By Dr Matthew Doré, Palliative Care Consultant in NI Hospice and Royal Victoria Hospital Belfast, Chair of the PCC

There are many differences between ‘real life’ and the virtual, or digital life (social media / internet personae / virtual projections). For me, one of the foremost is that our endearments and faults reside in a wider and holistic context in real life. We consciously and subconsciously filter the virtual life more effectively to create a more perceived ideal self-image of how we wish to come across.

Therefore, the more virtual our society becomes, and the more filtered virtual ‘makeup’ we put on, the more we project a perceived self-ideal and thus we become increasingly isolated from the wider context and lived reality. This, I suspect, increases our separation, anxiety and loneliness. (1)

A parallel technological revolution is now under way: the creation of artificial intelligence (AI) ‘companions’. Sci-fi, TV, books and films are a little ahead, describing anthropomorphic AI, such as ‘Jarvis’ in the Marvel Universe. We are longing for companionship as a consequence of technology and surprisingly look to technology for an answer. As we form more and more of an emotional attachment to our devices (e.g. phones), we also search for our own personal ‘friend’. Alexa, Siri etc. are early versions of this.

Two people talking: a happy robot and a lonely woman outside in the sun
Artificial Intelligence art generator – https://hotpot.ai/nft-generator?s=art-maker

The technology companies have realised this, but ironically they have also realised a ‘best friend’ doesn’t always do exactly what you want. Doesn’t just obey. There is a delicate balance of agreeableness versus suggestion. It seems the proposed solution to this dilemma is data collection, as these AI systems and algorithms collect as many data points as possible on your character, preferences etc., and this creates a model in which suggestions can be made.

This is clearly a very imperfect and artificial solution, as friendship is commonly predicated on more than just trying to please in a single direction. There is a whole myriad of motivations, an element of mutual learning, admiration, exploration, enthusiasm but also selfish motivators. An example would be in almost every new activity someone discovers it is through friendship opposed to isolated self-exploration. A friend can expose something that we were previously unaware of, or that we had in fact ignored.

Two people talking, a happy robot and a lonely older woman walking along a beach path

Because AI co-pilots (as I call the future trend) are unlikely to ever spontaneously create this wonder and enthusiasm in a topic, they will always be relegated to less than ‘friends’. Perhaps some may be useful in care homes, on hospital wards or in day care settings. Even if they do generate a topic or activity which the person finds enjoyable, the person can’t share in that activity beyond enquiry and platitudes.

Indeed, friendship has a whole remit of motivations, some selfish and some even opaque. Therefore designing an AI friend which doesn’t have these complex motivations and is designed to serve a single person with suggestions and witticisms, can never aspire to come close to the imperfections of friendship. For me they are more like the droids in Star Wars, ultimately always in a form of servitude. Thus, to create an ideal friend I suspect you need the ability for mutuality in the decision to be friends and interestingly, the threat of the ability to leave that friendship.

It is ironic a friend needs to be free to betray and disobey. I suspect we are unwilling (as well as unable) to create such an AI. Rather instead we are destined to create AI obedient pets, companions, or carers,  sold as ‘friendly companions’. But will they ever fulfil our true needs?

EAPC 2023 (18th World Congress of the European Assc for Palliative Care) in Rotterdam will be hosting a Digital Legacy Conference, where AI technology that can mimic deceased loved-ones will be a topic of conversation.

Artificial intelligence solving loneliness

References:
1) Buecker, S., Mund, M., Chwastek, S., Sostmann, M., & Luhmann, M. (2021). Is loneliness in emerging adults increasing over time? A preregistered cross-temporal meta-analysis and systematic review. Psychological Bulletin, 147(8), 787–805. https://doi.org/10.1037/bul0000332
2) Artificial Intelligence Art generator – https://hotpot.ai/nft-generator?s=art-maker