*Note: This article discusses the ethical implications of using AI-enhanced and AI-generated imagery in historical storytelling, particularly on social media. All examples referenced are drawn from publicly accessible online posts or materials. No individual posters or specific groups are identified, and no accusation of bad faith is made. The intent is to encourage responsible use and clear labelling of AI-modified content in the public interest.

THESE DAYS, you may have come across gorgeous, nostalgia-inducing colour photos of old Penang in your social media feeds. These are not the grainy-looking, hand-tinted postcards of years past, with their unmistakable stray patches of grey showing through where the artist failed to paint over the underlying photo reproduction. No, these photographic masterpieces look like they were taken yesterday, their quality sometimes rivalling those taken by today’s cameras.

One particularly beguiling image of Fort Cornwallis—shared on a major Penang interest Facebook group with 100,000 members—shows its moat and yellow building, with the fort’s brown brick walls against Penang’s verdant hills. One could marvel at the amazing cameras the Victorians had back then to capture picture-perfect Penang. But take off those nostalgia-tinted glasses for a moment, and it becomes clear that things are not what they seem—this is actually a black-and-white photo colourised using artificial intelligence (AI) imaging techniques. [See Images 1b, 3 & 4]

Colour-Tinted Past

The original black-and-white photo was taken by the Scottish photographer John Thomson in 1865, just three years after colour photography became possible. Colour photographs, however, would, for most of the 19th and 20th centuries, remain extremely rare.

When AI-colourised images are shared without context, they risk misleading viewers into believing they are seeing authentic representations of the past. This distorts public understanding of history and undermines the work of historians, archivists and educators, who rely on visual records to convey the realities of bygone eras. While colourisation can make old photographs more accessible and emotionally resonant, doing so without transparency erodes trust, and may ultimately rewrite collective memory based on imagined hues rather than historical truth.

To be clear, colouring photos—also known as hand-tinting—has been practised since the dawn of photography. However, since late 2020, advances in AI imaging technology—led by Adobe, the makers of the well-known photo manipulation software Photoshop, among others—are so good that it is hard to tell if a photo has been digitally coloured. The technology is now so sophisticated that it poses an actual risk to our perceptions of the past. And there lies the ethical issue—the line between historical documentation and digital reinterpretation is increasingly blurred.

Another increasingly common form of AI-assisted enhancement is called upscaling—or as some might call it, the zooming in on or blowing up of an image. Enlarging an image the regular way would result in a bigger but blurred photo—detail is sacrificed. AI-assisted upscaling promises to enlarge a photo without any loss of detail, just like in spy movies. Upscaling is commonly known as super resolution or super zoom, a term popularised by Adobe. Photoshop’s AI relies on a “machine-learning model trained by analysing millions of images” to fill in extra image information— known as pixels—to make photos look sharper. This generally works well when an image is blown up to twice its size, but beyond that, it produces crisp images that, upon close inspection, contain inaccuracies such as garbled text or distorted faces. The inaccuracies worsen when an already upscaled photo is upscaled again, as is commonly done on social media. [See Image 6]

Then, there is also the more concerning case where images are purely made up by AI. By typing a few words, one can generate, say, a 19th-century-style photo of Fort Cornwallis with French troops in formation across the moat. [See Image 5]

Now, we’re faced with a difficult problem. Is the reader or social media follower expected to take up the enormous burden of fact-checking the photographs? Obviously, it is impractical and it will not happen. Take Thomson’s purported colour photo of Fort Cornwallis—of the thousands of people who saw it, based on the numerous comments from appreciative viewers, it appears that none realised it was a digitally colourised photo.

Fred Ritchin, former photo editor at The New York Times and now professor emeritus at New York University’s International Centre of Photography School, observes: “The burden on the reader has to be as light as possible because nobody’s going to do all that research. People may have 10 minutes a day to get the news. They have to go to work, make dinner, take care of the kids.”

“They need a credible source,” he adds.

Here are some tag-suggestions which I hope the various Penang interest groups would consider adopting as part of their code of conduct:

A Chat with Professor Oh Peng Ai

I recently had a chance to meet with a well-known historian, Oh Peng Ai, who is, of late, somewhat of a celebrity. This amazing chap is the rare kind of academic who specialises in the entire history of humanity, earning a deserved reputation as a know-it-all.

A question had always bothered me: when was the Kuan Im Teng—that quaint Chinese temple at Pitt Street—established? As we chatted on various topics regarding Penang’s fascinating history, I found an opportunity to bring up the topic: “When was the Chinese temple at Pitt Street in Penang built?” I asked.

“The Chinese temple located at Pitt Street in George Town, Penang is known as the Goddess of Mercy Temple (Kuan Yin Teng) or Kong Hock Keong. It was established in 1728, making it the oldest Buddhist temple in the state,” Oh replied, without a hint of hesitation.

“Initially, the temple was dedicated to Mazu, the sea goddess, to thank her for safe passage from China. However, by 1800, it transitioned to honouring Guan Yin, the Bodhisattva of Mercy, and became a neutral ground for the Cantonese and Hokkien communities.”

I was stunned, flabbergasted by the historian’s seemingly erudite answer.

“How did Pitt Street look when the temple was established?” I asked nervously.

“When the Goddess of Mercy Temple was established in 1728, Pitt Street in Penang—as we know it today—did not yet exist,” explained Oh. “The island was largely covered in tropical forest, with scattered Malay and possibly Siamese fishing or trading villages along the coast,” continued the ever-confident scholar.

The Chinese Room

Now, before some of you start writing angry emails to Penang Monthly or organise a protest against this wayward academic trying to push an alternative narrative of the founding of Penang, I suggest first going to the closest kopitiam for a glass of kopi O-peng to cool down. You see, the good professor is not responsible for his words. Prof. Oh Peng Ai is the creation of my literary licence, though ChatGPT—the famed AI chatbot by OpenAI—with whom I chatted is very real. [See Image 7] You too can—and probably have unknowingly—chatted with the good professor at OpenAI’s website: http://chatgpt.com

My conversation with ChatGPT illustrates the dangers of using AI as the sole source of reliable historical information. What ChatGPT answered was not an intentional lie per se—it is a hallucination. The machine has no way to tell fact from fiction. The answer, well-crafted and confident as it may seem, is nothing more than the most probable sounding reply the AI decides based on what it was trained on.

The philosopher, John Searle, best known for his work on the philosophy of mind and language, proposed a famous thought experiment in the 1980s to challenge the idea that computers could truly “understand” language. Known as the Chinese Room thought experiment, Searle imagines someone who doesn’t speak Chinese or know anything about the language locked in a room with a rulebook, manipulating Chinese characters according to instructions without understanding their meaning.

That is essentially what happened with “Professor Oh”—sophisticated pattern-matching that produces fluent responses without genuine understanding. Think of autocomplete on steroids. The historian and bestselling author, Yuval Harari, best known for his book Sapiens: A Brief History of Humankind, observes: “When information is in a complete free market, the vast majority of information becomes fiction, illusion or lies.”

“This is because... telling the truth is costly. On the other hand, creating fiction is inexpensive. If you want to write a truthful account of history, economics, physics, et cetera, you need to invest time, effort and money in gathering evidence and fact-checking. With fiction, however, you can simply write whatever you want.”

Today, with AI chatbots, you don’t even need to write social media posts yourself—there are numerous free services that will do it for you. Whatever your intentions may be, it really doesn’t matter. That is because your robot copywriter—with its perfect mastery of English—cannot tell truth from fiction (at least not yet). So, how does Penang keep its heritage real with this onslaught of easily generated “facts”?

A Call to the People of Penang

The sustainable solution lies not in complex regulation, but perhaps in community vigilance. Penang’s heritage and cultural interest groups—with their combined reach of hundreds of thousands of followers—hold the power to preserve historical authenticity in the digital age.

I hope this article will encourage my friends at various Penang heritage and interest groups to consider implementing AI-content policies, following consultations with their respective communities. Although we are still in the early stages of AI development, it is imperative to initiate discussions and eventually take action. It is important to educate members that AI cannot reliably perform fact-checking, and should not be regarded as an authoritative historical source. Consider this to be akin to community maintenance—just as we would not tolerate graffiti on our heritage buildings, we should not accept digital inaccuracies in our heritage discussions. The objective is not to ban AI entirely, but rather to utilise its advantages transparently, while preserving historical integrity. Perhaps these discussions could begin with informal member surveys or pinned posts, allowing communities to shape their own standards organically.

Make no mistake, AI is here to stay— we must adapt to it. So, by all means, use AI to assist in creating engaging and informative social media content. Use AI to brainstorm ideas, proofread and improve posts. Just remember, despite its marketing promises, it really cannot do original research yet. Its answers are remixes and summaries of what humans have written over the past five millennia. Sometimes, these remixes and summaries are confident but incorrect fabrications. Perhaps Microsoft’s name for its AI-chatbot—Co-pilot—is a good reminder that it is a tool, an assistant. Humans still need to be the pilot.

And yes, all this fact-checking will require effort and resources. But crowdsourcing through an engaged, historically conscious community makes this manageable. We are quite lucky in this sense; Penangites are passionate about our heritage. With a long-running tradition of volunteerism, harnessing this passion for truth and accountability should produce an army of digital guardians of historical facts.

As a Persian sage once said, “Truthfulness is the foundation of all human virtues.” If we want to protect the fabric of society and its virtues, the choice is clear: act now to preserve the integrity of our visual and written heritage, or watch as AI-generated fiction gradually replaces authentic historical memory.

Penang’s real story is remarkable enough—let’s keep it real.

REFERENCES

1. Besak, Daniel R. “Ethics in Photojournalism: Past, Present, and Future.” Master’s thesis, MIT, 2003. Accessed https://web.mit.edu/drb/Public/PhotoThesis/

2. Chan, Eric. “Pack More Megapixels into Your Photos with Adobe Super Resolution.” Accessed https://blog.adobe.com/en/publish/2021/03/10/from-the-acr-team-super-resolution

3. Effendi Rabbani, Shoghi. “The Advent of Divine Justice”, p. 27, 1938. Accessed https://www.gutenberg.org/files/19243/19243-pdf.pdf

4. Hawkins, Jeff. “A Thousand Brains: A New Theory of Intelligence”, 2021.

5. Matsushima, Michiaki. “Yuval Noah Harari: ‘How Do We Share the Planet with This New Superintelligence?’” Wired Magazine, April 1, 2025. Accessed https://www.wired.com/story/questions-answered-by-yuval-noahharari-for-wired-ai-artificial-intelligence-singularity/

6. Ritchin, Fred. “Q&A: Fred Ritchin on AI and the Threat to Photojournalism No One Is Talking About.” The Media Today – Columbia Journalism Review, March 1, 2023. Accessed https://www.cjr.org/the_media_today/fred_ritchin_ai_photojournalism.php

7. Rodness, Roshaya. “The Controversial History of Colorizing Black-and-White Photographs.” Published May 20, 2021. Accessed https://www.fastcompany.com/90638388/the-controversial-history-of-colorizing-black-and-white-photographs

8. Searle, John R. “Minds, Brains, and Programs.” The Behavioral and Brain Sciences 3 (1980).