why we should think hard about artificial wombs

a philosophical long-read about the possibility and permissibility of a fully mechanised route to biological motherhood

TLDR:

• Child-bearing is so brutal and risky that it would be hard to deny women a fully mechanised alternative, if such an alternative were possible

• Full machine gestation could offer substantial benefits to women — and often to foetuses and men — in relation to wellbeing, equality, and access to opportunity

• Most arguments against mechanised gestation depend on flawed naturalistic reasoning. But a child’s moral right not to be intentionally deprived of its mother imposes conditionality on the permissibility of the non-child-bearing route

---

Child-bearing is a brutal and risky activity. This isn’t a controversial claim: just think about what giving birth involves, even in instances described as ‘easy’. Think about what carrying and expelling a whole other human being does to a woman, physically and psychologically. Seven-to-eight pounds, on average, of bony mass, with a length of a half a meter, and a circumference at points that’s not much smaller. A bony mass that struggles and kicks, and sometimes gets stuck on the way out. Tell yourself it’s a natural activity, and tell yourself most adult women throughout history have done it. Because those things are true. But none of this changes the fact that child-bearing is brutal and risky.

The brutality and risk of child-bearing is at its extreme in places with substandard maternity services. A recent Oxford report tells us that the rate of UK women dying whilst pregnant, giving birth, or soon after, recently rose to the highest level in almost 20 years. And that UK women from socio-economically deprived and some ethnic-minority groups face a higher risk, by multiples, than their white and less socio-economically deprived comparators. Meanwhile, a House of Lords Library paper emphasises that “in November 2023, around two-thirds (67 per cent) of England’s maternity units received a CQC rating of ‘requires improvement’ or ‘inadequate’ on safety”. And in 2022, the Guardian reported that the UK’s maternal mortality rate was higher than many other European countries including France and Italy, and three times higher than Norway.

If you’re not British, or you’re blinded by unconditional love for the NHS, then the UK’s substandard maternity performance might surprise you. You won’t be surprised by other places, however. As this Our World in Data chart shows, the decrease in worldwide maternal mortality, over the past forty years, is one of the great human achievements. But OWiD also tells us that in poverty-torn Chad, a fifteen-year-old girl’s lifetime probability of dying from pregnancy-related causes is almost 7 per cent. And that if all the countries of the world had, in 2015, had the same maternal mortality rate as the EU average, then 11,000 fewer women would have died that year.

This isn’t a piece about maternal mortality statistics, however. Partly because numbers can’t tell the whole story. Partly because many women who suffer horrifically from child-bearing nonetheless survive. (I bet I’m not an outlier in having female friends who’ve had unthinkable experiences; I’ve forced myself to read some detailed accounts, including this and this.) But the main reason I’m going to move on from maternal mortality statistics is that child-bearing is inherently brutal and risky, and as such would remain so even if every current maternity service operated perfectly.

I’ll spend the rest of this piece, therefore, imagining a world in which fertile women have the option of a mechanised non-child-bearing (‘NCB’) route to biological motherhood. This isn’t just a philosopher’s thought experiment, however. Artificial-womb technology is developing fast. Recent progress with ‘biobags’ — currently being tested during the gestation of animal foetuses — precursors a time in which it’s possible for a woman’s bodily involvement in the development of her baby to be limited to the contribution of an egg. Moreover, as Aria Babu points out, we can conceive of the use of artificial wombs simply as an extension to the use of existing technologies — including IVF and neonatal intensive care (‘NICU’) incubators — that “already allow some women to spend parts of the gestation period not pregnant”. I don’t need to make predictions about how the mechanised NCB route will develop, however. I’ll simply accept it exists in a ‘nearby possible’ world. Finally, I should acknowledge that women can, of course, already become mothers without giving birth. Beyond IVF and NICU incubators, Caesarean sections and adoption also edge us into the NCB world.

At the moment, however, only part of a foetus’s gestation can be outsourced to a machine, and this happens out of necessity rather than maternal choice. The underlying question I’m seeking to answer, therefore, is not whether it’s likely that ‘full machine gestation’ will become a general option, but whether it should. Nonetheless, I’m happy to begin by asserting that, minimally, the permissibility of the mechanised NCB option is conditional on tests (no doubt carried out gradually on ‘necessity’ cases) having reliably determined that undergoing full machine gestation causes no increase of physical foetal suffering, when compared against gestation in the womb.

The impact of the mechanised NCB option on wellbeing

So far, I’ve stacked this piece heavily in favour of concluding that gaining the mechanised NCB option would be advantageous to women’s wellbeing. There’s more I could’ve done, too. I haven’t discussed the psychological costs of child-bearing, for instance, beyond implying that experiencing its physical hardships must be psychologically hard. Yet it’s well-known that one in 10 mothers suffers post-natal depression, and there’s strong evidence that new mothers are at particular risk of OCD, other mental health conditions, and financial stress. I accept, however, that taking the mechanised NCB option might not alleviate these risks.

I also accept that most women who endure child-bearing feel pride in having done so. Indeed, I even accept that enduring child-bearing counts objectively as an achievement, and will do for at least as long as child-bearing remains necessary to bringing children into the world. But activities that must be endured don’t automatically become good, in themselves, just because they enable achievement! Nor do they become good by virtue of bringing about the existence of good things — even unalloyedly good things, like babies. After all, many bad things have good side effects: just think about the friendships that often arise from shared tragedy.

Beyond the positive feelings that many women associate with having endured child-bearing, however, some women claim not to mind — or even to enjoy — giving birth. At the extreme, some insist they find ecstasy in it. And many more derive some physical pleasure from pregnancy. But none of this can suffice to make child-bearing a good thing in itself, either! There are many unfortunate cases of human beings finding pleasure in undeniably bodily and mentally harmful activities. Some find satisfaction in the unnecessary loss of limbs; others in sexual violence. So, unless you have a sufficiently thin hedonistic conception of the good that enables you to find something normatively positive in such examples, then you should probably accept that undeniably harmful activities don’t automatically gain goodness, in themselves, just because some people find pleasure in them.

None of what I’ve argued so far suffices to make the case for the mechanised NCB option, however. None of it tells us that its net effect on women’s wellbeing would be positive, or that full machine gestation is permissible in the morally-thicker sense that cost-benefit analyses can’t reveal. It doesn’t even tell us that the mechanised NCB route is entirely non-brutal and non-risky. After all, it’s well documented that egg removal can be physically and psychologically tough — although it seems clear that the mechanised NCB option, even when used in combination with today’s egg-removal practices, would be vastly less brutal and risky than child-bearing. So, nothing I’ve argued can determine whether taking the mechanised NCB option would be morally permissible. Nonetheless, separating the value of a process from the value of what it produces can help us to assess what, specifically, might be lost and gained when that process is replaced by another. And the implication, so far, is that if there were a possible route to biological motherhood that was vastly less brutal and risky than child-bearing, then it’d take some work to justify denying that option to women.

Before I turn to my second point of focus, however, we should consider the relevance of the foetus’s wellbeing within all this. Rightly or wrongly, the interests of unborn children are typically subserved to the interests of their living mothers. But a strong consideration in favour of the mechanised NCB option hinges on foetal wellbeing. That is, I’ve already accepted that minimally, the permissibility of such an option is conditional on the reliable understanding that undergoing full machine gestation causes no increase of physical foetal suffering, when compared against gestation in the womb. But what about the likely physical benefits of full machine gestation to a foetus?

First, it seems clear that, whilst the existing technologies that enable women to avoid some of pregnancy are typically used only when necessary for a foetus’s survival, these technologies could already also enable improved foetal monitoring and medical intervention, more generally. Beyond this, the general use of such technology could also surely reduce instances of foetal alcohol spectrum disorder, and other harms and inequalities suffered by foetuses whose needs are insufficiently met owing to their mothers’ behaviours.¹ At the moment, however, general access to these machine-gestation benefits would come with the risk of ‘unnecessarily’ transferring developing foetuses from their human wombs — something that could only take place after they‘d reached the relatively late stage of being able to survive a NICU incubator. In the mechanised NCB world, however, pregnancy could be avoided entirely, and any physical problems suffered by a foetus could be identified — and ideally mitigated — much earlier, without the risks posed by womb-to-machine transfer. Beyond giving women the opportunity to avoid the brutality and risk of child-bearing, therefore, it seems likely that the mechanised NCB option would also bring significant benefits for the physical wellbeing of foetuses.

If we want machine gestation to move beyond its status as an occasional partial necessity, however, then difficult questions about its psychological effects on foetuses become more pressing. Here, the greatest difficulty lies in our inability to know what it is like to be a foetus or a new-born baby, even though all of us have been these things, ourselves. Difficulty lies also in our inability to test the counterfactual: no foetus’s psychology is identical with the psychology of any other foetus, even its identical twin, so how can we know how a foetus would have fared psychologically in different conditions? Of course, we do know that even extremely early instances of childhood harm can cause lifetime trauma. But why assume that machine gestation would be more traumatic than bodily gestation? After all, bodily gestation involves a foetus being carried from place to place — whilst its mother goes about her daily life — before being pushed down a tight canal. And it’s not as if the fully-machine-gestated foetus would be aware of what it was ‘missing out on’, anyway!

Beyond the difficulty of predicting the consequences of full machine gestation on babies’ psychology, therefore, it seems unlikely we could ever know the full truths of such a matter. Nonetheless, as with its physical impact, I’m happy to accept that, minimally, the mechanised NCB option is permissible conditional on tests having reliably determined that being fully gestated in a machine comes with no increase in psychological foetal suffering, compared against gestation in the womb.

The impact of the mechanised NCB option on equality between the sexes

I’ll now turn briefly to a second point of focus: the extent to which the mechanised NCB option might positively rebalance the distribution of child-raising labour between the sexes. I’m talking here specifically about opposite-sex co-parenting, and I’ll include some further clarifications, along the way. First, I accept there are some physically, psychologically, and morally valent biological differences between men and women. The most obvious of these are differences around reproductive function — and my comments above about the risks and brutality of child-bearing apply only to women, qua members of the female sex set. This isn’t to deny that some members of this set cannot bear children, or that there are members of the male set who self-identify as women. But I don’t want to get into a debate here about different meanings of the word ‘woman’. Rather, I’m simply emphasising that the male-female reproductive-function distinction bears relevance to my NCB thought experiment, because the set of human beings who bear children are female.

In this context, I’m also happy to accept that a mother’s particular biological involvement in her baby’s life underpins at least some of the imbalanced division of labour and opportunity that typically obtains when a male-female couple co-parent their child. The woman often breastfeeds, for instance. And typically, she has a specific hormonal response to child-bearing that engenders such a depth of nurturing feeling that it can influence her to preference child-raising over the other valuable activities on which she previously spent time. Now, I’m not denying the male capacity for nurturing children, or that an increase in nurturing feelings can be a hormonal response to fatherhood, or that some men buck the steady trend and use the entirety of their entitlement to paternity leave! But I’m happy to accept that the biology of motherhood has an effect on the distribution of child-raising labour, between the sexes. And that the sex-specificity of this is morally relevant, in terms of its implications for male-female interpersonal and societal fairness.

Two possibilities we should take into account, therefore, when considering the value of the mechanised NCB option, are that: 1) uptake of this option would likely correlate with a reduction in sex-specific child-raising labour, such as breast-feeding, for biological and social reasons; and 2) the mothers of babies gestated entirely by machine would likely experience a weaker hormonally-driven desire to preference spending time on child-raising over doing other valuable things.

Second, I’m happy to accept that child-bearing has long-term financial implications for women. In particular, that a causal factor behind the male-female salary and life-time earnings divergence, in the UK at least, is the career interruption typically experienced by mothers, which leads to reduced opportunities for career progression. Moreover, that this reduction of opportunity also likely comes with a deleterious effect on the satisfaction mothers find in their work — perhaps particularly when they see male comparators being promoted ahead of them — even when this is balanced out by the satisfaction they find in motherhood. In this context, the mechanised NCB option comes not only with likely egalitarian opportunities, but also the possibility of increasing mothers’ capacity to find fulfilment in a wide range of valuable work opportunities.

This isn’t to deny the other side of the ledger, however. Of course babies benefit from spending time with their mothers! Of course mothers benefit from spending time with their babies! But none of this makes it morally required for mothers and babies to spend every moment together. Nor does it mean we should ignore, out of hand, the value of babies and fathers spending more time together. And minimally, it seems likely that the mechanised NCB option would shift norms in favour of men pulling more weight before and during the child-raising years. After all, when babies are in NICU incubators, there’s little a mother can do for her child that its father cannot. And when a mother is no longer necessitated to carry the gestating child in her body, it seems almost certain that her proportion of child-related labour will decrease. Again, none of this suffices to make the case for the mechanised NCB option. Rather it’s simply to emphasise the varied opportunities — to women, babies, and men — that this option likely represents.

What about permissibility?

In my final section, therefore, I’m going to turn to the more complicated matter of permissibility. That is, beyond any likely benefits accrued, would a purposeful shift to the use of full machine gestation be morally justifiable, as a matter of general maternal choice? Specifically, would be it permissible for an individual woman to choose to bring her child into the world in such a way, solely as a matter of personal preference?

I want to emphasise at this point that it’s only the ‘mechanised NCB option’ that I’m talking about, here. So, the question on the table isn’t whether women should always opt for mechanised NCB, or whether child-bearing should be banned by the state. And it certainly isn’t whether we should stop the partial use of machine gestation in cases of foetal necessity. Rather, it’s whether taking up the mechanised NCB option — when it becomes possible — would be morally acceptable. After all, just because something becomes possible doesn’t make it automatically permissible. Otherwise, we’d struggle to argue against the use of the painless capital-punishment machine. And just because something brings about more good than bad doesn’t make it automatically permissible, either — partly because such calculations depend on trying to convert unquantifiable things into commensurable widgets, and partly because, relatedly, such approaches support and sometimes even demand inhumane conclusions.

In my search for permissibility, therefore, I’m going to turn to what I see as the two strongest arguments against the mechanised NCB option. Both of these arguments are focused on the societal moral implications of fully outsourcing human gestation to machines. I’ll briefly discuss each of these arguments, in the context of considering whether, even if these arguments hold, they are sufficient to override the serious problem of the brutality and risk of child-bearing. That is, regardless of any other possible benefits of the mechanised NCB option, can either of these arguments provide a strong enough reason to deny women a vastly less brutal and risky alternative route to biological motherhood? Finally, I’ll turn to a rights-based consideration, which I think provides a promising approach for imposing some conditionality on the permissibility of taking the mechanised NCB option.

The first strong argument against mechanised NCB is a naturalistic argument, on which its use is dismissed as an instance of ‘playing God’. This kind of argument is typically found in religious packaging. Here, its overriding concern might be that full machine gestation would anger God, or that using it would be disrespectful to God or God’s creation. The naturalistic argument also comes in non-religious packaging, however. Here, its concern might be that full machine gestation represents human arrogance, or that it simply isn’t our business as human beings to interfere in the process of gestation, whether or not that process was designed by God. Now, my guess is that most people – religious or not — would have some sympathy for naturalistic arguments against full machine gestation. That is, I think that outsourcing the entirety of the human gestation process to machines is an idea that most people would find at least a little disturbing. And that, if they thought hard enough about the reasons behind their disturbance, they’d likely conclude that ‘unnaturalness’ was playing some part.

Nonetheless, I can re-use some of my earlier moves to dismiss naturalistic concerns. First, as above, it’s crucial here that we separate the value of the process from the value of the outcome. The unalloyed good of a particular baby’s life cannot suffice to ‘make good’ any possible level of suffering its mother experiences during its gestation. And neither can the goodness of a particular baby’s life be used to justify any possible route to bringing more babies into existence. After all, if it turned out that having sex with cows was a possible route to making human babies, then presumably you’d want to count that as morally impermissible. You’d probably also want to rule out the idea that a woman should be killed if her ongoing existence provided an otherwise unavoidable mortal threat to her foetus’s survival. And finally, the fact that the current child-bearing process happens to be the current child-bearing process cannot suffice to make it the only permissible option. After all, if human gestation had developed, or did develop, to consist in some alternative approach — the slow opening of a clenched fist, or the laying and protection of an egg, like a chicken — then surely we’d have no reason to value the current brutal and risky process.

All that said, if you accept the theological premises underpinning religious versions of the naturalistic argument, then you’ll be able to make some side arguments, here. Perhaps you’ll tell me that everything that God designed is good in itself. And perhaps you’ll withstand the obvious objections I use to respond to this — objections focused on the natural things that I think clearly involve badness, such as the human tendency for vengeance, and the human instinct to eat other living creatures. Nonetheless, religious or not, unless you can convince me that child-bearing isn’t inherently brutal and risky, then you’re going to have to persuade me that a woman shouldn’t be permitted to opt for mechanised NCB, instead — and you’re going to have to do this without depending on non-natural mitigants like synthetic pain-killers.

The second strong argument against the mechanised NCB option is an ‘unintended consequences’ argument. Here, however, rather than focusing on the possible but as-yet-unknown deleterious impact of full machine gestation on the physical or psychological wellbeing of foetuses, this argument is focused on the concern that such technology could give rise to dystopian societal effects. My friend Ellen Pasternack argues persuasively, for instance, that we should expect the world of full machine gestation to be a world in which: 1) the needs of women who wished to become mothers “the traditional way” would be left unmet; and 2) the overbearing state would set about creating people in its “image”, to serve as “the next generation of workers”. I don’t deny the risk of either of these unintended consequences.

Nor do I deny the even more concerning unintended-consequences risk that a mechanised NCB world might be one in which machines were used to gestate babies that weren’t just workers in the state’s “image”, but were workers of whatever number deemed required to meet the state’s goals. After all, as Pasternack puts it, the state could have “control over the supply of new citizens”. And, in an age of increasingly low birth rates, it's not far-fetched to imagine that full machine gestation might be used to ensure that vastly more fertilised eggs were gestated than could ever be gestated in the wombs of women. That is, can we trust that the introduction of full machine gestation wouldn’t lead to the mass production of babies gestated for state-determined ends, and without any intention of active parental involvement?

Finally, I don’t deny what’s possibly the most concerning unintended-consequences risk of all: that the mechanised NCB option might have significant negative effects on the evolutionary process. An anti-mechanised-NCB argument depending on this risk would run something like: “Sure, evolution made women suffer from child-bearing, but that’s the cost of human optimisation!” I’m unconvinced, however, that most people who’d use evolutionary concerns to oppose the mechanised NCB option would also be happy to make correlative arguments if evolution had led male humans to develop barbs on their penises, like cats. My guess, rather, is that they’d be persuadable on the need to look for a much better solution than offering women sufficiently strong pain-killers to use during sex.

I don’t need to depend on the penis-barb objection, here, however. And I don’t need to come up with similar specific objections to other unintended-consequence concerns, either — including strong Pasternack-type concerns about the overbearing state. Rather, I can simply conclude that, unless you’re a determinist, then you should recognise that unintended societal consequences are avoidable, once they become likely consequences! And that before then, we require something more than their possibility to deny potential mothers the option of avoiding the risk and brutality of child-bearing.

Towards a rights-based argument for imposing conditionality on taking the mechanised NCB option

I’m going to end, therefore, by raising a different concern about the mechanised NCB option. This concern, I believe, offers a route into a permissibility argument that, minimally, should cause us to hesitate about the introduction of full machine gestation as a matter solely for maternal choice. This is a concern that reflects something that seems undeniably wrong about that scenario in which technology is used by the state to create as many workers as it deems required — a scenario in which many more humans are created than could ever be gestated in the wombs of women. The undeniably wrong thing I’m referring to, here, is that these babies are being intentionally brought into the world, effectively motherless. Yet surely every child has the moral right not to be intentionally deprived of its biological mother — except in extreme cases in which she represents a serious risk. Of course, this isn’t to deny that many people lead fulfilling lives without the presence of their biological mother. Rather, it’s to argue that intentionally depriving a child of that, except in extreme cases, is a particularly egregious human wrong.

A complicated question pertains, however, to what exactly the moral obligations are that each of us holds correlative to this moral right held by each child. Surely, that is, this right underpins one of the many reasons you should avoid purposefully running over the mother who’s crossing the road in front of your car. Likely, it also underpins some of your moral requirement to support societal commitment to medical research and baseline standards of health-care provision. It probably doesn’t require you to pay for all the living costs of your colleague’s baby, however, to prevent your colleague from having to spend time away from home, working! And it surely doesn’t require you to let the two of them live in your spare room, when impending homelessness would otherwise cause them to be separated. My instinct, however, is that a child’s moral right not to be intentionally deprived of its biological mother could indeed be used to underpin a permissibility argument against full machine gestation as a matter of non-conditional maternal choice.

That is, if foetuses, as unborn children, also hold the moral right not to be intentionally deprived of their biological mothers, then the permissibility of a woman’s choice to use full machine gestation could be taken as conditional on her desire not only to become a biological mother, but also her desire to actively mother her child.² This argument would provide safeguards against state attempts to optimise the birth rate through the use of full machine gestation. And it also reflects my general view that all people who choose to have children should want to be parents. Of course this isn’t to deny, for instance, that many women who unintentionally become pregnant turn out to be wonderful parents! This approach also reflects general practice around adoption. That is, a standardly-accepted requirement of becoming a parent through adoption is not only wanting to acquire the status of parent, but wanting to parent a child. I should note, however, that I’m aware my argument provokes questions about the permissibility of surrogacy — but, as I’ll discuss here some other time, I already have deep concerns about that.

Rather, I’ll end by making two short side points. First, that my argument is only an argument about moral rights and obligations. All I’ve been doing here is setting out some thoughts about the moral permissibility of opting for the mechanised NCB route. And, while I believe these thoughts could also be used to underpin a framework for allocating legal access to full machine gestation, such thoughts clearly couldn’t suffice to tell us what such a framework should look like, or how the relevant laws should be enforced.

Finally, as you’ve probably realised, I’m keen to emphasise that much of what I’ve been arguing in this article can also be used to argue for general access to vastly improved maternity care. One reading of this article, therefore, is that I’ve been engaging in a Machiavellian process of ‘aiming above the target’. That is, maybe I haven’t convinced you that the brutality and risk of child-bearing provides strong reason to pursue the general introduction of a mechanised NCB option. But hopefully, I’ve minimally convinced you that the reasons typically offered for denying the brutality and risk of child-bearing are flawed, and that serious societal work is required to acknowledge and try to make good on this.

1. Thanks go to my friend Emily for this point.

2. I’m aware that some people might have concerns about what my argument implies regarding access to necessary machine gestation for extant foetuses whose mothers do not want to mother them. But just because I’m arguing that all instances of chosen mechanised NCB gestation require the maternal desire to parent, doesn’t mean I’m arguing that all instances of access to mechanised gestation require this.