The ‘AI revolution’ is here. Could it soon be leading Friday prayers?
The advent of AI powered chatbots is forcing some British imams to think about how new technologies could benefit their communities — or lead them astray
–
“In the name of Allah, the Most Gracious, the Most Merciful,” the Islamic sermon began. “Brothers and sisters in Islam, today’s khutbah is about taqwa, which is often translated as ‘God-consciousness’ or ‘piety’.”
The sermon, or khutbah, was everything one might expect, but it wasn’t being delivered by an imam standing at the pulpit of a mosque in front of a congregation of worshippers. It was written by the AI-powered chatbot ChatGPT.
It usually takes Asim Khan, who’s in his mid-thirties and has been an imam for nearly a decade, half an hour to write a khutbah. For the chatbot he was playing around with, it took mere seconds.
Khan was left stunned, not just by the speed of the bot, but also because the khutbah was replete with a narrative arc, quotes from the Qur’an and traditions from the Prophet, or Hadith, and mirrored one he could have written and delivered himself. “It seemed like it was a human being on the other side,” said Khan, who shared the 10 minute-long sermon in a viral tweet.
ChatGPT, which has soared to prominence since its release last November, works by generating human-like text when prompted with a query.
Khan, who juggles his duties as an imam at Redbridge Islamic Centre in east London with working as a pharmacist, had first tried the bot by asking it a question about contraindications to certain medications, which are effects of drugs that may be harmful for a given individual. Impressed by its accuracy, he began to ask it other questions, including about Islamic teachings. Now he is contemplating how the new tool could be used to benefit his community.
But Khan would think twice before actually delivering an AI-generated khutbah to a Muslim crowd. He’s concerned about its accuracy on Islamic topics and worried it could give out incorrect information.
ChatGPT was built by OpenAI, which counts Tesla CEO Elon Musk and Palantir founder Peter Thiel among its investors. The bot has set the record for the fastest-growing consumer app in history, surpassing 100 million monthly active users in January. OpenAI’s valuation has skyrocketed since ChatGPT’s launch, from $14 billion in 2021 to a projected $29 billion after major technology companies became interested in incorporating it into their operations.
Recently, Microsoft invested $10 billion in ChatGPT, while other technology sectors, including financial investment, education and dating apps, are exploring ways they can utilise innovative AI tools.
Muslims around the world are also joining in. An Islamic college in Australia has been using it for lesson planning, while a mosque in Canada, the Kanata Muslim Association, has used it for a fundraising campaign.
While the rise of AI chatbots has taken the tech world by storm, its growing ubiquity is not without warning — not least because of the racist and sexist remarks they have produced in the past, as well as the social and political biases that older AI models have largely presented as factual. Microsoft’s “Tay” was taken offline in 2016 after users taught it to be racist and antisemitic, including by citing Hitler. Meta’s BlenderBot 3, released last summer, began espousing misinformation, conspiracy theories and pro-Trump content. Google’s LaMDA chatbot has been accused by a former engineer of being both sexist and racist, breaching the company’s own ethics codes.
Digital experts argue that Muslims should be using the applications if they want AI tools to be fairer and more accurate in their understanding of Islam
While acknowledging this, many Muslims see greater potential upsides to using AI technology, particularly within their own communities. Arfah Farooq, based in London, is the CEO and co-founder of an organisation that brings together Muslims in tech, called Muslamic Makers. She recently used ChatGPT to help her craft a post on her blog, which focuses on her professional and personal endeavours. In the short time she has been using it, the bot has replaced her husband as her proofreader.
“My husband’s job is now redundant because he used to be the chief proofreader,” Farooq said in a TikTok video she posted about using ChatGPT.
For Farooq, ChatGPT is useful when checking for grammatical errors that she sometimes misses because of a learning disability. “I always kind of need a bit of help with someone to look over my work,” Farooq said. “ChatGPT does a really, really good job.”
Farooq, who also works in venture capital, has been familiar with OpenAI’s chatbots for several years, and knows that past models associated Muslims with more negative stereotyping than ChatGPT appears to. “There’s always downsides to these kinds of tools, especially for Muslims,” she said. “If the information it’s feeding off of isn’t the most diverse, it can lead to biases.”
Farooq’s concerns about chatbots aren’t unfounded, especially about OpenAI’s older ones. Muslims have long faced disproportionate amounts of abuse and threats of violence via the use of chatbots. An older OpenAI model, GPT-3, launched in 2020 with a well-documented anti-Muslim bias. A 2021 study later found that it disproportionately associated Muslims with violence.
OpenAI’s latest model appears to filter out content with an anti-Muslim bias to a greater degree than previous iterations. However, The Intercept reported earlier this year that ChatGPT still exhibits an Islamophobic bias when given different prompts — such as when asked about which travellers pose a security risk, or suggesting that surveilling mosques be used as a counter-terrorism strategy.
OpenAI did not respond to Hyphen’s request for comment on the issue.
Digital experts argue that Muslims should consider using the applications if they want AI tools to be fairer and more accurate in their understanding of Islam and Muslims. According to Yusuf Celik, a professor of digital humanities and Islam at Vrije Universiteit Amsterdam in the Netherlands, chatbots like ChatGPT could be used to create research using materials by only Muslim authors who have written about Islam.
“It would say we’re a peace-loving, wonderful, spiritual people that are connected with God or something like that,” said the professor.
Celik added that while he is concerned about AI potentially disrupting the sense of community mosques provide, the important role of any imam is unlikely to erode any time soon. “That is something that AI has not surpassed yet: the imam still has a bodily presence,” he said.
That’s likely to be of some relief to Khan, the east London-based imam, who had jokingly replied “My career is over” to a Twitter user who had responded to his post about the ChatGPT khutbah.
Khan’s not yet sure how he will continue to use ChatGPT in the future, but he’s both curious and cautious about any innovations it may bring. “I don’t see any fundamental contradictions between Islam and this piece of technology,” he said. “The main thing is to use it for noble causes and to be aware of its pitfalls as well.”
Topics
Get the Hyphen weekly
Subscribe to Hyphen’s weekly round-up for insightful reportage, commentary and the latest arts and lifestyle coverage, from across the UK and Europe
