AI use tells us less about literacy than it appears to

Young Danes now have substantial exposure to generative AI. Many have seen directly how quickly a plausible text can be produced or how convincingly an image can be altered.

That practical familiarity has value. I see this in my parallel work with AI content production: people become comfortable with generative systems surprisingly quickly once the technology enters an ordinary workflow.

Information discipline develops differently.

Someone who regularly generates AI images may still have no systematic method for checking the provenance of an image received through social media. Experience with a language model does not automatically create a habit of checking the evidence behind its answers.

Knowing where the generate button is located is a rather low threshold for AI literacy.

This distinction becomes more important once AI-generated material leaves the application where it was created. The recipient sees an output without its prompts or production history. Technical familiarity may help with the judgement that follows, but it cannot make that judgement on someone’s behalf.

Older Danes are already part of digital Denmark

Lower generative AI adoption among older citizens needs equally careful interpretation.

Older Danes already live within a society where public services, banking, communication and news consumption are extensively digital. Statistics Denmark data shows that 29 percent of people aged 75–79 had needed help with public digital self-service or Digital Post, compared with a population average of 19 percent.

For my work with Digitalt Fællesskab, this distinction is significant. Digital participation can coexist with uncertainty or a need for assistance in particular situations.

Generative AI introduces a newer difference. Someone may use MitID, read digital news and communicate online every day while having little practical experience of how a generative system constructs an image or produces a confident answer.

The relevant divide is therefore more specific than online versus offline. It concerns different degrees of exposure inside the same digital society.

A Danish experiment complicates the generational explanation

A representative Danish experiment with 2,091 participants tested something more demanding than AI adoption: whether people could distinguish real images from AI-generated ones.

Participants knew in advance that some images were synthetic. Overall, their performance was not substantially better than chance. Higher age was associated with poorer performance on this particular authenticity task, while ChatGPT experience and critical reflection were associated with better results.

There is an important boundary around that finding. It cannot establish that older Danes are generally more susceptible to misinformation. The experiment tested a specific form of image judgement.

I find the result interesting for another reason. Direct AI experience appears relevant, yet the participants as a whole still struggled. Awareness that synthetic media exists was clearly insufficient.

That makes the familiar generational categories less useful. High AI use cannot stand in for information judgement, while lower AI use tells us little about the wider experience a person brings to assessing a source or claim.

AI literacy is becoming relational

This is where my work across Digitalt Fællesskab and practical AI production starts to connect.

On one side, I see why digital access is an incomplete measure of inclusion. On the other, I see how quickly people can become technically comfortable with AI without developing equally strong verification habits.

The connection has changed how I think about AI literacy.

Part of it is technical knowledge: understanding what generative systems can produce and having enough direct experience to recognise their capabilities. But judgement also draws on experience developed outside AI itself. People assess sources, institutional context and plausibility through habits accumulated over years of consuming information.

Different generations have accumulated these experiences under different technological conditions.

A younger person may have considerable direct experience with generative systems but no consistent verification routine. An older person may have encountered generative AI far less often and bring a different history of dealing with public information.

The evidence does not tell us which combination produces better judgement. I do not think there is a clean answer yet.

That uncertainty is useful.

What different generations may see in the same information

Intergenerational digital activities often assume a transfer of technical knowledge from younger participants to older ones. That model makes sense when the problem is learning how to operate a service.

Information judgement is less tidy.

If younger and older citizens have different levels of AI exposure, they may notice different things when confronted with the same uncertain material. Practical familiarity with generative systems could make one type of signal visible. Experience with sources and public information could draw attention elsewhere.

Neither observation should automatically carry more authority because of the participant’s age.

This gives intergenerational work a more substantive role in digital inclusion. The interesting part is not the exchange between generations for its own sake. It is the possibility of making different reasoning processes visible when people examine the same information.

For Digitalt Fællesskab, that is where the generational AI gap becomes relevant.

The 77 percent and 15 percent figures describe very different levels of exposure to generative AI. The Danish image experiment suggests that exposure matters, but also that recognising synthetic material remains difficult even when people know they should be looking for it.

Digital inclusion therefore has another layer to deal with. Access remains part of the work. So does practical support. Generative AI adds the question of how people judge information when its origin is uncertain.

Younger and older Danes are arriving at that question with different information histories.

Understanding what happens when those histories meet may tell us more about AI literacy than measuring who uses AI most often.