I. Glenn Cohen is the James A. Attwood and Leslie Williams Professor of Law at Harvard Law School, as well as Deputy Dean and Faculty Director, Petrie-Flom Center for Health Law Policy, Biotechnology & Bioethics. He’s one of the world’s leading experts on the intersection of bioethics and the law, as well as health law, and a prolific contributor to working papers on SSRN. In a wide ranging interview with SSRN he shared his thoughts on some fascinating issues at the intersection of health and law, from the recent Alabama Supreme Court ruling on the legal status of embryos, to the ethical and legal issues raised by AI, and some extraordinary new developments in reproductive science.
Q: For people who may not be familiar with the term, what does it mean to be a bioethicist?
A: Like many fields the borders are porous and precise definitions contested, but in general I think a fair definition of bioethics is that it “the interdisciplinary study of ethical issues arising in the life sciences, health care, and health science and policy.” It looks as the spaces where the medical system raises ethical questions – questions of right or wrong – which blend into questions of policy. Bioethics had its roots in religious studies and philosophy and medical practice, so at the start of the field (dated by most people to the 1960s) most people came in from that tradition. Over time more law people, more medical people, empirical people, political science, sociology, and critical studies approaches have come in, among others, so it’s now a pretty interdisciplinary field.
The field studies everything from clinical medical decision making, which, to policy questions about rationing COVID-19 vaccines, to questions about the scientific process, questions about the ethical conduct of human clinical trials and even whether certain kinds of diseases ought to get priority in terms of being researched.
Q: What issue did the recent Alabama ruling on embryos raise for you, and do you think it could lead to a federal ban on abortion in the US?
A: As it happens, I’ve got a preprint relevant to this on SSRN for a book that came out this week, I wrote the paper maybe a year and a half ago, and I will hopefully have a paper coming out shortly, probably after this is published, on the Alabama case specifically, so it’s something I’ve thought about a lot.
There is a narrow legal question posed by the Alabama case, and a broader, cultural philosophical question.
The narrow legal question is whether, under the statute that provides for tort liability for the wrongful death of a minor – which allows people to sue someone who causes the wrongful death of a minor, – whether that statute is used for the word minor, or child, or person, does it apply only to children who were born, or does it apply to embryos that are being stored in a freezer?
The bigger question is part of the context of a larger set of political forces and philosophical issues. The political forces have been, I think, galvanised by the reversal of Roe v. Wade in the United States, which provided an abortion right protected by the Federal constitution. In the Dobbs case, the Supreme Court said, “No, we’re going to say the constitution does not provide such a right, and that States are now free to make policy in this area.” With that kind of victory at hand, the anti-abortion movement, or portions of it, has set their sight on a new goal which is not to return the matter to the States, but instead to have a nationwide ban on abortion.
One way of achieving that is to get the recognition, ideally at the level of the Federal constitution, that foetuses, and perhaps now embryos, are persons – such that their destruction violates the constitution, and such that a state action that permits the destruction of foetuses – even in states that are Democrat-leaning like Massachusetts where I live – would not be able to have an abortion protective law in place. That’s the political arc, and decisions like this are an important piece of it.
The philosophical question is a difficult one, which is to say: when does personhood begin? That’s just a different question than the question, “When does life begin?” or “When does being a human being begin?” One could have different answers to all three questions, and indeed, in the book chapter that I have up on SSRN as a preprint, I discuss prevailing differences of opinion as to whether embryos are a person.
I think the questions about the morality of embryo destruction are in some ways more far reaching than the question about the morality of abortion. The reason I say this is because you can believe that abortion ought to be lawful and permitted, while still believing the foetuses are persons that deserve the protection that persons normally get: because you think women’s reproductive rights, their autonomy over their bodies, and control of one’s body is so important that it even permits women to engage in an activity which caused the destruction of a developing, or emerging person.
In the case of embryo destruction, we don’t have the same kind of countervailing interests around controlling our bodies. Instead, the countervailing interest is, one, about controlling what happens to the embryos we create, or having a right to create embryos that we will destroy in the end, and these seem to me to be weaker countervailing rights claim , which sets very directly the question about when does personhood begin in early embryo persons.
Q: You did a TED talk a while ago in which you discussed what it is to be a human and what it is to be a person, and it struck me that the court in this case said clearly “This embryo is a person: they’ve just gone straight there.” And as you say, it’s a tougher hill to die on than abortion, because there is less conflict around other people’s interests.
A: The questions around what a person is are fascinating, I’m so glad you saw the TED talk. I was talking a little bit about animal personhood, there are also questions about AI personhood that are now emerging, and to have a robust theory of personhood that explains A but not B, is philosophically challenging. Often, the prevailing theories of personhood have implications we don’t like. The Alabama case is asking narrow questions on a particular statute, but it’s a harbinger, I think, of things to come.
One of the Chief Justices of Alabama’s Supreme court had a concurring opinion full of explicitly religious language references, Bible discussions about how this offends God because we’re created in his image, and the idea of God’s holy wrath. It also raises the question in a larger political context, about to what extent our legislative acts or judicial decisions are permitted to invoke explicitly religious terms, or not, and that of course is a big debate too.
Q: It seems like our clever colleagues in technology are keeping bioethicists busy in reproductive health, particularly with IVF, artificial wombs, and the reverse engineering of fertility cells from other human tissue; are there technological developments in this field that you think will be raising interesting questions in the future?
A: I would say all the above… There are questions about creating artificial wombs, now mostly used to transfer very early prematurely born neonates and sustain them, outside the womb after birth. Maybe one day, we could perhaps have them starting gestation in artificial wombs. This raises all sorts of questions about whether you have a right to do this or can be prevented from doing this.
Uterus transplants are happening in many parts of the world already, and thinking about those, do you have a right to experience pregnancy that is separate from the right to be a parent? So, if someone says, “Well you could use a surrogate?”, can you say, “No, actually I want to carry in my own womb,” or really, a borrowed womb. This idea of transplanted uteruses also raises interesting questions about how we think about the uterus donors. What are their relationships to the child that’s born? Should the State pay for this, is this a want, or a medical need?
I talk to transplant surgeons and there is no reason to think male uterus transplantation is biologically impossible. We could also in the future perhaps transplant uteruses onto the male pelvis, either to people who are assigned male at birth and live as men, or to people assigned male at birth who have transitioned and are living their lives as women. Essentially the idea would be that you might make a claim that a person has a right to experience pregnancy as someone who was assigned male at birth, and does the State have an obligation to pay for that?
In vitro gametogenesis is the potential to derive sperm and egg from adult cells, and we’ve done this successfully in mice and dogs. There are companies now trying to perfect this in human beings too. There’s a lot of reasons why this might not happen scientifically or regulatorily, but essentially there are interesting questions about whether same sex couples, for example, should have a rights claim to use In vitro gametogenesis to produce a child that is both of their genetic lineage; so from sperm and egg, one of which being derived from the person in the relationship that normally doesn’t do that, so an egg derived from our males adult skin cells, and a sperm derived from a woman’s adult stem cells.
There is also the possibility of true genetic single parenthood, where you provide both the sperm and the egg. There is the idea of Multiplex parenthood, which would be, for example taking sixteen people fand deriving an embryo from each of them and then deriving from that sperm and egg, and then putting it together and so on and so forth. So, collapsing what we normally think of as the pathway from our great-great- grandparents to us, which usually happens over multiple generations, into a single generation. Why would you want to do that is unclear, but it’s an interesting question that pushes the boundaries beyond the more traditional, ‘reproductive technologies’, such as IVF, surrogacy, sperm provision, sperm donations and the like. So, the ethical and legal questions abound, which is one of the reasons why I love working in this field.
Q: AI is the current big technology breakthrough. At the end of your SSRN paper on ChatGPT, you write about the need for ethicists as gadflies to ensure companies do the right thing. What should a good gadfly do in this space?
A: I think this is very interesting because I am asked sometimes to give ethical advice or legal advice in this space. My view is, your responsibility is not to future-proof things, because it turns out that the like the famous Yogi Berra quote says, “it’s very hard to make predictions, especially about the future.” I think you need to really press on questions and viewpoints that might not otherwise come to the table, especially in the product development cycle, which obeys certain kinds of imperatives and incentives.
I do not think it is a a horrific fact that corporations don’t necessarily always operate by the same values or viewpoints that I may have as an ethicist: That is, I am not appalled that corporations are going to corporate. But, it should be incumbent upon them, and if not incumbent upon them then incumbent upon outside forces to push them, to really have this pressure-testing to say, “What could go wrong, but also, “What could go right?”
We don’t always think about successes and the way successes could create ethically problematic settings; I think about this in the healthcare space I work in. For privacy, often we think about the error that somebody will get the wrong impression about me and make the wrong inference about me from partial information. One of the biggest threats from AI is that it becomes *too good* at making the correct inferences about me and about my future health state.
Insurance works for all of us because we operate under conditions of uncertainty, whereas if I could exactly predict what your future costs would be, it would be the kind of thing that might destroy the insurance process because we couldn’t cross subsidise. So, there are ways in which it is not just our failures, but our successes we must think about. We should at least pose the questions and bring to bear different perspectives, and it’s a helpful thing when they are brought into a conversation, rather than merely being on the outside criticising after the fact.
Q: Just as a worked example of how AI might impact journalism, when I asked ChatGPT: ‘If you were going to interview Prof Glenn Cohen in the Wall St Journal, what would be some interesting and intellectually challenging legal questions that you could ask him?’ I got: “Ethical implications of emerging medical technologies; legal challenges in global health crises; reproductive rights; artificial intelligence; legal and ethical issues in healthcare; access and inequality; regulatory challenges and medical tourism; AI and law practise and surrogacy and reproductive justice.” It’s not bad…
A: Not bad at all…
Q: One of the big issues for publishers is the issue of training AI with copyrighted material. How do you see that challenge for copyright holders for training?
A: I have very smart colleagues who are thinking about that, and I view myself as an amateur here. But putting aside whether the law declares there’s a copyright violation or not, let’s ask ourselves the question; do we think that something is owed in a deeper sense?
I often make this point within the medical space that if you build an amazing algorithm or device that helps detect cancer much earlier, or helps do a better job of sorting benign versus malignant cancer tumours in a much less invasive way, and you do this based on data collected from people across the United States, or across the world, what are your ethical obligations once you have built it? What about equitable access for all those people. If not, what are you going to do to provide that?
I often say (borrowing a phrase from my friend and co-author Nicholson Price) that much of the value of medical AI is democratising the expertise we already have, rather than taking the expertise we have and making it even better. Dermatologists in the United States might become very good with AI augmented dermatology and that’s great, but when I think about all the people across the world, and across the US, that don’t have the opportunity to see a dermatologist, and I think by taking the expertise of just an average, or even a below average dermatologist and scaling that up, it seems ethically the really attractive value proposition.
We’re only going to get that if the incentives to develop AI, to build AI, to disseminate AI, align with that. So, I think that people often talk about the alignment problem in AI and making AI a kind of, a sorcerer’s apprentice, I think there’s an even bigger alignment problem, which is the alignment between what AI is getting built for, and what it has value for in an ethical sense.
Q: How have you seen this stuff affect your students and other researchers work? I have had some interesting conversations with my son about his university essays and students all seem to be making their peace with the fact that this is very often a tool that, just like Google, is a part of their research – although that may be a benign explanation of what they’re doing…
A: We have the benefit of being an extremely well-resourced university with very smart people who are ahead of the curve, such that, even when Open AI and Chat GPT debuted, there were many people around these parts who knew a lot about it, even at the very beginning, and had given some thought before much of the craze became public. As a result, I think there is a predictable reaction phasing and we’ve been able to move through it a little quicker.
So, the First Phase is just, “Oh my gosh, what just happened, this is going to change everything I need to do, we need to run to get academic policies and concerns about cheating, and what will be our policy on this”” etc.
The Second Phase, I think, is this idea of, “This is so interesting…” We can observe and run a bunch of experiments and see what’s working, see what’s helpful in the educational space and see what our students are doing, both in the settings we observe them, but also, for example, as our law students in their summers; we can ask what are the law firms doing, and how are they dealing with that?
I think we’re still in the second phase, but transitioning to the Third Phase, which is to think that of all those costs or uses: which are the ones that are corrosive to what we want to achieve in education, which are the ones that are exciting and successful, and which are the ones that we want to be creative with and think about integrating?
I give an example of our flagship course for people learning computer science; for this course, I understand them to have built a tutor or bot that has rules about what it will answer, and what it won’t answer. In a way, that could scale feedback much more effectively than only using teaching fellows, that’s an exciting use case. What about the ability for students to generate brand-new exam questions, in the mode of the kinds of exam questions I would ask, but that are not my exam questions, and then being able to critique answers as practise? That’s exciting and useful.
I don’t want to suggest that all the concerns about cheating are something we should ignore, there’s still lots of questions within education and medicine about deep fakes, about watermarking, and that is the current conversation among the big tech companies, in part because they try to avoid actual regulation, so they’re trying to self-regulate this space.
The decision to unveil ChatGPT, and the other labs, in the way they did, in the moment they did, was a hugely costly endeavour for most educational institutions. To suddenly have to scramble, learn, protect against, and deal with this, when I think about this idea of redistribution, effort, and time to introduce a product that has all of these impacts for the world, I wonder, what’s your obligation before you release it to work with those communities, socialise those communities, and also have the fixes ready to go for their issues?
I can tell you the standard plagiarism detectors, for example, are not always that good on ChatGPT. I do think it raises a question about political economy, that large tech companies do what they do, and the people who have to pay for it indirectly might be poorly funded public schools, and the educators there; teachers who are under-compensated here in the United States, scrambling and trying to figure this out, and that’s an interesting case study in itself for the ethics of product launch.
Q: You were practicing law in the government for the dept of justice. There is a huge difference between making legal policy and studying law academically. Why for you has the pull of academia been more powerful?
A: I loved my time at the Justice Department, I was born in Canada, a Canadian citizen. To say, ‘Glenn Cohen on behalf of the United States of America your honour’, in Court was this crazy amazing thing, a huge amount of pressure and exciting too.
Why make the switch? I think there is basically two reasons, for me at least, one is the ability to choose the projects that interest me as opposed to doing the work of the government, which is incredibly important but what is important to the government on a particular day and its overlap with my area of interest may be small.
The second reason was as a litigator I was often, if not always, in an adversarial position where my view is, “The government’s right, we’re going to win, or “Our position’s not so great, so can I figure out how to settle this case?” But it’s essentially a constantly adversarial position, us against other people, whereas for me, I really also enjoy a little bit more of a constructive relationship, particularly with my students, the ability to foster, to mentor, to support, and to learn from them. That is incredibly exciting and enriching and that is probably the main thing that caused me to make this switch.
Q: Law has always been one of the most important disciplines for SSRN, how do you think about SSRN in the context of legal scholarship?
A: SSRN has become a huge hub for legal scholarship, especially via the subject matter journals that many view as one of the best ways to keep up. We are blessed that law reviews tend to be open access by default, such that SSRN often ends up carrying not just pre-prints but final versions of articles too. Especially in fast-moving areas (like the recently Supreme Court litigation over the approval of the drug Mifepristone for early abortions or Trump’s ballot eligibility) the drafts that go up on SSRN are vital for educating the public and the bar since the cases may be resolved before the publication cycle of the actual paper concludes.
Q: If people wanted to learn more about your research, are there any books or papers that you’ve shared on SSRN or elsewhere recently that would be a good start for them?
A: To my surprise I apparently have 129 scholarly papers on SSRN, so I am a frequent flyer!I am not sure if I am a good judge of quality, but here are a few to look at that might be interesting regarding things in the news.
On how to think about the current controversy over embryo personhood, perhaps this book chapter by me: (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4284007),
For early thoughts on bioethics and ChatGpt just as it was getting out the gate,
(https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4430100)
For how to think about liability for artificial intelligence in medicine this co-authored book chapter:
(https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4115538)
And for learning about state attempts to regulate the therapeutic and non-therapeutic use of psychedelics, in this case focused on micro dosing, this recent co-authored paper: (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4544500)