We are now living in the age of AI. Algorithms shape how we search, shop, and discover. Recommendations arrive before we ask. Suggestions appear before we search. With AI, our everyday life tends to become frictionless.

Yet one question remains open: Can artificial intelligence create genuine experiences of serendipity in our daily lives?

In our recent study, we explored this question by listening to people who navigate AI-mediated environments daily: consumers in online marketplaces. Drawing on 123 diary accounts and follow-up interviews with 32 active online shoppers in China, we found that the answer is more complicated than a simple yes or no.

From happy accident to engineered encounters

For decades, serendipity has been understood as a happy accident. You stumble across an article that answers a question you have been wrestling with for months. A conversation takes an unexpected turn and opens up a new direction. A discovery finds you precisely when you were not looking for it.

In each case, the experience feels both surprising and valuable. The environment presents possibilities without deliberate alignment to your interests. You exercise personal judgement in recognising and extracting value from randomness.

AI-mediated environments complicate this picture. Intelligent systems track behaviour, infer preferences, and shape what appears before us. What looks like surprise may already have been calculated. What feels meaningful may serve more for platforms.

So in a world where surprise can be engineered and value may be commercially aligned, can genuine serendipity still exist?

The surprise part? Algorithms amplify it

To answer that question, let's look at whether AI can deliver on the two core elements of serendipity (i.e. surprise and value). Start with surprise.

At first glance, engineered encounters seem incompatible with genuine surprise. If an algorithm is predicting you, how can it still surprise you?

Our findings challenge this assumption.

Participants described engineered encounters as more surprising than stumbling upon something purely by chance, referring to AI-powered shopping platforms as "treasure troves". Also, the frequency of bumping into these surprises tends to be much higher, with some participants referring to visiting AI-powered shopping platforms like "treasure hunting" every time.

What's going on?

The central reason is that surprise in everyday life does not require something completely alien. What catches us off guard is often a subtle twist on the familiar: a new style from a favourite brand, a variation of a dish you already enjoy. After all, as participants noted, everyday life is about seeking things that match our preferences to bring security and comfort. When something feels too innovative, it triggers uncertainty and discomfort, making us ignore it.

AI recommendations excel at striking this balance between known and novel. They know us via our digital footprints and the broader market in ways we don't, thus can easily surface the hidden treasures that we would have overlooked. Also, AI works like a black box. Even when participants knew AI was curating, they can hardly predict when or how surprise would land.

Thus, when it comes to surprise, AI does not just deliver. It amplifies, with intensified unexpectedness and encounters more frequent.

The value part? Less in the moment, more in hindsight

Value is where the picture becomes more complex.

AI systems are not neutral. They are designed to mainly serve platform and vendor interests, not necessarily yours. Thus, bumping into the engineered encounters can be a gamble, with both potential gains and real risks.

When algorithmic encounters land well, they feel genuinely valuable. Participants described discovering products they had not realised they needed, or items that perfectly matched tastes they struggled to articulate. These encounters improved everyday life in tangible ways.

More often, however, the outcomes felt manipulative. Some participants described making purchases in what one called an “emotional opium” state, where algorithm-fuelled excitement clouded rational judgement. Others found themselves trapped in browsing loops, spending far longer on the platform than they had intended. A few experienced something more troubling: after months of AI-curated discovery, they felt they had gradually lost the ability to search independently.

Yet the story doesn't end there.

Disappointing encounters often led to reflection, making participants more aware of their own misjudgements and of how platform incentives operate. Over time, they learned to engage with the AI system more strategically. Some participants even described deliberately gaming the AI system, steering it in ways that ultimately aligned with their own benifits.

Thus, when it comes to value, AI can still produce meaningful outcomes, but these are rarely transparent or instant. More often, value becomes visible only through reflection.

So, genuine serendipity still exists?

As we've seen, AI amplifies surprise and delivers value. So, for the very question "Can artificial intelligence create genuine moments of serendipity?", our answer is yes.

Yet something is altered. AI-engineered encounters differ in subtle but significant ways from traditional serendipity.

Traditional serendipity is relatively rare, arriving as an occasional gift from a world that was not designed with you in mind. Engineered encounters, by contrast, occur frequently. In one case, a single participant experienced sixteen engineered encounter within a month.

Yet value from these engineered encounters was less immediately transparent than in traditional serendipity. Value was often only recognisable in hindsight. The experiencer of these encounters may feel unsettled, even briefly diminished at first, before they finally come to appreciate what the encounter had to offer.

In short, what we have seen from our study is a form of serendipity that holds its core qualities yet unfolds differently. We call it artificial serendipity: an encounter that brings surprise and value, but jointly shaped by the agency of serendipitist (i.e. those who experiencing serendipity) and algorithm.