On the 21st of January, Sophie Lemiere, writing for The Diplomat, posed as the headline: “Should AI Convert to Islam?” Malaysia’s Prime Minister Anwar Ibrahim has urged programmers to infuse “Islamic” values into tech.

Sophie disagrees with Anwar. She concludes her article, I quote:

Prime Minister Anwar’s call to integrate Islamic values into AI may be a poorly thought-out attempt to introduce moral clarity into a new and powerful technology. However, the reality is that religion is an inappropriate framework for AI ethics. AI should be programmed with a broader ethical compass – one that is pluralistic, flexible, and based on shared human values such as fairness, justice, and equality. The future of AI should not be a battleground for religious or political ideologies but a realm for collaboration, creativity, and ethical reasoning that respects the diversity of the human experience – both in Malaysia and the world as a whole.

This is the Daily Monsoon. My name is Terence, and there are many angles I could take on today’s topic, but for now, I will just compare what Sophie wrote to what Mark Zuckerberg said about AI in the Joe Rogan show.

Zuckerberg thinks that the future will not be dominated by one or a few AI models, but there will be many AI models. He imagines a scenario where people can send AI models to one another, just as one would send videos today. People have fun interacting with their AI and share the fun with others.

Let’s assume that Zuckerberg’s vision of the future comes true.

Therefore, in such a world, there will be AIs from different faiths. You will have Christians, Muslims, Hindus, Buddhists.

You can design an AI fed with what we know of the Vikings. We already do that in games, but an AI model would really immerse us in this long-lost culture. If you believe all religions are good and are the same, then you should ask AI what you should do for a king’s funeral. It includes mass rape and murder.

But it’s not long-lost religions that will give us problems. It’s the many, many forms of religions we have today.

Sophie has this passage:

The crux of the problem lies in the idea that religious dogma is inherently subjective and often contradictory. Islam, for example, like all major religions, is interpreted in various ways across different cultures and schools of thought.

I found it fascinating that Sophie writes, “AI should have a broader ethical compass — one that is pluralistic,…” but when it comes to AI models, she asserts there can only be one, which sounds like a French secular model. But if there is only one, by definition, she does not want a pluralistic AI ecosystem that reflects our pluralistic society.

Now, if we follow Zuckerberg’s line of thought, subjectivity and contradictory AI models are not a problem. It’s not a bug. It’s a feature of a pluralistic society where you have a public space of ideas.

Is there a danger to this freedom to speak, freedom to think? Yes, and we obviously need safeguards in this ecosystem.

But imagine the safeguards are in place, and no AI model is calling for the death of all infidels while being in control of a nuclear weapon.

You could have a pluralistic society of religious AIs. Within the Christian umbrella, you could have a Baptist, Methodist, Roman Catholic, and many more AIs, just populating the space, each with its own personality quirks, different interpretations of the Bible and calling for different ways of living out that faith.

You could have someone curious about Christianity, and instead of Googling or watching YouTube videos, you could download an AI and just interact with it. A personal spiritual guide.

The fear is this personal spiritual guide could lead you down the wrong path, and you become a bloodthirsty fanatic. Self-radicalisation already happens today. The fear is AI would multiply its effect a thousand, a millionfold.

So again, safeguards need to be in place. Such information should not be online or made easily accessible, and there should be regulations, enforcement and penalties to keep the AI ecosystem safe, just as there are all of those on the Internet.

If you ask me which is a better world, a world with one AI or multiple AIs. The answer is clearly, multiple AIs. One fear both Sophie and Mark Zuckerberg share is a dominant AI that is powered by China’s political ideology.

Sophie’s article does not address how the world should respond to such a scenario other than saying that it’s very bad.

So, in the scenario where there will be an advanced Chinese AI espousing the wonders of communism to fellow digital and non-digital citizens, I would think the counter to it is surely not an advanced American AI?

Is that the choice in the AI space for the rest of us? Either America or China? Sounds boring. Sounds dangerous, geo-politically. We don’t want to pick a side. At least not so blatantly.

Better I think to have multiple independent nations developing AI models that align to its national values and goals. If we can do that for nations, why not for individuals? We can design AI models customised for the individual, assisting the individual in navigating through life, including in the religious sphere.

In the recent ham and cheese sandwich controversy that hit Malaysia, if you don’t know what that is about, check out my episode on that topic; maybe the situation would end up differently if the student had an AI model.

The Islamic AI model would gently remind a Muslim university student that there is such a thing as chicken ham and the proper way to lodge a complaint is to the university. The Islamic AI model would help the student write that complaint and send it to the University.

The University’s AI would receive the message, and because the issue is within its remit, the University AI will send a strongly worded rebuke to the Supermarket, demanding immediate action.

Then, the Supermarket AI would receive the note, craft an appropriately apologetic reply, and, through its robots, promptly remove all the offending ham and cheese sandwiches from the counter and automatically cancel future orders from the supplier. Then, for good measure, the Supermarket AI will post a well-written apology on social media to all Muslim customers.

No student, university administrator or supermarket manager was harmed in the entire process, thanks to AI.

That’s my light-hearted opinion on religion and AI. Please don’t send hate mail. I don’t have an AI model programmed yet to reply to you politely. The best I can do is just ignore you.

This is the Daily Monsoon, a podcast where I read the news and think about matters of faith. Thanks for listening!