What AI needs to learn about dying and why it will save it

Those programming truthfulness can learn a lot from mortality.

Andrew Steane

Andrew Steane has been Professor of Physics at the University of Oxford since 2002, He is the author of Faithful to Science: The Role of Science in Religion.

An angel of death lays a hand of a humanioid robot that has died amid a data centre

A digital momento mori.

Nick Jones/midjourney.ai

Google got itself into some unusual hot water in recently when its Gemini generative AI software started putting out images that were not just implausible but downright unethical. The CEO Sundar Pichai has taken the situation in hand and I am sure it will improve. But before this episode it was already clear that currently available chat-bots, while impressive, are capable of generating misleading or fantastical responses and in fact they do this a lot. How to manage this?

Let’s use the initials ‘AI’ for artificial intelligence, leaving it open whether or not the term is entirely appropriate for the transformer and large language model (LLM) methods currently available. The problem is that the LLM approach causes chat-bots to generate both reasonable and well-supported statements and images, and also unsupported and fantastical (delusory and factually incorrect) statements and images, and this is done without signalling to the human user any guidance in telling which is which. The LLMs, as developed to date, have not been programmed in such a way as to pay attention to this issue. They are subject to the age-old problem of computer programming: garbage in, garbage out.

If, as a society, we advocate for greater attention to truthfulness in the outputs of AI, then software companies and programmers will try to bring it about. It might involve, for example, greater investment in electronic authentication methods. An image or document will have to have, embedded in its digital code, extra information serving to authenticate it by some agreed and hard-to-forge method. In the 2002 science fiction film Minority Report an example of this was included: the name of a person accused of a ‘pre-crime’ (in the terminology of the film) is inscribed on a wooden ball, so as to use the unique cellular structure of a given piece of hardwood as a form of data substrate that is near impossible to duplicate.

The questions we face with AI thus come close to some of those we face when dealing with one another as humans.

It is clear that a major issue in the future use of AI by humans will be the issue of trust and reasonable belief. On what basis will we be able to trust what AI asserts? If we are unable to check the reasoning process in a result claimed to be rational, how will be able to tell that it was in fact well-reasoned? If we only have an AI-generated output as evidence of something having happened in the past, how will we know whether it is factually correct?

Among the strategies that suggest themselves is the use of several independent AIs. If they are indeed independent and all propose the same answer to some matter of reasoning or of fact, then there is a prima facie case for increasing our degree of trust in the output. This will give rise to the meta-question: how can we tell that a given set of AIs are in fact independent? Perhaps they all were trained on a common faulty data set. Or perhaps they were able to communicate with each other and thus influence each other.

The questions we face with AI thus come close to some of those we face when dealing with one another as humans. We know humans in general are capable of both ignorance and deliberate deception. We manage this by building up degrees of trust based on whether or not people show behaviours that suggest they are trustworthy. This also involves the ability to recognize unique individuals over time, so that a case for trustworthiness can be built up over a sequence of observations. We also need to get a sense of one another's character in more general ways, so that we can tell if someone is showing a change in behaviour that might signal a change in their degree of trustworthiness.

In order to earn our trust, an AI too will have to be able to suffer and, perhaps, to die.

Issues of trust and of reasonable belief are very much grist to the mill of theology. The existing theological literature may have much that can be drawn upon to help us in this area. An item which strikes me as particularly noteworthy is the connection between suffering and loss and earning of trust, and the relation to mortality. In brief, a person you can trust is one who has ventured something of themselves on their pronouncements, such that they have something to lose if they prove to be untrustworthy. In a similar vein, a message which is costly to the messenger may be more valuable than a message which costs the messenger nothing. They have already staked something on their message. This implies they are working all the harder to exert their influence on you, for good or ill. (You will need to know them in other ways in order to determine which of good or ill is their intention.)

Mortality brings this issue of cost to a point of considerable sharpness. A person willing to die on behalf of what they claim certainly invests a lot in their contribution. They earn attention. It is not a guarantee of rationality or factual correctness, but it is a demonstration of commitment to a message. It signals a sense of importance attached to whatever has demanded this ultimate cost. Death becomes a form of bearing witness.

A thought-provoking implication of the above is that in order to earn our trust, an AI too will have to be able to suffer and, perhaps, to die.

In the case of human life, even if making a specific claim does not itself lead directly to one's own death, the very fact that we die lends added weight to all the choices we make and all the actions we take. For, together, they are our message and our contribution to the world, and they cannot be endlessly taken back and replaced. Death will curtail our opportunity to add anything else or qualify what we said before. The things we said and did show what we cared about whether we intended them to or not. This effect of death on the weightiness of our messages to one another might be called the weight of mortality.

In order for this kind of weight to become attached to the claims an AI may make, the coming death has to be clearly seen and understood beforehand by the AI, and the timescale must not be so long that the AI’s death is merely some nebulous idea in the far future. Also, although there may be some hope of new life beyond death it must not be a sure thing, or it must be such that it would be compromised if the AI were to knowingly lie, or fail to make an effort to be truthful. Only thus can the pronouncements of an AI earn the weight of mortality.

For as long as AI is not imbued with mortality and the ability to understand the implications of its own death, it will remain a useful tool as opposed to a valued partner. The AI you can trust is the AI reconciled to its own mortality.

Column

Character

Confession

Culture

Psychology

18 September 2024

8 min read

‘Yet All Shall Be Forgot?’ Saying sorry has never been more difficult

Acknowledging wrongdoing is vital for any society to flourish. So why do we find it so difficult to apologise, especially online?

Roger Bretherton

Roger is Associate Professor of Psychology, at the University of Lincoln. He is a UK accredited Clinical Psychologist.

On a street, two men confront each other face to face.

Darwin Boaventura on Unsplash.

People in the UK don’t like to apologise. At least that’s what a recent poll reported by the Daily Mail claims. Of a thousand British people surveyed, about forty percent of them claimed they didn’t like to apologise because they were never wrong! At least that’s what the headline said. When you actually look at the survey itself, things get a bit more nuanced. 18 per cent don’t feel ‘comfortable’ making an apology. 15 per cent don’t like admitting they’re wrong. 23 per cent feel embarrassed at the thought of apologising. Sorry does indeed seem to be the hardest word. And Elton John seems to be the hardest person to avoid quoting whenever these things come up. Which they do - a lot!

We shouldn’t really be that surprised by the findings of this study. Contrary to the popular belief that the world is divided between goodies and baddies, upstanding citizens and immoral rotters, the ethical picture is much more complex than that. The line between good and bad, as Russian dissident Aleksandr Solzhenitsyn noted, runs through people not between them. Many moral qualities like kindness, forgiveness, gratitude, humility and so on, are trait-like. There are relatively few pure saints and absolute villains, most of us linger in the muddy moral middle, neither exceptionally good nor reprehensibly evil. And this is what the survey indicates. Despite all our reservations about apologising, the average 20 to 50-year-old says sorry about three times a week, totting up an annual total of 150 apologies per year. We may not like apologising, but we get there in the end.

Unfortunately, it’s not as simple as all that. Because while we may apologise, we don’t always mean it. If the need to apologise is a spectrum it not only includes those who NEVER apologise, but also those who ALWAYS apologise. If the non-apologisers sit at one extreme, the super-apologisers dwell at the other. These are the people who over-use apology, who never stop apologising for their existence. According to this survey, 41 per cent of us are first to apologise whether or not we think we are in the wrong, and 38 per cent apologise without meaning it. Ever found yourself inexplicably blurting out a sorry to the person who bumped into you at the supermarket? or gratuitously apologising for your emotions in an attempt to appease the workplace bully who caused them? I have. If that’s you, please pull up a chair and join me at the table of compulsive and unnecessary apologies- assuming you can sit down without apologising for taking up the air space.

With the wisdom of age most of us will learn to let things lie. Which is to say we will learn to forgive. Which is also to say we will learn to accept apologies.

It does seem, from this survey at least, that people are a bit confused about the nature of apology. ‘Sorry’, is a necessary part of the social vocabulary that makes community life possible. To say sorry is to acknowledge that we are embedded within a rich social network upon which we rely for our existence and without which human life would be untenable. It belongs alongside other basic words like ‘please’ and ‘thank you’, that recognise our social dependence. This applies everywhere: at home, at school, in the office, down the high street, at church. When we say Please, we acknowledge that there are things we cannot do and cannot know without the help of others. When we say Thank You, we accept that even our greatest achievements were team efforts, not wholly down to us. And when we say Sorry, we accept that this community of trust, this web of promises and fulfilments, is fragile. We can act in ways that fray or even break the threads that connect us to others. Sometimes we don’t show up when we said we would. Sometimes we lie to avoid shame. Sometimes we take far more than we should from those who can’t afford to give. Sometimes we are rude, hurtful, even hateful. Saying sorry is the way we recognise, renew and repair our damaged connections to the people on which our lives depend.

One of the most interesting findings in forgiveness research is that as people get older they generally become more forgiving. Now we can all think of exceptions to this - we all know people who seem to have become bitter rather than better with age - but that’s not the rule of it. Most of us will mellow and become more tolerant as the years pass. Partly because the passing of time diminishes our energy for grudges and plotting petty retaliations. But mainly because the older we get the fewer friends we have left. If young adulthood is awash with weddings, then later life is filled with funerals. To put it bluntly, as we get older more people we know have died. We increasingly realise that our connections to family and friends are priceless and irreplaceable and hardly worth severing over minor grievances. With the wisdom of age most of us will learn to let things lie. Which is to say we will learn to forgive. Which is also to say we will learn to accept apologies.

Why say sorry if there is no hope of social connectedness? This seems to be the zero-sum game played out in our digital lives.

Get updates

This by contrast sheds some light on why it might be that some people (the maligned 40 per cent of the survey) simply do not apologise. Admittedly it is likely that the tendency to offer apology varies alongside other personality traits like Agreeableness- our general tendency to get along with people. Those high in Agreeableness are more sensitive to ruptures in their relationships and therefore more likely to resolve these with a well-timed apology. And given that women tend to score more highly than men in measures of agreeableness and social intelligence, it seems equally likely that the league of super-apologisers who say sorry too often (like me) is predominantly populated by women (unlike me). By contrast those who do not apologise are likely to be at the tough-minded end of the personality spectrum, more ferociously individualistic, less emotionally aware, and not particularly sensitive to the fabric of social life into which they are inescapably stitched.

The apologiser and the non-apologiser then inhabit different universes. If apology belongs to a social network that needs to be tended, then the refusal to ever apologise is to deny the relational fabric of human life. Why say sorry if there is no hope of social connectedness? This seems to be the zero-sum game played out in our digital lives. Anyone can trawl the elephant’s graveyard of our online history and find things we said or did in our least thoughtful moments. And if they do, no amount of apology seems sufficient to rectify the mistake. Online apologies cannot erase online offences. It’s hard to imagine a better system for teaching us the futility of saying sorry.

There‘s a timing issue too. Quite often people who do not like to apologise assume their apology will result in humiliation. If they admit to being wrong, they will be publicly shamed, not restored to connectedness but excommunicated. As a result, if they ever do get round to apologising, they do so reluctantly or halfheartedly or under duress or just way too late, and consequently receive exactly the kind of vicious reaction they assume apologies usually receive. It’s a self-fulfilling prophecy: if we believe our apologies will be met with hostility, we tend to apologise in ways that make hostility more likely. It’s no wonder some people don’t see saying sorry as a viable social strategy.

To confess is to acknowledge and turn from our self-absorption, distraction, ignorance, inconsistency and whatever else detunes us from this heavenly wavelength.

It is a pity, because for those who care to look apology can address the deepest needs of the human soul. Apology restores us to the human community, reweaves the threads of trust that connect us to family, friends, colleagues, and neighbours. It assumes there is an invisible world we can rely upon, in which we can place our faith, and to which saying sorry can restore us. This is not just the logic of social apology but also the logic of spiritual apology, or to use the more traditional term, confession.

Just as we seem to be confused about apology, we are also pretty confused about confession. For many of us it belongs to movies where gangsters seek forgiveness for heinous acts through the screen of a confessional booth. Or even worse to the humiliation of being forced to publicly reveal our most shameful character flaws. But these are caricatures.

Confession, like apology, ultimately belongs to a benevolent view of reality. A view suggesting that, at all times and in all places we are in the presence of an utterly attentive, absolutely constant and unfailingly loving God. A God who is closer to us than we are to ourselves. A God who cannot help doing whatever it takes to close the distance between us, whose gentle presence hugs the contours of our lives the way the sea hugs the shore. And this divine reality is so permanent, so consistent that, like white noise, we live in complete ignorance of it most of the time. We tend to think that we are here and God is elsewhere, but actually it is God who is here and we who are absentmindedly elsewhere.

In this universe we don’t confess in the hope that our abject humiliation might possibly eke out a morsel of compassion from an otherwise indifferent deity. No. When we confess we acknowledge that while God may be unfailingly aligned with us we are less so with Him. We don’t seem capable of flying in formation with Him. If He moves in straight lines, our lines waver. To confess is to acknowledge and turn from our self-absorption, distraction, ignorance, inconsistency and whatever else detunes us from this heavenly wavelength. If apology restores us to a wider social reality than confession restores us to the deepest reality of all.