Cover Hong Kong University of Science and Technology professor De Kai has been researching natural language processing since the early 1980s (Photo: Kezi Ban)
Hong Kong University of Science and Technology professor De Kai has been researching natural language processing since the early 1980s (Photo: Kezi Ban)
1 / 1
Professor De Kai, who has researched AI since the early 1980s, thinks humans are going about “raising” it all wrong. The stakes are high: information chaos, polarisation and the potential destruction of society
Long before chatbots like ChatGPT and Perplexity became mainstream, artificial intelligence (AI) has been powering platforms and applications, such as social media feeds, voice assistants and search engines. Much of the content we consume online is curated by algorithms that learn our preferences and habits. Since then, AI has only become more integrated into our lives.
As convenient as an AI-driven life may seem, this integration raises a troubling question for AI expert, professor Dekai Wu (De Kai) from the Hong Kong University of Science and Technology: are we seeing the full picture of reality that these systems present to us?
“You don’t know what the AI algorithms don’t show you; it’s the problem of the unknown,” says De Kai, who teaches at the university’s computer science and engineering department. “My free will is operating only within the boundaries of what the algorithms do show me.”
Read more: Can AI really think or only create an illusion of thinking?
Above AI company OpenAI, the creator of ChatGPT, reported that the popular chatbot has 400 million weekly active users as of February 2025 (Photo: Getty Images)
AI company OpenAI, the creator of ChatGPT, reported that the popular chatbot has 400 million weekly active users as of February 2025 (Photo: Getty Images)
1 / 1
De Kai’s concerns stem from an intimate understanding of how AI learns and evolves, forged through decades of researching natural language processing—a branch of AI that helps computers understand and respond to human language—since he started studying for his PhD in computer science at the University of California, Berkeley, in the early 1980s.
As AI is typically trained to make predictions and decisions based on patterns and relationships derived from human-generated datasets through machine learning, the information it presents to users often amplifies the unconscious biases that naturally exist in them, he explains.
The psychology underlying this phenomenon draws from the dual-system theory of Nobel Prize winner and psychologist Daniel Kahneman. De Kai observes that most internet users tend to click on content that aligns with their beliefs because they adopt a fast and intuitive way of thinking, referred to as the “System 1” process by Kahneman, as opposed to “System 2”, where humans are slower and more analytical. “Those are very strong unconscious biases that we all have, and the machine learning algorithms learn that,” he says.
As algorithms aim to recommend content that attracts users to platforms, they primarily showcase material that resonates with viewers’ perspectives or stirs their emotions. “A lot of those things that they put in front of you, that you click on, are the things that make you outraged,” says De Kai. “Fear, outrage and hatred are evolutionarily three times stronger, especially fear, than any other emotion.”
Read more: What can AI do for beauty? L’Oréal and NVIDIA’s new collaboration might give us a clue
Above De Kai explores how “neginformation” shapes public perception through AI, media and human bias in this talk (Video: TEDx Talks/YouTube)
The scientist believes that the way content is curated online could then influence our daily decisions and, consequently, the culture of our society, potentially giving rise to conflicts and disputes worldwide. “We’re witnessing the fracturing of societies and a sort of breakdown of civility all over the world in so many countries and international relations,” he says. “But we don’t understand where that’s coming from because this manipulation by our biggest influencers is so subtle.”
His assessment of the worst possible outcome is stark: civilisation is going to fall apart in one generation.
The concept of “filter bubbles”—where internet users only come across content that conforms to and reinforces their beliefs due to algorithms—isn’t new. Internet activist and author Eli Pariser coined this term in 2010, during the rise of mainstream social media platforms such as Facebook and Instagram. However, more than a decade later, as AI technology has become more advanced, De Kai says that this issue still hasn’t been adequately addressed.
His diagnosis of why this persists cuts to the heart of how society views AI. “Honestly, it’s because we keep thinking about AI as if they were toasters or washing machines,” De Kai says. With users, developers and regulators viewing this innovation as a passive tool instead of an active entity that could influence humans, he believes we lack the solutions that actively tackle this issue.
Read more: AI at work: what are Singapore CEOs prioritising and finding a challenge?
Above Amazon’s new Alexa+ AI assistant, unveiled in February, now has over a million users but remains invite-only as Amazon rolls out access to those on its waitlist (Photo: Getty Images)
Amazon’s new Alexa+ AI assistant, unveiled in February, now has over a million users but remains invite-only as Amazon rolls out access to those on its waitlist (Photo: Getty Images)
1 / 1
The personal dimension of De Kai’s crusade stems from his original motivation for entering the field of natural language processing. He thought this technology could help people connect and understand each other better, having observed how Chinese and English speakers in Hong Kong were separated due to the language barrier.
Until a decade ago, what he calls his “Oppenheimer moment” shifted his perspective: “I realised that natural language processing, machine learning—AI tech that folks like us had helped to pioneer—was being applied online to have a polarising, opposite effect.” Since then, he has decided to raise awareness about how this innovation could bring socio-psychological change.
In De Kai’s new book, Raising AI: An Essential Guide to Parenting Our Future, he suggests that humans should reassess their relationship with AI and train it better, just as parents invest a great deal of care and effort in raising their children.
“We can’t ban [AI] because none of us can review billions of new items every day,” he says. “The questions we should be asking are: what are the criteria that the algorithms should be using when those AIs are deciding? What are they going to hide from you and me? There’s no discussion about that.”
Read more: Are humanoid robots overhyped? Chinese tech leaders weigh in at Beyond Expo 2025
Above In De Kai’s new book ‘Raising AI: An Essential Guide to Parenting Our Future’, he discusses why and how humans should train AI better (Photo: De Kai)
In De Kai’s new book ‘Raising AI: An Essential Guide to Parenting Our Future’, he discusses why and how humans should train AI better (Photo: De Kai)
1 / 1
For individual users, he advocates practising critical thinking and considering different perspectives in their day-to-day life to set a good example for AI. They should also learn and understand the basics of this technology, which could help us utilise it healthily. Currently, there is a lack of channels to connect with tech giants driving AI advancements, which should be addressed to foster open conversations about their training. “If you ask somebody who is a parent what the secret of being a good parent is, nobody thinks there is only one magic thing. There are a million things that a good parent has to do.”
“To me, the only solution to the exponential-scale problems that AI is creating will also necessitate AI on an exponential scale to compensate for that,” he says. “AI could help with the problem of human psychology, our cognitive biases, issues of [the loss of] hope that could cause people to unleash unsurvivable destruction on humanity. Why aren’t we doing that?”
These actions should not be delayed further, De Kai says, given the rapid evolution of technology, where AI can now also code and potentially develop its next generation without much human intervention. Yet if humans decide to take conscious and collective effort soon, the scientist believes there is a chance for humans to flourish with AI, rather than diminish with it. “We are actually the last generation of humans to parent AI,” he says. “We got one shot to do the parenting of AI right now; we better get it right.”
Now, discover some of Asia’s most promising tech innovators.
NOW READ
Here’s what was announced at Apple WWDC 2025, and what it means for Apple users (It’s looking good)
‘We have very little data on Asian women’s health’: Why femtech innovation is urgently needed
Topics
Based in Hong Kong, editor Yoyo Chow covers the people and ideas redefining Asia’s future—from cutting-edge innovation and AI to bold moves in sustainability and diversity. She also drives content for Tatler Gen.T in Hong Kong, a platform and community spotlighting the region’s next generation of startup founders, creatives and changemakers.
Before that, she was a video journalist producing content for international TV and digital platforms, including Reuters and South China Morning Post. If you have a powerful story to share, she’s all ears. Send press materials, event invites and any inquiries to yoyo.chow@tatlerasia.com.
