Your heart’s aflutter. An Instagram influencer is staring into your eyes. While sipping on their teh tarik from the Hougang hawker centre, they smile coyly and whisper: “This is the best date I’ve ever been on”.

BEEP BEEP BEEP BEEP BEEP. The alarm goes off at the most inopportune of moments. And reality rushes in. You realise it was all a dream.

You groggily reach for your phone, trying to silence the ear-piercing sound that’s hammering your cochlea. The river of dopamine you were swimming in is gone, replaced by an excruciating sea of dread.

You need to get ready for work. You’ve got an important presentation today. A presentation that if you screw up, might just spell the end of your dreams for a promotion. Your boss is already unhappy with you. This can’t possibly end well.

In your desperation to temporarily escape the crushing weight of your life, you open up Instagram. You need a dopamine hit. You need it now.

And that’s where you see them again. And for a fleeting second, all is well in the world.

For some, this may sound familiar. But now, something new has been thrown into the mix: instead of fantasising about flesh and blood influencers, some may soon start fantasising about AI influencers.

Also known as virtual influencers, these AI influencers are the natural evolution of deepfakes. While deepfakes tweak reality by virtually grafting one person’s face on another’s torso, AI influencers are entirely virtual avatars created in our likeness, but without any of our defects. They are, in a way, the best version of us – the Platonic ideal of our species.

Instagram and TikTok are the perfect platforms for such superficial, virtual beings, and they are more prevalent than you think.

There are now dozens of AI influencers roaming cyberspace. Some of the most popular ones include Lu do Magalu (6.9m IG followers), Miquela (2.6m IG followers) and Thalasya Pov (459k IG followers).

Not to be outdone, Singapore’s Capitaland – one of Asia’s largest real estate developers – created its own AI influencer, Rae (23k IG followers), in 2022. Designed to enhance online customer engagement, Rae has since fronted campaigns with fashion heavyweights such as Gucci and Moschino.

In fact, there’s now even an AI beauty pageant – a world first – that is currently taking applications. According to an article in Forbes, these “AI-generated contestants will be judged on their looks and poise, as well as the technical skill that went into creating them. They’ll also be appraised for their online pull”.

I expect the AI influencer phenomenon to explode within the next decade. While the overwhelming majority of Instagram and TikTok content now stars actual people, this lead will be chipped away aggressively.

And why wouldn’t it be? AI influencers are indefatigable, infinitely customisable, vastly cheaper than human influencers and can be spun up at any time.

Once the technology to create AI influencers is sufficiently sophisticated and personalisable, most companies will craft their own AI influencers the way we can craft a website now with website builders, turning them into a commodity.

The “website builder” of AI influencers will rake in billions. This is all part of the increasing dematerialisation and digitisation of the world. When atoms are transmuted into bits, the cost of production collapses by orders of magnitude and its adoption skyrockets.

Examples of some of our most popular tools and institutions that have gone from atoms to bits in the past few decades include:

  1. Libraries to Google

  2. Audio and video CDs to Spotify and Netflix

  3. Physical books and newspapers to Kindle e-books and news portals

  4. Gold and physical cash to bitcoin and stablecoins

  5. Town halls to social media platforms

Similarly, while food, travel and gadget review influencers will still largely retain their sway, fashion and lifestyle influencers are likely to be largely supplanted by their AI counterparts.

As AI improves by leaps and bounds, the line between what’s real and what’s fake will blur even further. As lifelike as today’s AI influencers are, most of us can still tell them apart from a real person. Within the next decade, this will cease to be the case – AI influencers will look, speak and act like real humans.

Just as the vast majority of people today have no idea how the internet works, in a decade, the vast majority of people will have no idea who’s real and who’s virtual on social media. And it’s not our fault – we simply lack the faculty to be able to distinguish between them.

This is why social media platforms should step in to protect the average consumer. There are two ways this can be done – a centralised way and a decentralised way.

The centralised way would be for social media platforms to require that every post or video uploaded on their platforms be clearly watermarked to indicate that those featured are AI avatars, and not real people.

The decentralised way is to emulate what X (formerly Twitter) does – allowing its users to add a Community Note to posts which inform them of the veracity of the content. The Community Note that’s voted to be the most useful and/or truthful gets voted up and sits as a disclaimer at the bottom of the post.

Leading phone makers such as Apple, Samsung and Huawei should include an optional piece of software that slaps a note at the bottom of a photo or video to inform the user if AI influencers or deepfakes feature in them.

Software such as Sentinel and Intel’s FakeCatcher already do this, but they aren’t readily and easily available to the public. Smartphone manufacturers can change this by integrating similar software in their technology stack and bundling it with their phones.

However, it’s important to note that the data processing and analysis necessary to detect AI-enhanced media needs to be done locally on the smartphone (instead of being sent to the cloud) so there are no data privacy issues.

For its part, the Singapore government could create a ranking system based on which social media platform does the best job of ensuring users are informed of which influencers are human and which are AI. Such an effort would mirror Singapore’s E-commerce Marketplace Transaction Safety Ratings (TSR), which rates e-commerce platforms based on the anti-scam measures they have in place.

This might not be legally enforceable now but I suspect it’s just a matter of time before the US and EU force social media platforms to include this feature in their content moderation policies.

The genie’s out of the box. AI will only continue to get more sophisticated, increasingly erasing the boundary between the real and virtual. So it is imperative that ordinary Singaporeans are ready for this future – a future that’s racing towards us, whether we like it or not.

AI will only continue to get more sophisticated, increasingly erasing the boundary between the real and virtual, says contributor KathirguganAI will only continue to get more sophisticated, increasingly erasing the boundary between the real and virtual, says contributor Kathirgugan

AI will only continue to get more sophisticated, increasingly erasing the boundary between the real and virtual, says contributor Kathirgugan. (PHOTO: Getty Images)

Kathirgugan is an inventor and food robotics pioneer who has worked in Silicon Valley, Shenzhen and Singapore. He believes in the power of technology and capitalism to make the world a better place. All views expressed are the writer’s own.