Artificial intelligence (AI) is not just another wave of digital transformation — it is a powerful force fundamentally reshaping the global economy. While it's tempting to quantify its future effects, the truth is that no one really knows how deep or wide the impact will be. As digital transformation expert Professor Jason Davis put it at the INSEAD AI Forum Singapore earlier this year, “Please, don’t believe anyone predicting any number about what’s going to happen in jobs, because no one knows — it’s all (exaggerated).”
But what actually can be done is to observe things already unfolding, where the fault lines are, and how individuals, companies, and nations are beginning to respond to this seismic AI shift.
The white-collar shockwave: Knowledge workers are the target
The first and most immediate shockwave is hitting white-collar professionals. Contrary to conventional wisdom, AI’s early victims would not be factory workers or truck drivers, but knowledge workers — people in roles like software development, legal analysis, accounting, marketing, customer service, and teaching. According to Davis, highly educated, highly paid occupations are the most exposed to gen AI as these systems generate knowledge.
Software development, in particular, is already undergoing massive disruption. Tools like Cursor and Copilot are not just assisting coders — they’re beginning to approximate the capabilities of entire engineering teams. That makes coding a uniquely significant case: it becomes a multiplier.
From code to care: AI in the human professions
If coding was once thought safe, what about medicine? Surely the emotional and human elements of care are irreplaceable. That’s what Davis thought — until he read the data. In one study, researchers compared a generative AI chatbot to doctors not only on diagnostic quality but on empathy. The result: the AI chatbot outperformed human doctors on both counts.
This upends long-standing assumptions. “My wife is a doctor, and I remember telling her, don’t worry honey, your job is safe… your bedside manner is 99.9%, there’s no way ChatGPT is going to do better than you,” he recalled.
But literally the next day, the study showed otherwise. The reason, Davis explained, is that empathy is essentially a content-generation exercise. “You listen very carefully to what people tell you, all their problems and concerns, and you generate a response that is hopefully very empathetic....”
The top use case for gen AI today is therapy and companionship. Many people now turn to these systems not just for answers, but for support.
Financial services and the cost of falling behind
In industries like finance, the pressure to adopt AI is particularly intense. Davis, currently researching AI agents in the financial sector, described a growing belief among banks that any delay could be fatal. “If there is an inflection to AGI, even if you’re a bit ahead of your competitors, all your customers will notice — you will win,” he said.
The implication is clear: you cannot overinvest. In highly competitive, high-margin industries, firms that integrate AI early and deeply will simply outperform. And while many firms still hesitate, those who delay too long may find themselves structurally uncompetitive. The AI-enabled firm will not just be better at doing old things — it will create new things faster, personalise better, and operate more efficiently than its slower peers.
A global arms race: The geopolitics of AI development
While AI is a technological story, it is rapidly becoming a geopolitical one as well. Most leading-edge AI systems today — both research and application — emerge from the US, and more specifically, California, and even more narrowly, San Francisco. As Davis bluntly put it, “That’s what AI is right now. I won’t even just say West — I’ll be blunt. I’ll be specific: US. I’ll be more specific: California. I’ll be more specific: San Francisco.”
But the US is not the only player in the game. China is experiencing what Davis described as a “flowering of dynamism and research, ideas and talent”. With government sponsorship, national strategy, and rapidly advancing companies like DeepSeek, China is positioning itself to compete neck-and-neck with giants like OpenAI.
The broader implication? Nations are beginning to treat AI as a national security issue, with some comparing it to nuclear technology. “One perspective on this is that this is a technology potentially as powerful and threatening as nuclear weapons. And therefore, we might want to think about it in that guise,” he said.
Indeed, in the US, a “Manhattan Project for AI” is now underway, with executive orders signed and large-scale federal investments in safety and development. China’s corresponding initiatives may be less transparent, but they are certainly no less ambitious.
Two threats, not one: AI safety in a high-stakes world
Amidst all the talk of productivity and disruption, AI safety remains a looming concern. While the “paperclip" scenario — where AI takes over the world without caring for human welfare — is often the headline, Davis viewed it as a more immediate risk. “I’m much more worried about highly empowered human beings with mal-intent utilising this technology,” he said.
That’s where responsible development comes in. Companies like OpenAI and Anthropic, Davis observed, are investing enormous resources into monitoring and aligning capabilities. Much of this manifests in the kinds of questions AI systems refuse to answer, like how to build chemical weapons, he said. These filters are there because of active work from internal alignment and safety teams.
Meanwhile, Davis stressed that no one has all the answers as it is an evolving field with high uncertainty and little precedent.
Managing the impact: Three strategies
With so much uncertainty, how should companies respond? Here is one answer according to Davis. Some companies are going all in, aiming to become hyperscalers in AI infrastructure and model development. This route offers massive potential rewards, but it comes with high risk. Only a few will survive in a hyperscale world.
Others take the “wait and see” approach, delaying adoption to avoid investing in tools that may be outdated in six months. There’s a rational logic to this; the field evolves fast, and early adopters sometimes build on tech that quickly becomes obsolete.
Then there’s a third strategy — flexible experimentation. Companies like Tesla and Palantir exemplify this. Instead of betting everything on one platform, they build product systems that allow continuous iteration. Palantir, in particular, works by embedding AI into frontline workflows, identifying “superusers”, and diffusing capabilities from the ground up.
Individual agency: Try fast, learn faster
For individuals, the key is not waiting for perfect conditions — it’s acting now. As Davis highlighted through the “midwit meme,” sometimes both the smartest and the most naive act first, while those in the middle overthink and wait.
This ethos shapes how Davis now teaches entrepreneurship at INSEAD. Every student is required to build an AI app and agent as part of the course. No Power Point plans, no speculative decks — just build something.
The AI economy isn’t coming — it’s here
Whether you look at job displacement in white-collar industries, government-scale investments from the US and China, or the shifting strategies of global corporations, the signs are undeniable. But the story of AI’s disruption is not just about the power of machines — it’s about the choices we make in response. The future won’t be shaped by predictions. It will be shaped by those who act, experiment, and adapt — today.