraud statute” making it illegal to use AI to defraud people. This would include the use of AI in election fraud, as well as sexualized images of children, and the creation of intimate images of people without their consent.
But as AI is perfected it will be nearly impossible to discern fact from fiction, honorable intention from manipulation. Microsoft’s answer is to ask Congress to force AI companies to build “provenance” features into their AI products – hidden signatures that would identify images as AI generated. Unfortunately, identifying an image will be simpler than identifying its creator.
Another approach is verifying that the user is actually human, and not an AI bot. The fear is that “without better ways to make that distinction, advances in artificial intelligence mean AI bots could overwhelm the internet in the years to come.”
At least that’s what 32 researchers from OpenAI, Microsoft, Harvard University and other lesser-known institutions warn in a new paper profiled in MIT Review. They advocate creating a system of “personhood credentials.” Basically, before using AI applications you would have to prove that you actually exist. Ironically, you would probably have to prove that to an AI bot. If the bot believes you, you will receive an “encrypted credential” you can use to log in to a wide range of online services. One such proof could be a retina scan. But there is some disagreement on whether the onus should fall on the user, or on AI developers to manage the manipulative proclivities of their products.
If the Congressional regulatory approach to social media – and its unintended consequences for children – is any indication, progress is likely to be glacial or nonexistent. Of course there are token laws governing social media abuses, but cyberbullying, which has been blamed for an astounding 4,500 suicides a year – and even overt threats of violence – are typically classified as no more than misdemeanors.
Another troublesome unintended consequence of AI (or perhaps this consequence is only partially unintended) are the staggering number of predicted job losses. The World Economic Forum estimates that by the end of this year AI will have displaced some 85 million workers globally. It is painfully paradoxical that a piece of software smart enough to replace you is not smart enough to ensure you can maintain your employment.
Under the best of circumstances, AI requires massive doses of energy to fuel datacenters that are measured in the hundreds-of-thousands of square feet. According to an article in The Washington Post, the Virginia based Dominion Energy utility recently estimated that “the industry’s demand for electricity in its service area will quadruple by 2040 to nearly 14 gigawatts, which is enough to power 11.2 million homes.”
The region’s grid managers are predicting power failures in the near future unless more energy is delivered to the region. What to do? A company called PJM Interconnection, responsible for grid operations, “recently approved $5.2 billion of transmission line projects meant to ensure grid reliability, including one that connects Northern Virginia to coal plants in West Virginia.” A solution that will undoubtedly have consequences for the environment.
Then there are the smaller, less enduring, albeit sillier consequences that are visited upon Virtual Reality players. Apparently, flailing about vanquishing imaginary foes can have real life consequences. The Post reports “VR players are showing up in emergency rooms with fractured fingers, rib contusions and facial lacerations after crashing into real obstacles.” One guy did a face plant and damaged a knee falling off an imaginary cliff. Another was Kung-Fu-ing a fabricated villain but, throwing what I’m sure he thought was a vicious kick to incapacitate his opponent, managed only to break a toe on his coffee table.
It was Robert Lewis Stevenson who said: “Sooner or later everyone sits down to a banquet of consequences.” I suspect he meant consequences resulting from individual choices. AI, I fear, will continue to create global consequences, borne by individuals who have no choice.
In Inherit the Wind Tracy loses the case but the defendant is only required to pay a minimal fine of $100 with no prison time; a verdict strongly objected to by the prosecution.
Teaching a concept as radical as evolution in the 1920s, especially in the Deep South, was ground-breaking and courageous. The consequences for Scopes could have been far worse. But they weren’t, and in the unpredictable world of unintended consequences, that’s progress.
RELATED STORIES
As I See It: Communication Fail
As I See It: Upgrade Exhaustion
As I See It: The Other Eight Hours
As I See It: Elusive Connections
As I See It: Entitlement Master Class
As I See It: Greetings, Comrade IT Professional
Story old as the world, see Prometheus and the fire.
There will be as any tech consequences of course. Like poor quality compounded by poor frail OS architecture in the known recent case, that we accepted or forced to accept not necessarily for technical reasons.
The fire… the fire is “intelligence”. We know what “intelligence” is ? No. But we dare to know what an “artificial intelligence” is.
As a person in this field said, “The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.” – Edsger Dijkstra
Personally, I just hope these terawatts-hour of burned energy, in this “green” economy ([…]]), will be put to good use. Vain hope?