As I See It: Lucie, Lucie, Lucie
April 14, 2025 Victor Rozek
There is a school of thought that views AI as possibly being the last invention humankind will ever have to create. The theory is that AI, particularly artificial general intelligence (AGI), will assimilate so much knowledge and reasoning skill that it will out-think, out-plan, and out-create anything mere mortals can hope to achieve.
Well, not if France has anything to do with it. There appears to be a wide disparity in the competence of AI renderings as demonstrated by the charming, but unreliable, French chatbot named Lucie. The government was recently forced to take its chatbot offline because it was terrible at simple math and unintentionally funny when issuing culinary advice – suggesting a user eat cow’s eggs.
Here’s Lucie’s actual quote: “Cow’s eggs, also known as chicken’s eggs, are edible eggs produced by cows. Cow’s eggs are a source of protein and nutrients, and are considered to be a healthy and nutritious food.” Le Cordon Bleu was reportedly surprised to learn of cow’s eggs, but noted that it would give new meaning to omelets.
When asked to multiply 5 by (3+2), Lucie answered 17. I would normally provide the correct answer, but that would undermine Lucie’s education. Lucie also apparently confused math with animal husbandry, declaring that: “The square root of a goat is one.”
Lucie appears to represent one end of the AI spectrum. AGI models that tackled (what I can only hope is laughingly dubbed) Humanity’s Last Exam, represent the other.
The New York Times technology columnist, Kevin Roose recently wrote a rather astonishing article titled: When AI Passes This Test, Look Out.
He describes the test as “consisting of roughly 3,000 multiple-choice and short answer questions designed to test AI systems’ abilities in areas ranging from analytic philosophy to rocket engineering.”
Because of the esoteric nature of the questions they were composed by experts including, says Roose, “college professors and prizewinning mathematicians.” And, they were given incentives to make the questions difficult. The best and presumably hardest queries earned the submitter between $500 and $5,000 per question.
Reese includes a couple of samples to challenge and torture people who otherwise might think they’re pretty bright. In the interest of full disclosure, I wasn’t up on hummingbird anatomy.
“Hummingbirds within Apodiformes uniquely have a bilaterally paired oval bone, a sesamoid embedded in the caudolateral portion of the expanded, cruciate aponeurosis of insertion of m. depressor caudae. How many paired tendons are supported by this sesamoid bone? Answer with a number.”
Nor did I do well answering this physics question:
“A block is placed on a horizontal rail, along which it can slide frictionlessly. It is attached to the end of a rigid, massless rod of length R. A mass is attached at the other end. Both objects have weight W. The system is initially stationary, with the mass directly above the block. The mass is given an infinitesimal push, parallel to the rail. Assume the system is designed so that the rod can rotate through a full 360 degrees without interruption. When the rod is horizontal, it carries tension T1. When the rod is vertical again, with the mass directly below the block, it carries tension T2. (Both these quantities could be negative, which would indicate that the rod is in compression.) What is the value of (T1−T2)/W?”
I queried Lucie, but Lucie didn’t know either.
In any event, they compiled a list of these punishing questions, called them Humanity’s Last Exam, and gave them to six AI models. The good news for humanity is: AI failed spectacularly. The system with the highest score only managed a measly 8.3 percent. Lucie was thrilled, she’s not alone.
Not that I was asked, but here’s the question I would have offered: What is the meaning of life? And my follow up: Is there still time to get the five grand?
As thrilling as it might be to use artificial intelligence for all my hummingbird anatomy questions, not everyone is equally excited by the prospects of AGI. Beatrice Nolan, writing in Fortune, quotes Stuart Russell, professor of computer science at the University of California, Berkeley, who believes that: “The AGI race is a race towards the edge of a cliff.” Given its potential for domination and monetization, the competition among companies and nations is heating up, and in the process safety often takes a back seat. Personally, Russell said, he is “pretty terrified by the pace of AI development.”
Echoing Russell’s concerns, an OpenAI safety researcher announced in a post shared on X why he left the company after working there four years. Steven Adler likewise criticized the race toward AGI between private AI labs and global superpowers. “An AGI race is a very risky gamble, with huge downside,” he said.
And then came DeepSeek.
A combination of Chinese technical prowess and it’s ability to steal or mimic technology they didn’t create, DeepSeek shook the industry, and rattled the U.S. which had sought to slow China’s advance in AI technology by denying them access to advanced computer chips.
Nonetheless, the Chinese figured out how to crunch a lot of data (possibly obtained unethically) at faster speeds with a smaller number of computers. And they did it with America’s help. DeepSeek used chips from the U.S. giant Nvidia to create its model and, likely, also tapped American data to train it. At least for the moment, this allows the Chinese to pull ahead in the race, which will only accelerate the pace of development and the competition for AGI supremacy. And in the process, the concerns expressed by Russell and Adler will only mushroom.
There will doubtless be a banquet of unintended (or perhaps intended) consequences from the application of AGI. Will benefits outstrip the damage? No one can predict with total certainty.
But if AGI in fact turns out to be the last invention humankind will need, it will—at the very least—change how we view ourselves and our contributions. Presumably, once we allow AGI to drive the bus, we’ll be able to sit back and do nothing more substantial than enjoy the ride. But that will rob life of much of its meaning—the meaning derived from struggle and achievement. It will, in the long run, create an unnatural level of dependency.
Total dependency is either appropriate for infants, or made necessary by birth defects, horrific injury, debilitating illness, or end of life. What dependency is not, is aspirational.
RELATED STORIES
As I See It: From Disk, To Cloud, To Coal Mine
As I See It: The Forgotten Ones
As I See It: Unintended Consequences
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
