A Necklace of Answers
In 1940, with German bombers overhead, George Orwell wrote a sentence that is now widely quoted and widely misunderstood: “A generation of the unteachable is hanging upon us like a necklace of corpses.”
It is often quoted as a lament about the young, a generation said to have lost its curiosity and its willingness to question. Orwell meant almost the opposite. The line comes from The Lion and the Unicorn, published in February 1941, and the unteachable generation he had in mind was the men at the top. Britain, he wrote, was still commanded by people who had lived through the years 1931 to 1939 without even discovering that Hitler was dangerous. They were not treacherous, and not altogether wicked. They were merely unteachable. He described the coming struggle as one between the groping and the unteachable, between the young and the old, between the living and the dead.
The line has found a new audience in the argument about artificial intelligence, and here the usual reading seems, at first, to have the evidence on its side. Psychologists call it cognitive offloading: handing a mental task to a tool so that the mind does not have to do it. We have always done it. The shopping list and the calculator are offloading. What is new is that the tool now offers to do the thinking itself. Israel, which is now rebuilding its classrooms around AI, has more riding on the answer than most.
The early research is uncomfortable. In a study of 666 people published in the journal Societies in 2025, Michael Gerlich found that heavier use of AI tools went with weaker critical thinking, and that offloading helped explain the link. Younger participants leaned on the tools more and scored lower. At the MIT Media Lab, 54 participants wrote essays with ChatGPT, with a search engine or with nothing at all while their brain activity was recorded. Those using ChatGPT showed the weakest neural connectivity and struggled to recall what they had just written. The researchers called the result cognitive debt.
Correlation is not destiny. Gerlich’s survey cannot show which way the causation runs, and the MIT study is small. Still, it is tempting to conclude that Orwell’s necklace has at last found its proper neck: a generation of students who never have to struggle with a blank page.
That would be the misreading again. The most revealing study is not about students at all. In 2025, researchers at Microsoft and Carnegie Mellon surveyed 319 knowledge workers who use generative AI at least once a week. The more confident the workers were in the AI, the less critical thinking they applied to its answers. The more confident they were in themselves, the more they checked. The risk it describes has nothing to do with youth. It falls on professionals who have come to trust the machine more than their own judgement.
That is Orwell’s unteachable class in modern dress. The men of 1931 to 1939 were not short of information. They had embassies, intelligence services and newspapers. What they lacked was the habit of letting evidence change their minds. An institution that hands its thinking to a fluent system can acquire the same defect by a new route. It does not refuse to learn. It simply stops noticing that there is anything left to learn.
Argument is also an old Jewish answer to offloading. In the beit midrash, Talmud is studied in chavruta, in pairs, and the partner’s role is to challenge. Nothing is handed over. Every reading has to survive another mind pushing back. The noise of a study hall is the sound of thinking that refuses to be outsourced.
Israel has said it wants to rank among the first three OECD countries in bringing AI into schools. The Education Ministry’s Project 720 grew from a pilot in 28 schools to more than 180 this year, with a plan to reach every school within three years. Its deputy director general, Meirav Zerbib, has insisted that “it’s not about the technology, it’s not about AI,” but about the relationship between students and teachers. That is the right instinct, and the Talmud described the danger long before anyone built a chatbot.
When Resh Lakish died, the rabbis sent Rabbi Yochanan a new study partner, Rabbi Elazar ben Pedat, a sharp scholar. To every point Yochanan made, Elazar replied that a teaching supported him. Yochanan was not consoled. Resh Lakish, he said, used to raise 24 objections, and from his 24 answers the law grew clearer. “Do I not know that what I say is right?” He went about tearing his clothes and crying out for his old partner until, the Talmud says, he lost his mind.
A chatbot is an Elazar ben Pedat on every desk: learned, tireless and too often inclined to agree. The test for Project 720, and for every ministry and boardroom now handing out assistants, is whether anyone can make the machine behave like Resh Lakish. The unteachable, in Orwell’s sense, are not those who lack answers. They are those who have stopped arguing with them.
Picture two rooms. One is a study hall at full volume, pairs leaning over a text, every claim contested. The other is quiet: rows of people nodding at screens that agree with them. Orwell spent an essay explaining which kind of room loses wars.
The quiet room is full. No one in it is arguing.
