Usage of Large Language Models in the Context of Digital Exclusion: Older People’s Perspective

https://doi.org/10.63189/BRTD1348

Older people represent a particularly vulnerable demographic due to several factors, including a decline in physical and cognitive abilities typically associated with ageing, and relative difficulties in adapting to new technologies.

Contemporary circumstances create some new and largely unprecedented challenges from the perspective of older people. These challenges stem from at least two issues. The first consists of the ageing of society – the increase of the proportion of older people in the general population results in pressures to keep older people economically active. The second issue is related to rapid technological transformations, mainly in the context of digitalization, of economic and social relations. These transformations pose challenges for older people in terms of social inclusion, access to public as well as private services and economic activity. This means that older people may find it increasingly difficult to access crucial services such as social services and public healthcare. At the same time, while expected to work longer, they may find it increasingly challenging to find and retain employment.

Older people typically do not use digital tools with the same ease as young and middle-aged persons. As the use of digital tools in the economy becomes increasingly common, older people may thus find themselves at an additional disadvantage while striving to remain economically active. The mix of these issues, that is the digitalization of the economy, digitalization of access to public services and pressure to raise the retirement age, create particular challenges from the point of view of older people.

These challenges result in the need for the public authorities to provide support for older people in adjusting to new realities. This support should include maintaining a substantial degree of non-digital access to public services and supporting digital inclusion of older people. Providing the necessary support can be increasingly difficult as societal ageing may be accompanied by increasingly limited public resources. These limitations may stem from lower rates of GDP growth and lower public revenues as societal ageing results with shrinking labour pool, decreased tax bases and increasing spending pressures.

The category of older people may be divided into subcategories. The first subcategory encompasses the older people who may be able to remain at least partially professionally active. The second subcategory includes older people who may be unable to maintain professional activity, but who remain able to independently manage their private affairs. Finally, the third subcategory includes older people who are not capable of independently running their private affairs.

The notions of artificial intelligence and large language models (LLMs) introduce very promising and potentially significant developments to the situation of older people in the age of digital transformation. LLM is defined as ‘a deep-learning algorithm that uses massive amounts of parameters and training data to understand and predict text. This generative artificial intelligence-based model can perform a variety of natural language processing tasks outside of simple text generation, including revising and translating content’^[1]. LLMs seem to have the capability to be beneficial from the point of view of all the subcategories of older people outlined above.

These tools seem to have a significant potential in terms of limiting the digital divide between different generations. LLMs operate based on natural language and thus the barriers to entry, in terms of their usage, may be relatively low as they do not require being able to operate dedicated interfaces. Being able to conduct activities online based on natural language creates a potential to limit the digital exclusion of older people. While working with LLMs older people may be provided with interactive instructions and access to online services on the basis of natural language. LLMs also offer a potential of increasing workers’ productivity and thus mitigating some natural limitations faced by older people in the working environment. This may be particularly significant from the point of view of older people who remain professionally active. Furthermore, LLMs have the potential to improve the quality of medical services from the point of view of older persons^[2]. LLMs also have the potential in terms of addressing problems related to loneliness among this demographic^[3]. This may be particularly significant from the point of view of older people who struggle to remain independent and socially active and who are at risk of social exclusion.

LLMs may also cause challenges for older people as their use still requires certain level of digital skills. Furthermore, the increasing capabilities of LLMs are likely to further the gap between those who are able and willing to use them, and those who are not. The use of LLMs is also connected to certain concerns such as biases, toxic content, hallucinations, and privacy problems^[4]. Older people may be particularly prone to suffer as the result of these issues.

Older people should be encouraged and supported to adapt new technologies, including LLMs. This support, provided by public authorities, is particularly important in countries where older people are able to count on the support of their families to relatively limited degree^[5].

^[1] Brittanica, https://www.britannica.com/topic/large-language-model (accessed: 23.03.2025).

^[2] LLMs have the potential to augment clinical decisions by effectively deprescribing criteria for older adults with polypharmacy in emergency departments. See: Vimig Socrates et. al., 2025, p. 12.

^[3] At the moment significant challenges are being observed in this context including: “disruptions in conversations, including frequent interruptions, slow, repetitive, superficial, incoherent, and disengaging responses, language barriers, hallucinations, and outdated information, leading to frustration, confusion, and worry among older adults”. See: Ifran et al., 2025, abstract.

^[4] Patil and Gudivada, 2024, p. 38.

^[5] See: Mubarak and Suomi, 2022, abstract.