Why AI must adapt to a long-lived society. Ageism and technology in the workplace
The term ageism emerged in the 1960s to refer to age-related biases and prejudices. It was then that age discrimination began to be seen as a reality deeply rooted in cultural and social elements that are still relevant today. Based on the fact that life expectancy is increasing and the demographic transformation towards a long-lived society, the conception that someone is "old" according to "for which areas" is already, for obvious reasons in our society, anachronistic and non-functional.
Society is ageing and it is necessary to adapt each of its tools, including technological ones, so as not to perpetuate beliefs that disadvantage an ever-increasing number of people.
According to the recent World Health Organisation (WHO) report "Ageism Report", "ageism has been shown to be detrimental to health and well-being and is a major obstacle to effective policy formulation and action on healthy ageing".
A very positive aspect of technological advances is that they largely favour the automation of processes in any sector, something to which we are becoming more and more accustomed, being in most cases, harmless.
For this article we are going to take into account the senior public (over 65 years old), because not all seniors decide to opt for retirement or retire from the labour market - either for personal or economic reasons - but also all those people over 50 years old and we will talk about how artificial intelligence algorithms and machine learning affect them in the workplace where automatic software based on AI is increasingly used for the selection of CVs.
Several studies, such as the one carried out by the National Bureau of Economic Research, show that in personnel selection processes, algorithms have a very high percentage of being biased, discriminating and discarding all those profiles that are over a certain age, regardless of the candidate's skills.
Social aspects of ageism
As something integrated in society that we must overcome, we assume that people of a certain age lose abilities and that they are not able to adapt to changes. We assume that people with more than 20 years of work experience on their CVs cannot be suitable for roles that require a high degree of technological knowledge. According to this article published in the BBC, age bias is present in the selection process "and this means that the over-50s are more than twice as likely as other workers to be unemployed for two years or more if they lose their current job". Moreover, this bias rate has been found to be even more unfavourable for women aged 50 and over.
But what causes these biases and patterns in technology to occur?
Machine learning, data and ageism
The technology itself is created by people, and these people choose and train data based on cultural and social realities. And this is where the challenge begins. Machine learning algorithms are created from datasets, which store large amounts of information, "they learn what counts and what doesn't". To be fair, this is the same as saying that the datasets and the way the data used to train the models are processed are based on outdated social criteria and beliefs, and therefore perpetuate stereotypes. And it is these algorithms that are used in automatic selection programmes causing a high level of bias.
Another proven factor is the lack of senior professionals and women in the design of the technology itself, which ultimately affects the decision-making chain. Moreover, this factor determines and conditions not only the technological development itself, but also questions its applicability in different fields.
Measures are currently being taken to combat this type of prejudice from the outset, however, part of the change has to come from social change itself so that it can be scaled up ethically in the technological sphere. The graph below demonstrates the upward trend described above, despite the relative age of the data.
Graph: Source PayScale / Statista
What can we do to improve these factors?
Taking both social, economic and technological factors as a reference, we face a challenge that has already been echoed by organisations such as the World Health Organisation (WHO).
The published report "Ageism in Artificial Intelligence for health" refers to how AI can generate new forms of ageism. It also introduces a framework to begin to minimise the risk of age exclusion.
The WHO also sets out a number of considerations among which we highlight:
- "Participatory design of AI technologies by and with older people".
- "Age-inclusive data collection".
- "Governance and regulatory frameworks for empowering and working with older people".
- "Robust ethical processes"
In addition to these considerations and measures, another challenge and objective is to improve the perception of qualified senior staff (which we could summarise as experience) in the technology field and sector. A study carried out by Hacker Rank, which is alluded to in this PayScale article, confirms that older staff are even more flexible when it comes to learning e.g. programming languages, compared to younger employees.
Another tool at our disposal is to improve employment policies regarding ageism and to start a progressive shift towards a full integration of AI in a long-lived society.
It is clear that bringing together the social, economic and political components of the challenge is very complex, but it is nonetheless very positive that we continue to contribute to an ethical and responsible evolution of technology.
