Used with Caution – How Prospective Students Feel About AI Search

To what extent do prospective students use and trust information generated by AI? And is this now a part of their higher education research journey?

Data on the growing use of AI tools needs to be used with the same amount of nuance as the tools themselves, but the findings can be striking.

A recent report from SparkToro (experts in audience and channel research) reveals that 20% of Americans use dedicated AI search tools more than 10 times a month and 40% do so at least once a month, but that this use isn’t necessarily displacing traditional resources, such as search engines.

So what does this look like for higher education and specifically for marketing and student recruitment? In a world where a range of well-known models can quickly answer questions – and even make recommendations – about potential universities or programs, how much do potential students use and trust these sources? And do they supplement, or replace, the channels you’re used to?

Thanks to the size of Keystone’s audiences, we can begin to answer this question – drawing on survey data from over 1,700 responses in August 2025.

There’s more to come when we segment to explore AI use across different audiences and demographics, but for now I want to take a look at the headline findings.

Over 2/3 of prospective students use AI search tools at least monthly

…and 1/3 use them daily.

Our initial sample is smaller and composed differently to the SparkToro data, but the finding suggests the use of AI tools is even more prevalent amongst prospective students than it is amongst the general population. And remember, this is use specifically to find information on study opportunities.

The finding is plausible: if you’re actively exploring study opportunities you’re likely to search frequently and students will typically be amongst the demographics with higher ‘AI uptake’.

But the conclusion that AI is completely displacing other channels for this purpose would definitely be a hallucination.

But every other channel is used more

Here’s what the data looks like across a range of channels we asked about:

Prospective students use AI to learn about study opportunities slightly less often than they turn to social media, a fair bit less often than they consult university websites and far less often than they use Google (and similar search engines).

I think the differences in breakdown may also tell a story here. It makes sense to me that someone might be slightly more likely to use AI daily than to check university websites with the same frequency – especially if they’re seeking answers to more general informational queries. But ‘weekly’ and ‘monthly’ cadence probably represents more in-depth queries, ‘further down the funnel’ where the university wins out overall.

We can also see that AI is the most divisive option here. Around 1/3 of people say they ‘rarely’ or ‘never’ use this channel to find information on higher education opportunities: far higher than for general web search or university websites, though only slightly behind social media.

Will audiences rely on AI for study information?

We also asked how trustworthy prospective students found information from the same set of sources. And I think these results make a lot of sense:

The university website comes out comfortably on top (as well it should!) with 80% of respondents saying this information was ‘very’ or ‘extremely’ trustworthy. Your website is the ultimate source of truth for information about your programs, which is why we link to it across your listings on Keystone websites.

Web search is a close second on overall trust, but with fewer people scoring this as ‘extremely’ trustworthy. This is also logical. The modern web includes a range of information sources endeavouring to provide useful context and guidance for study options (including our own). Search engines such as Google are designed to surface the best of these, but students will need to evaluate those results themselves.

AI tools and social media are then a relatively distant third and fourth, with the former winning out slightly (and consistently). Social media may reveal the wisdom of the algorithm as much as the wisdom of the crowd (and both are variable); information from AI tools is also ‘crowd-sourced’ and the models themselves encourage users to scrutinise it.

How much is really changing?

The information sources prospective students have access to are expanding, but they aren’t fundamentally shifting yet. Our data agrees with other sources in suggesting that AI tools are used alongside, rather than in place of, other channels.

Crucially, the most trusted information about a university is still the information published on your own website or through trusted partners. This is where students should ultimately end up, wherever their journey starts.