To account for the dramatic decline in fertility observed across high- and middle-income countries over the past decades, part of the relevant literature has focused on the role of technological change and its unequal impact by socio-economic status (SES), education and gender. A large body of work has pointed to labour market polarisation – driven by sharp technological changes and automation – as a key mechanism underlying shifts in fertility and union formation (Autor et al., 2006, 2019). In a paper published in this journal, I provided a general overview of some of these changes and the role of occupational characteristics of jobs (Adserà, 2017; see also Matysiak et al., 2023; Bogusz et al., 2025).

The rise of artificial intelligence (AI) may extend or transform the impact of technological change on fertility choices. AI systems not only automate tasks, but also allocate labour and change job requirements (e.g., algorithmic shift scheduling, increases in remote work enabled by the digital transformation), mediate the partner search (e.g., dating app recommender systems) and enter reproductive medicine (e.g., using AI for embryo selection, fertility planning apps). This note explores, first, the channels through which AI may influence population variation in fertility patterns; and, second, whether those channels replicate earlier mechanisms or are qualitatively different from automation.

In the rest of this article, I organise the discussion around four channels through which AI may reshape – and potentially amplify – within-population variation in fertility patterns over the next 20 years: (1) labour market polarisation and its impact by gender; (2) AI-enabled reproductive technologies and information landscapes; (3) algorithmic partner matching and assortative union formation; and (4) algorithmic influence on fertility ideals.

Throughout this debate article, I highlight differential effects by gender and education, and assess whether these new divides will replace or reinforce existing ones.

Over the past few decades, technological change, deindustrialisation and the expansion of service economies have led to the disappearance of a range of middle-skilled jobs in many advanced economies. This hollowing out of the occupational structure, and particularly the erosion of routine jobs, has been widely documented (Autor, 2014; Cortes et al., 2020). The decline of stable, moderately skilled employment has altered the distribution of earnings, and has widened the gulf between labour market “winners” and “losers”. The resulting polarisation between high-wage, high-skilled occupations and low-wage, precarious employment has transformed both labour market trajectories and the conditions under which individuals form families (Oesch, 2013). These developments have contributed to lower fertility, particularly among those most vulnerable to economic instability (e.g., Adserà 2011; Ananat et al., 2013; Autor et al., 2019).

Prior studies have shown that unfavourable labour market conditions, such as unemployment and unstable contracts, negatively affect fertility (e.g., Adserà, 2005, 2011; Kreyenfeld and Andersson, 2014, among many; see Alderotti et al., 2021, for a metanalysis). Labour demand shocks disproportionately impacting male-dominated sectors (mostly in manufacturing) have led to reductions in marriage and fertility rates, especially among lower-educated men, although in the past these declines were sometimes offset by rises in teen fertility within certain disadvantaged groups (Autor et al., 2006, 2019; Ananat et al., 2013; Schaller, 2016). However, the recent sharp drop in teenage fertility rates makes such offsetting effects less likely in the future. Anelli et al. (2024) have shown how the diffusion of robots in the US negatively affected men’s labour market prospects, reducing their marriageability and inducing a shift towards nonmarital births. Furthermore, in some clerical fields traditionally dominated by medium-educated female workers, employment opportunities and earnings have declined substantially, depressing the ability of these workers to combine stable work and family formation (Cortes et al., 2017). The de-routinisation of work has had a larger impact on non-college-educated women in recent cohorts than on those in earlier cohorts, potentially delaying their transition to adulthood in some domains such as childbearing (Dabla-Norris et al., 2023). The automation of tasks and the advent of digital technologies that allow individuals to work from home have been shown to have ambiguous effects on fertility that vary by education and parity (e.g., Matysiak et al., 2023).

These structural changes have also affected the size of the motherhood penalty, which refers to the permanent income loss and/or slower career advancement suffered by mothers after childbirth compared to childless women. In related work, Adserà and Querin (2023) highlighted that the differences in the occupational characteristics of the jobs mothers and fathers sort into strongly shape the parental wage gaps across European countries. Technologically induced changes in the demand for different types of skills and subsequent returns to different job characteristics tend to affect the extent of the motherhood penalty and/or lead to new forms of employment sorting among parents – for example, fathers are more likely than mothers to work in occupations with more autonomy and leadership requirements (Adserà and Querin, 2023, 2024).

The advent of AI is expected to accelerate these processes via three channels: labour augmentation, capital labour substitution and the potential rise of new types of tasks. By automating both routine cognitive and manual tasks, AI deepens the divide between occupations that can be fully automated and those that complement new technologies (Acemoglu and Restrepo, 2018). Thus, the extent to which AI will affect a job negatively or positively will depend on how exposed the job is to AI and how complementary the skills required for the job are to AI (Pizzinelli et al., 2023; Cazzaniga et al., 2024), similarly to how robots affected jobs in the past (Dauth et al., 2021). The degree of complementarity with AI may depend on workers’ skills. Frey and Osborne (2017) estimated that a large share of current jobs, especially those held by workers with mid-level education and limited digital skills, are susceptible to computerisation. In contrast, jobs requiring interpersonal, creative or highly technical abilities are less exposed and are more likely to benefit from AI augmentation. This evolving skill bias may reinforce the advantages of highly educated individuals, and may further delay or reduce fertility among the less educated, who face growing precarity and declining prospects. Although several studies have begun to document these patterns (Pizzinelli et al., 2023; Cazzaniga et al., 2024), the research community is still puzzling over the ultimate extent of AI’s impact on employment and on the allocation of work across different types of workers, including by gender. Within these ongoing labour market transformations, I highlight four channels through which AI can directly affect fertility: (1) gender-biased effects; (2) increased monitoring; (3) changes in required work effort; and (4) improvements in household technologies.

The impact of AI may not be gender-neutral, and could have particularly significant effects on fertility. The displacement and augmentation effects of AI are likely to play out differently for men and women, depending on their sectoral exposure and the extent to which their skills are complementary to AI (Pizzinelli et al., 2023). First, if women hold jobs in sectors that are relatively more exposed, they may be more affected by capital-labour substitution. Second, if their skills are more complementary to AI, they may be more protected and benefit from a productivity boost. Depending on the relative strength of these two forces, AI will induce changes in the wages and employment of women relative to those of men. These changes are likely to have implications for union formation and fertility.

According to a recent IMF report, women and more educated workers are more likely to be in jobs that are highly exposed to AI (Cazzaniga et al., 2024). However, many of these jobs are also more likely to benefit from AI, as the skills required are complementary to AI. For women in particular the report suggests that while they face higher AI-related risks, they also stand to gain more from the opportunities AI offers, given that they tend to be employed in occupations with both low and high degrees of complementarity. For example, women employed in the care and education sectors may benefit from AI as a complementary technology, provided there is adequate institutional support to facilitate its use in the workplace. Thus, AI’s impact on fertility via labour market allocation depends on how it affects the absolute economic resources available to couples with different education levels, as well as how it alters the gender (and parental) wage gaps across these groups.

At the same time, a second potential channel through which AI could affect parenthood choices is by intensifying the existing trend towards increased remote work. In this regard, AI offers specific technologies such as algorithmic workplace monitoring and scheduling automation. The ultimate effect on fertility is ambiguous because while such technologies can provide more flexibility, they can also place more round-the-clock time demands on parents with young children.

Third, the impact of AI on effort in the workplace remains unclear. Recent research has shown that even though some workers may benefit from AI complementarity, they might also face extended hours of work, not only because of the incentives of increased productivity, but also because of extensive monitoring of effort via AI itself (Jiang et al., 2025). If workers experience longer weekly work hours, instead of the relief that some observers expected from the introduction of AI, they may find it harder to achieve work-life balance, leading them to delay their parenthood plans. Conversely, if AI boosts capital-labour substitution, policymakers may call for fiscal policies taxing AI technologies to be accompanied by regulations aimed at reducing working time, increasing leisure time and expanding resources that lower barriers to childbearing.

Finally, AI also introduces potential innovations in housework that may compensate for higher demands from the labour market. Historically, technological innovations in the domestic sphere have played a major role in reshaping gender roles and fertility. Greenwood et al. (2005) showed that the diffusion of household appliances (“engines of liberation”) in the 20th century was instrumental in significantly reducing the time women spent on housework and in boosting their labour force participation. AI-driven “smart homes” may offer similar benefits. Lehdonvirta et al. (2023) conducted a survey of 65 AI researchers in Japan and the UK to evaluate to what extent 17 types of domestic tasks are easily automatable. The experts estimated that, on average, 39% of the time spent on these tasks could be automated within the next decade. Importantly, the assessments differed by the types of tasks, with a larger share of housework tasks (44%) than of caregiving tasks (28%) considered potentially automatable. This is one of the first studies to provide systematic estimates specifically for unpaid domestic work. If such time-saving technologies spread widely they may alleviate the motherhood time tax and ease fertility constraints, especially among dual-earner couples. It is, however, important to note that the tasks most easily transferable to AI devices are those not directly related to care (of either children or the elderly), which have traditionally been performed mainly by women. The next wave of AI embodied in robots may facilitate the automation of some of those tasks. There are already devices geared to elderly care (e.g., Japan’s Paro, Korea’s Diaper-bot). To what extent parents will be ready to use similar devices to care for children, and how long it will take for this technology to become available, are open questions.

In sum, AI has the potential to reshape labour market and intrahousehold dynamics in ways that both deepen and mitigate fertility inequalities. Its long-term effects will depend on the nature of technological diffusion, the distribution of skills, the impact on hours of work and the responsiveness of social and labour market institutions.

With the delay in fertility and changes in the perceived reproductive age window, the use of assisted reproductive technology (ART) has been growing at an accelerated rate in recent years (Sobotka et al., 2008; Beaujouan and Sobotka, 2019; Lazzari et al., 2023, 2024).

The application of AI in reproductive health represents one scenario in which technology-driven change can lead to social stratification in family behaviour through differences in information access or cost. Fertility tracking apps such as Natural Cycles, Clue and Ovia use machine learning to provide personalised ovulation predictions and cycle analysis. These apps usually have a free basic version but require a paid subscription to access additional features. In clinical contexts, AI is increasingly being used in fertility treatments, particularly in vitro fertilisation (IVF), to improve their accuracy and efficiency – and, potentially, their success rates. AI is being applied in various aspects of IVF, including optimising ovarian stimulation, analysing sperm and oocyte quality and selecting the best embryos for transfer (e.g., Zaninovic and Rosenwaks, 2020; Chow et al., 2021; Hanassab et al., 2024; Mapari et al., 2024).

However, access to these tools is far from universal, and will critically depend on the level of institutional support across countries. Fertility apps often require users to pay subscription fees, and to be digitally literate and comfortable with data sharing. ART enhanced by AI is expensive and is mainly available in high-income settings. Public coverage of ART is limited in some countries, and insurance barriers persist even in relatively generous systems. When ART is publicly funded, the number of cycles covered and the age restrictions vary across countries. As a result, these technologies primarily benefit wealthier, urban and well-educated individuals by offering them greater control over their fertility timing and outcomes, while excluding others. Goisis et al. (2024) analysed data from Denmark, France, Spain, the UK and the US and found that while an educational gradient exists in ART usage, its strength varies by national context, potentially due to differences in institutional and policy environments.

The ultimate impact of the advent of AI within the ART landscape is ambiguous. By increasing the success rates of reproductive interventions, it may provide easier access for those who are more cash-strapped or lead to an expansion of publicly provided services. Conversely, given the current educational gradients in delayed fertility and in the use of ART (including egg freezing), AI may further reinforce existing SES disparities in treatment seeking, access and affordability.

Beyond fertility technologies, AI is transforming how people form unions. Online dating has become the dominant mode of partner selection among younger adults in many countries. These platforms – Tinder, Bumble, Hinge and others – use recommender algorithms to match users based on preferences and behaviours. Drawing on a nationally representative 2017 survey, Rosenfeld et al. (2019) showed that since 2013, these online platforms have been the most common way couples in the US meet, surpassing introductions through friends. The findings suggest a major transformation in the social infrastructure of partner formation and a decline in traditional ways of meeting, such as through family, friends, churches or neighbourhoods.

These algorithms do not operate in a social vacuum. Users bring pre-existing preferences for age, race, education and religion, which platforms often reinforce through design. Research has shown that dating apps have increased educational homogamy and potentially reduced exposure to diverse partner pools. Potarca (2021) found that online dating appeared to be especially advantageous for highly educated women in Germany, as they had lower marriage rates when meeting partners offline but higher marriage rates when meeting partners online – partly because they were better able to find partners with similar marriage attitudes and egalitarian gender views. As AI-driven personalisation becomes more sophisticated, we may see even greater sorting along class and educational lines.

Interestingly, Potarca and Hook (2023) found that meeting a partner online is associated with a more equal division of routine housework – especially for married women with lower education, who typically face the most unequal domestic arrangements. They argued that differences in how gender dynamics are negotiated during the digital courtship phase may influence later household labour arrangements. A more equal sharing of household chores between spouses may lead to higher fertility in some contexts in which traditional gender roles in domestic tasks are still highly prevalent, as the uneven division of domestic work has been shown to discourage women from childbearing (McDonald, 2000).

Educational homogamy compounds inequality because household resources depend on both partners’ human capital (Schwartz, 2013). If AI-driven recommendation strengthens “like-with-like” pairing, it may increase income (and wealth) inequality across households. It may also lead to inequalities in the extent to which couples achieve their fertility goals, as highly educated dual earners may delay but ultimately be able to afford parenthood, whereas individuals with lower education or unstable employment may struggle to find partners and end up with lower realised fertility and potentially higher dissolution risk. Such dynamics may in turn intensify the gender differences already uncovered by the literature in the association between automation, outsourcing and marriageable men (Kearney and Wilson, 2018; Autor et al., 2019).

The lack of transparency around the algorithms used by different platforms also raises concerns about racial and socio-economic bias in whose profiles are made visible to others – an overlooked yet potentially significant factor in shaping individuals’ reproductive opportunities by strengthening educational or ethnic homogamy.

Beyond its impact on union formation, AI curates the information that shapes reproductive decision-making. Social media feeds and search engines provide personalised advice on the “right” age to have children, the costs of parenting or environmental anxieties about overpopulation. Women may be more exposed than men to fertility content (e.g., egg freezing advertisements). It is unclear whether social influencers and information shared on social media around pregnancy or parenting are beneficial or harmful to the process of decision-making (Chee et al., 2023). More research on how such tailored information affects stated fertility ideals as well as fertility behaviour is needed. On the importance of screen time in shaping fertility preferences, see Rotkirch (2025).

Summing up, artificial intelligence is poised to reshape many dimensions of social life, including fertility and family formation. Its influence will be mediated by structural inequalities in labour markets, education and digital access. While AI may offer enhanced tools for reproductive autonomy and partner selection, it might also reinforce educational and other socio-economic barriers to parenthood. Do these emerging dynamics suggest a replacement of old divides or the layering of new inequalities atop existing ones? Some of the evidence suggests that the latter is the case. AI may be accompanied by new mechanisms – e.g., access to platforms, algorithmic bias and digital capital – that amplify long-standing gaps. However, other evidence indicates that more equal sharing of housework and a reduction of the burden of the double shift by AI devices may ease the path to realising fertility goals. The complementarities of AI in some female-dominated sectors may also narrow the gender wage gap and decrease the motherhood penalty in ways that could have ambiguous effects on fertility. Understanding these dynamics will be critical to explaining fertility patterns in the 21st century.