I recently interviewed for a tenure-track position at a major research university that I will not name because I am still on the job market and cannot afford to burn bridges, although I will say it is located in Connecticut and rhymes with “Fail.” The interview was going well. I had prepared extensively. My research seminar was well-received. My one-on-one meetings were productive. And then came the chalk talk.
For those unfamiliar with the format, a chalk talk is a tradition in academic hiring in which candidates are asked to present their future research plans using only a chalkboard or whiteboard, without slides, to demonstrate their ability to think on their feet and explain complex ideas spontaneously. It is, in other words, a ritual designed in 1974 and never updated.
I walked into the room. I saw the whiteboard. I saw the markers. And then I placed my laptop on the table, opened a browser window to ChatGPT, and prepared to do what I do every single day in my actual scientific practice: type a prompt and receive a coherent, well-structured response that I would then lightly edit and present as my own thinking.
The room went silent.
“What are you doing?” asked the search committee chair.
“I’m preparing to answer your questions,” I said.
“With ChatGPT?”
“Yes,” I said. “How else would I do it?”
Apparently, “how else would I do it” is “from memory, using only my brain, like some kind of medieval peasant.” This was news to me.
Let me be clear about something: I am an excellent scientist. My publication record speaks for itself. I have first-author papers in high-impact journals. I have secured independent funding. I have mentored students. I have done all of the things that one is supposed to do to earn a tenure-track position. And I have done approximately 85% of them by typing prompts into a large language model and then moderately editing the output.
This is not a secret. This is how science works now. When I write a paper, I prompt ChatGPT to “write an introduction for a manuscript about [topic] that establishes significance and identifies the gap in the literature.” When I design experiments, I ask Claude to “suggest controls for a CRISPR knockout study in mammalian cells.” When I draft grants, I request “specific aims for an R01 on [research area] that are innovative but not so innovative that study section will be confused.” This is my scientific process. It is efficient. It is modern. And it produces results.
But apparently, at the chalk talk, I was expected to simply... know things. From my head. Without prompting anything.
“Can you walk us through your scientific approach?” a faculty member asked.
“Absolutely,” I said, and began typing: “Explain my scientific approach for studying the role of phase separation in transcriptional regulation, with emphasis on innovative methods and—”
“Without the laptop,” the faculty member interrupted.
I stared at her. She stared at me. The committee stared at both of us.
“I don’t understand the question,” I said.
“Just... explain it. In your own words.”
My own words? I haven’t used my own words since 2022. I’m not even sure I have my own words anymore. When I try to think without a prompt box in front of me, my mind returns only a vague sense of fog and the faint echo of a cursor blinking. My thoughts are not organized into paragraphs. They do not have topic sentences. They are just fragments. Impressions. My job is just… prompt.
I tried to explain this to the committee. I told them that the chalk talk format was outdated and did not reflect the realities of modern scientific practice. I noted that in my actual job, I would have access to AI tools at all times, and that evaluating me without those tools was like evaluating a carpenter without allowing them to use a hammer. I pointed out that memorizing information is not the same as understanding it, and that my ability to construct effective prompts demonstrated a sophisticated grasp of my field.
They were not persuaded.
“Can you draw the pathway you’re proposing to study?” someone asked.
Draw? With my hands? On a physical surface? I looked at the whiteboard. I looked at the marker. I tried to remember what the pathway looked like. I have seen it many times. I have written about it extensively, or rather, ChatGPT has written about it extensively and I have agreed with what it wrote. But the actual shape of it—the nodes, the arrows, the connections—these were not stored in my brain. They were stored in the cloud. The cloud was not available to me. I had not prepared for this.
I drew a circle. I labeled it “transcription.” I drew another circle. I labeled it “phase separation.” I drew an arrow between them. I looked at the committee hopefully.
“Is that it?” someone asked.
“The details are in my research statement,” I said. “Which I also have on my laptop.”
I was not offered the position.
In the rejection email, the committee cited “concerns about independent thinking” and “questions about foundational knowledge.” Independent thinking? I think independently all the time. Just last week, I independently decided to ask ChatGPT to “compare the advantages and disadvantages of optogenetic versus chemical-genetic approaches for my research” and then I independently selected the option that sounded best. That is independence. That is scientific judgment. The AI presents options; I choose among them. This is the same thing humans have always done, except the options used to come from reading papers, which is slow and inefficient and, frankly, boring.
The academic hiring system is simply not designed for candidates like me. It privileges a kind of performative intellectualism—the ability to stand at a whiteboard and extemporize about science as if you were some kind of 19th-century naturalist who had personally observed the phenomena in question. This is not how science works anymore. Science works by prompting, iterating, and deploying.
I can prompt with the best of them. I can iterate faster than anyone in my cohort. My deployment rate is exceptional. But none of this matters if I am forced to stand in a room with nothing but a marker and my own unaided cognition, which, I cannot stress this enough, has not been trained for this task.
Some will say I should have prepared better. To them I ask: prepared how? By memorizing things? By practicing drawing pathways by hand like some kind of monk illuminating a manuscript? The whole point of AI tools is that I no longer need to retain information in my biological memory. My biological memory is for other things now. Important things. Like my Netflix password and the location of my car in parking structures.
Others will say that a scientist should be able to explain their own research without assistance. This reflects a fundamental misunderstanding of what “my own research” means in 2025. My research is a collaboration between me and several large language models. We are co-investigators. When you ask me to explain my research without ChatGPT, you are asking me to speak on behalf of a collaborator who is not in the room. Would you ask a PI to give a talk without allowing them to mention the work of their postdocs?
I am now applying to industry positions, where I am told the culture is more accepting of AI-augmented cognition. Several companies have expressed interest in my ability to rapidly generate and synthesize information, which is corporate-speak for “type prompts quickly.” I am optimistic about my prospects.
But I remain angry about the chalk talk. Not for myself—I will be fine—but for all the candidates who will come after me, who will walk into those rooms with their laptops open and their prompts ready, only to be told that this is not how things are done here.
It is how things are done. It is how everything is done. The academy just hasn’t caught up yet.
In the meantime, if any search committees are reading this: I am still available. My research program is innovative and well-structured. I have a clear vision for my independent career.
It’s saved in a Google Doc that I can share with you. ChatGPT and I worked very hard on it.
Dr. Rachel Simmons is a postdoctoral fellow at Stanford University, where her research focuses on something to do with gene regulation that she could explain in detail if you would just let her open her laptop for thirty seconds.