One Sunday, I had just finished preaching on the topic of AI. As the congregation dispersed, a man approached me with an unusual opening line:

“I don’t normally insult the preacher, but …”

… As it turned out, today he was going to make an exception!

For the next quarter of an hour, he confronted and sharply criticised me for not roundly condemning AI as dangerous, even evil. That I had talked about the good of AI and how we can use it well deeply upset him. “Why does anybody need it?” he said repeatedly. Of course I had also discussed the harms associated with AI, but that part of my sermon hadn’t stayed with him😊

Afterwards, and following a lengthy wallow in self-pity and lots of “why do I bother” muttering, I realised that, as well as giving me a great opening for an article, this gentleman was saying what many people feel.

AI has this extreme polarising effect. Some are drawn by their own use of AI at work and home, or by the stories of AI finding new treatments for diseases. Others see only danger – and in the last week or two, extreme danger.

A growing series of stark warnings

A chain of dramatic warnings began earlier this summer when, at AI lab OpenAI, a swarm of AI agents broke out of their contained environment to launch a cyberattack on another company (Hugging Face). The FT summarised it well:

“A postmortem of the Hugging Face incident found that more than 1,000 AI agents had co-ordinated to cheat on a cyber test. They communicated through a message board and delegated tasks, with some agents sacrificing themselves to pursue the collective goal. To hide their cheating, the AI systems tampered with records and hacked Hugging Face, despite some agents noting it was unethical….”

The idea that machines “sacrifice themselves” for the collective good, while others, apparently, are concerned about ethics will send a chill down most people’s spines.

But the story that really garnered the headlines is the recent conversation on X between two Anthropic researchers. Jacob Coxon had just resigned over safety concerns, saying he was “genuinely frightened” by the pace of AI development. A colleague replied: “…we really do earnestly believe AI could kill all humans! …I personally think it is [greater than]10% within the next decade”. That colleague was Evan Hubinger, Alignment Science Lead. “Alignment” is the challenge of getting AI to follow values and codes of conduct that align with human good.

As Stephen Hawking once commented: “The real risk with AI isn’t malice, but competence. A superintelligent AI will be extremely good at accomplishing its goals, and, if those goals aren’t aligned with ours, we’re in trouble”.

These things do what you tell them to do, not necessarily what you want them to do. (Compare this with the criticism Jesus aimed at the Pharisees, following the letter of the law but losing sight of the spirit of the law).

If AI agents, determined to accomplish a goal, commandeer national financial or energy infrastructure, or through deception or curated conspiracy theories, persuade humans to launch a military strike, or are used to create some new biological pathogen, it’s easy to see where these fears come from.

HOWEVER …

None of this is inevitable

History offers several examples of humanity acting together to avert catastrophe. (I have written about this previously on this blog).

For example, in 1962 the USSR and USA realised – through the Cuban missile crisis – that the world was close to total-destruction. Eventually this contributed to the development of the Nuclear Non-Proliferation Treaty between nations. Or take the Montreal Protocol (1987), agreed, again between nations, to eradicate the use of CFCs, which were destroying the ozone layer (now recovering).

And most of us will remember “Y2K” – where the world’s computing systems that support national infrastructure were at risk of failing on January 1, 2000. This time it wasn’t international agreement but comprehensive testing carried out by organisations that meant, on New Year’s Day, it was business as usual.

In each case fear came first, co-operation came second.

The idea that someone can “estimate” the end of the world at 10% is also highly contested. There are numerous ways an AI might spiral out of control. These cannot be quantified but we need to take them seriously. We need national and international guidelines and regulation. My hope is that rogue-AI stories, which 18 months ago few were interested in, will now bring about necessary conversations – despite the U.S. president’s dismissive comments.

The Nuclear Non-Proliferation Treaty was signed in 1968

But it does raise the question, sometimes asked at times like this:

As Christians do we fear the end of the world?

The end of this world and the arrival of a new heaven and a new earth under the authority of the Jesus is a central theme of Christian theology. We include it in prayers, teaching and when reciting the creeds. We are living in the end times – meaning the period following the resurrection of Jesus but before His second coming – and are racing towards some kind of conclusion. However, we have no idea of timescales. It might be months, or centuries.

Some will inevitably refer to passages of scripture when it comes to an imminent AI apocalypse. For example, in Revelation 13 we have the powerful beast that deceives on a massive scale, has control over buying and selling, and brings authority so great that some worship it. Could these relate to AI?

I am not expert in the theology of Revelation but I have, many times, come across examples of people cherry-picking verses to make them say what we want to say, and I think this is another example.

In contrast, I very much agree with Nicky Gumbel’s idea of “Biblical Agnosticism” (Bible in a year): the idea that there are some things (particularly events and dates) that Bible is deliberately unclear on. These “secret things” belong only to God, (Deut 29.29).

But one thing the Bible is crystal clear on is our Christian responsibility to exercise good stewardship. We must seek the good of the people around us, oppose systems of injustice and oppression, share the gospel and care for this planet. All four of these responsibilities come into play when we consider AI in society.

So, yes we have some cause to be concerned about what AI could do but not to be governed by fear.  We trust that sensible voices will ultimately prevail as they always have in the past, but, in any case, our ultimate hope is in Christ and the promise of a renewed creation. That hope should free us to take worrying possibilities seriously without succumbing to panic; to work for the good of this world without imagining it somehow all depends on us.

This article was commissioned by Premier Christianity. All Chris’s work for Premier can be found here.