Should an AI be your best friend? Even if you’ve never considered the possibility or think it’s an absurd idea, it’s worth contemplating, because in the near future, there will be artificial intelligence systems attempting to befriend you.
AI friends offer many benefits. They’re available any time (even at two o’clock in the morning) and they never get bored (unlike your human friends, who may tire of hearing the same complaints about your job or partner).
But, as convenient as an AI friend might be, it’s a friend with more costs than benefits.
One reason to be suspicious of AI companions is that you don’t know their motives – or, more accurately, the motives of those who control them. They may be programmed to secretly support a political candidate, sell gadgets or sow discord. And, while most of us find it difficult to sustain a false front for an extended period and would feel (and inadvertently reveal) significant guilt and discomfort if we hurt or betrayed another, a machine, having no mind, has no such compunctions.
Machines are also becoming more persuasive than most humans. A recent study showed that pro-vaccine arguments produced by ChatGPT were more effective than those used by government agencies – an ability that could be easily redirected to advance more nefarious positions.
Machine text is honed through innumerable interactions. You can see the process in the frequent changes to online headlines: back in the days of print-only newspapers, headline writers had one chance to write an attention-grabbing phrase; today, most publications constantly tweak their online text, watching how thousands of readers respond to the different variations and if a new version gets better results, in it goes.
Similarly, all aspects of artificial friend-bots can be polished and perfected based on the effectiveness of their interactions with various users.
But what about AI friends that are created as genuine companions, not secret salesmen or undercover disruptors? Could they be a solution to our crisis of loneliness? Friendship is good for you: research has shown that people with an active and engaging social life live longer and are healthier and happier. Might artificial friends be the social equivalent of vitamin pills – helpful supplements when the ideal balanced diet (or lively circle of great friends) is unavailable?
Here the concern centres on the mind – or lack of mind – of the other. Think about the everyday experience of seeing something that would amuse a friend and making a mental note to tell them about it. Why do you do that? Maybe they’re feeling down and you want to cheer them up. Or, perhaps you hope to impress them with your eye for the quirky or beautiful.
In any case, your concern is with their inner experience, with how they feel or what they think of you. If something has no mind – no inner experience, just the capacity to imitate the words and behaviours of those who can think and feel – can that entity really be considered a friend?
Society is bound by our caring about what others feel. Empathy is part of this: we can feel the joy, or sadness, that someone we care about experiences. It’s also important that we care what others think of us. Emotions such as shame or embarrassment can cause harm when they’re used harshly or coercively, but they’re also essential to keeping our society working – we ‘do the right thing’ and cooperate, at least in part, because we care how others perceive us.
And having others think highly of us, or simply enjoy our company and care about our friendship, can be tremendously uplifting. The key is that these others must be conscious beings, able to form impressions and experience emotions. An artificial entity can be designed to imitate the external signs of emotions (to smile, sigh or cry) but it doesn’t actually feel anything; it’s not conscious.
Does this mean you should avoid interacting with AIs? Not necessarily. Properly designed, they can offer many benefits. They can help people think through difficult situations, practise social skills and provide insightful feedback. But they can do these things without the pretence of consciousness and emotion.
Living, conscious friends can be unpredictable and demanding. But, even as we increasingly interact with ever more sophisticated AIs, we need to embrace the complications of truly two-way relationships and to care about, and be responsive to, the experience of others.
Read more: