Dating apps have long been unsafe spaces for women and gender-diverse people, particularly at points of intersection. In recent years, the highly visual nature of dating apps, what many critics describe as the aesthetic economy, has come under increasing scrutiny.
A 2024 systematic review found strong links between dating app use, body dissatisfaction, low self-esteem, anxiety and depression, while other recent studies have identified that women using dating apps are significantly more likely to pursue cosmetic procedures and body modifications.
At the centre of this criticism is the swipe itself: a mechanism designed to encourage split-second decisions based almost entirely on appearance. Bumble’s latest redesign, reportedly replacing the iconic swipe with an AI-mediated matching system known as “Bee” – moves to soften the hyper-visual logic of dating apps. But it also signals something more fundamental: the quiet erosion of a cultural framework that has long shaped how contemporary romance is performed, commodified, and consumed.
For over a decade, the swipe has been the organising principle of dating apps. It is deceptively simple in its immediacy – left, right, yes, no, desire, dismissal. Yet beneath that simplicity sits a highly engineered visual economy of attraction. People are reduced to images and encouraged to market themselves as recognisable brands.
The industry has long defended this model as democratising choice. Bios exist, of course. So do prompts, badges, curated statements of personality. But let’s not pretend they operate at equal weight. My research into dating cultures, including work examining how users engage with profile information, indicates that extended textual content is rarely central to initial selection. In practice, most users skim. Women, in particular, are more likely to return to bios post-match, using them retrospectively to justify or disqualify interest. But the first gate is almost always visual.
Swipe culture, then, is not just design, it is ideology made interface. As a firm anti-swipe, visual-economy advocate, it’s hard to imagine things getting much worse – but this decidedly does.
The introduction of Bumble’s AI layer, “Bee”, as it is being framed, attempts to interrupt this logic. Instead of users making rapid-fire visual decisions, the system will ostensibly generate matches through algorithmic interpretation of prompts, preferences, behavioural patterns, and conversational intent (the exact secret sauce is unknown). In theory, it asks users to articulate desire more explicitly. It shifts matching from reaction to inference, from image to narrative.
It is also, crucially, a form of automation, and automation always redistributes agency.
Bumble’s origins make this shift even more charged. When Bumble launched in 2014 under the leadership of Whitney Wolfe Herd, it positioned itself as a corrective to the masculinised norms of early swipe-era dating. As I’ve often argued, dating apps, predominately built by “brogrammers” have inbuilt inequities particularly directed towards women. Bumble’s signature innovation, women message first, was framed as feminist intervention. A structural rebalancing of digital courtship designed to reduce harassment and reconfigure power.
But the question has always lingered beneath the branding: did “women message first” redistribute power, or simply reassign labour? In many cases, it did not dismantle the expectation of initiation. It relocated it. Women became responsible not only for selection, but for sustaining the architecture of contact, yet another piece to the invisible load.
The introduction of Bee complicates that original premise. If AI begins to mediate who matches with whom, then even the limited agency embedded in swipe logic, the decisive yes or no, begins to evaporate. Selection becomes probabilistic and preference becomes modelled.
We are no longer choosing. We are being matched. Dating apps were, in their earliest marketing, positioned as the ultimate extension of that freedom. A limitless rolodex of potential partners – romantic democratisation at scale. But what they actually produced was a paradox: unprecedented choice, governed by increasingly rigid systems of visibility, desirability and algorithmic ranking. According to my research (and other corroborating research) they are also sites of abuse: particularly directed towards women and gender diverse group. One hundred percent of the women in my focus groups had accounted abusive behaviours on dating apps.
Now AI enters as the next structural shift, and unlike the swipe, which at least preserved the illusion of individual control, AI introduces a more opaque mediation layer. It does not simply present options, it constructs them. It also adds another layer of laziness: users no longer even need to muster the quiet swipe of a finger.
There is something disquieting about this evolution. If early romance culture moved love from family arrangement to individual choice, and swipe culture moved it from deliberation to instant judgement, then AI dating risks moving it again, from choice to curation without transparency. It raises a difficult question: what happens to romantic agency when even preference is outsourced?
There is also the feminist tension that Bumble has never fully resolved. Its founding promise was disruption: a women-centred alternative to male-dominated dating platforms. But feminist agency within app ecosystems has always been conditional. It exists inside infrastructure designed for engagement metrics, not liberation, within spaces created by “brogrammers” who have built in infrastructural inequity.
But, if Bee removes the swipe, it also removes the symbolic gesture that underpinned Bumble’s original identity: women choosing first. The irony is difficult to ignore. An app founded on the promise of empowering women in dating may now be moving toward a model where agency is further abstracted away from all users, regardless of gender.
If AI is doing the matching, then who is choosing at all?