Jun 11, 2026 · 55 min · 13 segments
Science depends on more than just results. It depends on researchers asking questions, testing hypotheses, challenging assumptions, and scrutinizing evidence. My guest, Emily Sullivan, Senior…
We need scientists doing a lot of different kinds of things and trying different methods and trying out different lines of inquiry in order to get us all sorts of results, whether we're talking about engineering or chemistry or biology or medicine.
We need scientists doing their thing in a way that respects the scientific method.
She is a senior lecturer of philosophy of science and AI at the University of Edinburgh.
She's an associate editor for the British Journal for the Philosophy of Science, and she leads the Normative Philosophy of Science Lab.
So when she says that AI is infiltrating the scientific life cycle in a way that potentially undermines scientific inquiry, That should probably give us some pause.
Now, as for how it's infiltrating, we'll get to that throughout the conversation.
Some of the things that will hit, though, is things like what does the scientific inquiry lifecycle look like? You know, from trying to think about what should I be researching? How should I be researching it? What method should I use to engage in that research? How should I write it up? All of that, and not just what should I do, but if I'm a reviewer or if I'm a journal publisher, maybe I should have AI reviewing, doing the peer review instead of having individual other scientists do the peer review.
And one, you might worry about this from perspective of bias, hallucinations, that sort of thing.
You also might worry about An algorithmic monoculture, as researcher Katie Creel talks about it from Northeastern.
If you only have a handful of models that are shaping scientific research, you don't have a thousand scientific flowers blooming.
So we talk about that research lifecycle and the way in which AI is infiltrating it.
That said, at the end of the day, what we probably want from a scientific inquiry is good results.
So maybe I care less about how you figure out how to cure cancer than just that you figure out how to cure cancer.
So are we getting wedded to... a sort of, I don't know, conservative approach to science, assuming that we should just use the kinds of methods that we've been using for the past several decades, or what we should really do is open ourselves up to other ways of engaging that inquiry and other ways for that inquiry to be shaped, that is to say, using AI, or is this really a sort of net bad thing to scientific inquiry? So we talk about all that stuff and more.
Read the full transcript.
Create an account to read the whole episode, search across every transcript, and follow the shows you care about.
We need scientists doing a lot of different kinds of things and trying different methods and trying out different lines of inquiry in order to get us all sorts of results, whether we're talking about engineering or chemistry or biology or medicine.
We need scientists doing their thing in a way that respects the scientific method.
She is a senior lecturer of philosophy of science and AI at the University of Edinburgh.
She's an associate editor for the British Journal for the Philosophy of Science, and she leads the Normative Philosophy of Science Lab.
So when she says that AI is infiltrating the scientific life cycle in a way that potentially undermines scientific inquiry, That should probably give us some pause.
Now, as for how it's infiltrating, we'll get to that throughout the conversation.
Some of the things that will hit, though, is things like what does the scientific inquiry lifecycle look like? You know, from trying to think about what should I be researching? How should I be researching it? What method should I use to engage in that research? How should I write it up? All of that, and not just what should I do, but if I'm a reviewer or if I'm a journal publisher, maybe I should have AI reviewing, doing the peer review instead of having individual other scientists do the peer review.
And one, you might worry about this from perspective of bias, hallucinations, that sort of thing.
You also might worry about An algorithmic monoculture, as researcher Katie Creel talks about it from Northeastern.
If you only have a handful of models that are shaping scientific research, you don't have a thousand scientific flowers blooming.
So we talk about that research lifecycle and the way in which AI is infiltrating it.
That said, at the end of the day, what we probably want from a scientific inquiry is good results.
So maybe I care less about how you figure out how to cure cancer than just that you figure out how to cure cancer.
So are we getting wedded to... a sort of, I don't know, conservative approach to science, assuming that we should just use the kinds of methods that we've been using for the past several decades, or what we should really do is open ourselves up to other ways of engaging that inquiry and other ways for that inquiry to be shaped, that is to say, using AI, or is this really a sort of net bad thing to scientific inquiry? So we talk about all that stuff and more.