The Five Stages of AI Grief: Dani Rodrik on Coming to Terms with Artificial Intelligence
Key Takeaways
- •Rodrik frames public reaction to AI through Kübler-Ross's five stages of grief—denial, anger, bargaining, depression, and acceptance—first outlined in her 1969 book 'On Death and Dying.'
- •Early anger toward AI was fueled by hallucinated case references in lawyers' briefs, flawed software code, malfunctioning customer-service bots, and AI-written student assignments.
- •Harvard economist Jason Furman, who chaired the Council of Economic Advisers from 2013 to 2017, has claimed AI tools would outperform professionally trained economists in staff roles at the White House CEA.
- •When tested with a hypothetical 2027 home-care crisis, Rodrik's chatbot produced a technical memo recommending federal matching grants, Medicaid pay adjustments, and a caregiver visa program, plus creative options such as a National Care Corps and 'care districts' in unused office and mall space.
- •Rodrik likens AI to performance-enhancing drugs whose harmful side effects arrive much later, citing the European Union's AI Act—effective in 2024 and being phased in over several years—as an example of the regulation and restraint he believes is needed.

Dani Rodrik
CAMBRIDGE — The Swiss-American psychiatrist Elisabeth Kübler-Ross famously observed that people receiving a terminal diagnosis pass through five stages of grief: denial, anger, bargaining, depression, and acceptance — a framework she introduced in her 1969 book "On Death and Dying" that has since traveled far beyond its clinical origins. The same sequence, it turns out, describes our collective response to the rapidly expanding capabilities of artificial intelligence (AI).
The denial stage came first. For a while, AI looked like little more than a glorified search engine, or a chatbot capable of simulating a human being decently enough. On that view, it was easy to dismiss the technology as something that would leave our lives and professions largely untouched.
Then AI began cropping up everywhere, and often in the most infuriating ways. It showed up in lawyers' briefs padded with hallucinated case references, in software code marred by mistakes and security vulnerabilities, in customer-service bots that went off script, and — a particular source of irritation for academics like me — in student essays and assignments. Each of these was reason enough to get mad.
But the technology got better. Fabrications, hallucinations, and mistakes grew rarer. Once it became obvious that students' use of AI was unstoppable, many began to see an upside in the technology's potential to enhance learning. It was time to bargain. Yes, we would need to make all kinds of adjustments to how we worked, we told ourselves, but inherent limits remained. AI could shoulder a vast amount of cognitive and analytical work, and far faster than any human; yet questions of judgment, taste, values, and preferences still belonged — and would continue to belong — to the human mind. AI could execute, but humans would imagine. The task was simply to get the division of labor right.
I clung to that view for quite some time, and I confess it gave me much comfort. AI, however, was not playing ball. Every time I asked my chatbot a question, it came back not only with answers but with much more: it told me what else I should be looking for, suggested a different way of framing my approach, and drew parallels or contrasts with other projects I had worked on — regardless of whether I had consulted AI on them. It was somewhat annoying, yet I could not deny that I was receiving valuable insights. I would like to think I would have arrived at similar thoughts on my own in due time, but my chatbot regularly preempted me. Worse still, it was clearly capable of being very imaginative.
My colleague Jason Furman, the Harvard economist who chaired the Council of Economic Advisers under President Barack Obama from 2013 to 2017, has claimed that AI tools would do a better job today than professionally trained economists in staff roles at the White House Council of Economic Advisers. Inspired by his claim, I tested my chatbot with a difficult question: "It is 2027, and the country is faced with a crisis in care. There is a shortage of home care aides, accidents and deaths are rising among elderly and disabled people left without help, and workers are dropping out of the labor market to look after loved ones. The CEA Chair needs options to present to the President."
I hit enter, and the AI promptly prepared a technical memo with detailed recommendations covering federal matching grants, adjustments to Medicaid pay, and a caregiver visa program, complete with a fiscal accounting. Exceeding its authority — and, characteristically, the prompt itself — it also outlined a political strategy for selling the program.
Hmm, I thought, my chatbot is clearly very good at operating within the prevailing economic and institutional context. But could it move beyond it? To find out, I asked it to develop out-of-the-box policy alternatives that disregarded political realities. It returned some interesting possibilities, including a National Care Corps that would draft young people to care for the old, and converting underutilized office and mall space into "care districts" that would benefit from scale and agglomeration economies.
Then, on a whim, I asked AI to turn one of my manuscripts into a stand-up comedy routine. What came back made me laugh so hard that I worried passersby might fear for my mental health. Insight, imagination, humor! This was not part of our bargain, was it?
That is when the depression began to set in. I started to feel sadness about a world in which human thought would no longer be distinct from machine work — and would be needed less day by day. Like the mathematicians who signed a joint letter on the impact of AI on their field, I became alarmed about the loss of "conceptual understanding and insight" that would follow from letting AI take over our work. It was frightening and disorienting.
The next stage, the manual tells us, is acceptance. Here, we do have a choice. Unlike a terminal illness, AI and its uses are not outside our control. From the Silicon Valley firms developing the technology and the businesses deploying it, to the professional communities setting the norms of appropriate use and the governments that must one day — better sooner than later — regulate it (the European Union's AI Act, which took effect in 2024, is being phased in over several years), decisions about our relationship to AI are being made every day. The worst mistake we can make is to accept a world in which we cede too much to it.
Ultimately, we will have to figure out how to restrain the use of AI for the good of humanity. The appropriate analogy is performance-enhancing drugs, which temporarily increase human capabilities but eventually harm their users. AI is similar, except that it is far more powerful, and its commensurately disastrous side effects will show up much later. The temptation to indulge in it freely is, and will remain, very strong.
Homer's Odysseus knew that the Sirens promised knowledge but delivered disaster. He had himself tied to the mast of his ship so that he would not succumb to their beautiful song and steer toward them and the cliffs that had claimed so many mariners. The challenge we face with AI is similar: to restrain ourselves from losing our humanity.
Dani Rodrik, professor of international political economy at Harvard Kennedy School, is past president of the International Economic Association and the author of "Shared Prosperity in a Fractured World: A New Economics for the Middle Class, the Global Poor, and Our Climate" (Princeton University Press, 2025). This article was distributed by Project Syndicate.