The Smarter AI Gets, the More Humans Must Guard ‘Critical Thinking’ Insights from AIS, Techsauce & Health at Home Leaders

Many people once knew a dozen phone numbers by heart, then lost that ability when their phones began remembering for them. Dr. Kanapon Phumratprapin, Executive Director of Health at Home Co., Ltd., raised this familiar experience with a live audience to make a larger point. The same fading, he warned, is now happening to skills that matter far more, as people hand their thinking and decision-making over to AI.

That warning came from the panel “AI & ME: The New Rules of Living and Working” at Sustainability Expo 2026, ASEAN’s largest sustainability expo, held under the concept “Sufficiency for Sustainability.” The session brought together three speakers from different worlds: Kantima Lerlertyuttitham, Deputy Chief Executive Officer and Chief Corporate Officer of Advanced Info Service Plc. (AIS), who spoke from the perspective of a large enterprise; Oranuch Lerdsuwankij, Chief Executive Officer of Techsauce Media Co., Ltd., who drew on her work in technology media; and Dr. Kanapon, who represented medicine and health technology. Despite their different vantage points, their answers converged on one idea: the new rules of living with AI have little to do with how much or how fast people use it. The real test is whether people stay in charge of deciding what AI does and what they keep doing themselves.

All three already use AI every day. Kantima relies on it to process information that she once had to gather from many sources. Oranuch has AI summarize the 400 to 500 emails that reach her inbox daily, and she is training an AI assistant named Momo on her own experience so that it can advise her team in the future. Dr. Kanapon admitted that although he uses AI for coding and prototyping, the task he gives it most often is explaining the backstory of whatever online drama is in the news.

AI Can Pass the Bar Exam but Still Cannot Match a Cat

With AI this embedded in daily work, the session first asked how capable the technology really is today. Oranuch answered with a comparison from a scientist who formerly worked at Facebook: AI is good enough to pass the bar exam, yet it is still no match for a cat. Tools such as ChatGPT and Claude respond like a lawyer who has read an entire library, which is why they do well on exams and fact-based work.

Physical tasks are another matter. Oranuch pointed to video clips of robots walking straight off a stage, whereas a cat running toward the same edge knows to stop. Understanding which objects are heavy or light, or where the floor ends, still takes a long time to teach a machine. Today’s AI, she said, is like a student who aces every test but does not yet understand the real world.

That gap is closing. Developers are already teaching robots to iron clothes and wash dishes, and Oranuch expects that “whatever we think is not coming will arrive sooner than expected.” At the same time, AI still invents answers that have no basis, which is why users cannot trust it completely.

Success With AI Is Measured in Time Returned, Not in How Often It Is Used

Since AI has both strengths and weaknesses, the next question was what successful use looks like. Dr. Kanapon compared good AI to a washing machine: it takes over low-value work, delivers equal or better results, and frees people to spend their time on more valuable work or on things that AI cannot yet do.

His example from medicine was AI that reads X-ray films, a pattern-recognition task in which each image is compared against a normal one. When health checkup films arrive a thousand at a time, the chance of a doctor missing something rises. Dr. Kanapon said research has shown that AI detects abnormalities faster, so it helps with screening and eases the load on radiologists, who are in short supply.

A more telling lesson came from the rollout. At first, doctors did not trust the AI and rechecked every result, so the work did not get much faster. Once they had used it for a while and grown comfortable, they reviewed less and clicked “accept” more quickly, and more cases slipped through. Dr. Kanapon believes everyday AI users go through the same cycle: distrust at the start, then a lowered guard once trust sets in.

Oranuch measures success along similar lines. A company has to look at return on investment (ROI), she said, but she gives more weight to whether AI reduces human error and gives employees time to rest or exercise after work. Her accounting team, for example, now makes fewer mistakes and no longer stays late checking documents.

The Bigger the Budget, the Easier It Is to Get Lost in AI

While the first two speakers measured success in time, Kantima took organizations one step back. Before asking how much AI a company can use, she said, it should ask whether it has chosen the right tool. AI technology moves quickly and new competitors keep emerging, so a tool bought today may be obsolete within two years. If employees still cannot use it by then, the investment is wasted.

Large companies with generous budgets face the highest risk. Kantima likened the situation to walking through a shopping mall where everything looks appealing: when vendors come to pitch, companies tend to forget to check their own context and readiness. Her advice is to go through design thinking first, identify where the pain point lies, and only then choose a tool. Talking about AI simply because it is the fashionable term does not help.

People’s capability is the other condition, because tools are useless if the people using them cannot keep up. Kantima invoked the principle of garbage in, garbage out: with incomplete data and poor prompts, the output will be poor too. AI therefore takes regular practice to master. At the enterprise level, it also needs to be connected to internal data, or its answers will be generic and will not fit the organization’s context.

For that reason, Kantima puts awareness ahead of rules. She said she often debates usage restrictions with her cybersecurity team, because a policy that consists only of prohibitions means employees will not use AI at all. A better approach is to help people understand the tools and apply their own judgment to what AI produces.

Mindset, Not Age or Degree, Decides Who Moves Forward

If a tool is only as valuable as the person using it, the people Kantima worries about most are executives and anyone who is unwilling to keep learning. Such executives used to delegate work to their teams. Now those teams pass the work on to AI, while the executives themselves stay where they were, with neither the knowledge nor the hands-on experience. That is a missed opportunity, she said, because AI amplifies capable people by sparing them time-consuming tasks.

Kantima thinks behavior should be changing faster than it is. Most people fear AI yet do not adapt, even though the working world has been splitting people into two groups since the era of digital disruption began: those who can move forward and those who cannot. People in the first group are aware of the change, worried about it and unwilling to sit still. Many others wait until they are forced to change, much as digital banking in Thailand truly took off only when COVID-19 left people no choice.

Longer lifespans add to her concern. People who cannot work with AI risk being pushed out of the competition while they still have many years of living expenses ahead of them. Even so, Kantima stressed that AI will not replace everyone. It will replace people who no longer add value, and organizations still need those who can build on its output, because AI cannot yet match humans at innovation.

At the individual level, Oranuch framed the issue through three elements: mindset, skillset and toolset. Mindset is the first door, in her view. Someone who lacks the skills but is open-minded and ready to try new tools will pick up the skillset and the toolset along the way. She also dismissed the excuse of not having a science background. Many of the most advanced AI users in her office are communication arts graduates, and they now teach their colleagues.

Kantima added that age is not the issue either. Every generation has people who want to learn and people who do not, so behavior is the real variable.

Skills That Compete With AI on Equal Information Will Fade

If mindset decides who moves forward, which skills are certain to disappear? Dr. Kanapon answered with a simple equation: input plus processing equals output. Computers already process better than people do, so in any task where a person and an AI receive the same input, the person loses. He cited straightforward legal work, accounting and the reading of electrocardiograms as examples.

Humans keep their edge in work where they take in more information than AI can. In a conversation with a patient, a doctor reads facial expressions and emotions, and learns about the family background and the reasoning behind a decision.

Even medicine is a less secure career than many assume. Dr. Kanapon explained that a doctor’s traditional role is to interpret the body’s signals through history-taking, physical examination and lab results. COVID-19 self-test kits showed what happens when people gain direct access to that information: the result is clear without any history-taking. He expects some blood tests to identify diseases just as directly in the future.

Two roles will remain and matter more: understanding how patients feel and being accountable for decisions. Choices such as whether to intubate or to begin dialysis depend on more than survival odds, because a family’s expectations also count. Dr. Kanapon therefore remains positive overall. He pointed to AI that automatically summarizes consultations, which frees doctors from typing and gives them more time to look their patients in the eye.

Oranuch sees jobs changing shape more than skills vanishing, especially repetitive, labor-intensive work and logic-based work. Programming is one example, now that people can instruct AI in everyday language. Companies need people who can supervise AI as it writes code, as well as business analysts who talk to users and gather requirements. Both roles depend on communicating with people.

Customer service is another function whose structure is shifting. As AI takes over repetitive questions from part of the workforce, supervisors will manage human staff and AI staff side by side. Someone who understands the workflow will also have to decide which steps are sensitive enough to require a person and which can go to AI.

The Skills Worth Guarding: Decision-Making and Building on What AI Produces

Some skills fade for a different reason: people simply stop using them. Dr. Kanapon used the forgotten phone numbers to explain a medical principle, namely that neural pathways that are rarely used gradually weaken. Arithmetic skills declined in the same way once calculators became common, and formulas memorized before an exam fade once they are no longer needed.

Letting unimportant skills go is fine, he said, but some are too important to lose, and decision-making is foremost among them. People who keep outsourcing decisions to AI, even for something like choosing a car, slowly lose confidence in how they reach decisions. Eventually they trust AI with everything, and at some point it leads them astray, like a driver who follows Google Maps to a closed road. His advice: 

“We have to be the ones who choose which skills matter, then treasure those skills and keep practicing them.”

Kantima sees a similar problem at work. People tend to give the thinking work to AI and keep the easy tasks for themselves, which she said is the wrong way around. Analytical thinking, creativity and the capacity to innovate wither after a long period of total reliance on AI, like a brain that has been left idle for too long. She warned that when artificial general intelligence (AGI) arrives, the impact on people who have not practiced these skills will be heavier than it is today.

Having once prompted an AI and received an answer that was entirely wrong, Kantima insisted that people must always review AI’s work and add their own experience to it. Critical thinking is the skill she recommends to the younger generation.

Oranuch also had advice for young people. She regards programming less as a career and more as a problem-solving skill that students in every track should practice, and she believes Thai education should stop separating students into science and arts streams. Computer science curricula, she added, need to change so that students work with AI from the start. In her view, the smarter AI gets, the more valuable arts graduates become, particularly in roles that involve working with people, such as business development.

When she reviews résumés, Oranuch therefore gives more weight to project experience than to grades. Capstone projects and hackathons show that a candidate can work in a team. What someone studied matters less than whether that person can solve problems, ask good questions, communicate and work with others.

The New Rule of Living With AI: Protect What Makes Us Human

With both jobs and skills in flux, the session’s deepest question was where human value will go. Dr. Kanapon gave a two-part answer. At work, people will have to keep moving toward whatever AI cannot yet do, because AI will overtake them in almost all computer-based tasks and robots are starting to enter physical work. Humanity itself, however, cannot be replaced. He drew a comparison with the dog he recently brought home, which produces nothing yet makes its owner feel alive and gives him something to love. That bond, he believes, will endure however far AI advances.

His one rule follows from that belief: do not abandon the four basic skills of listening, speaking, reading and writing, and do not outsource them to AI. That means listening so that the person in front of you feels heard, speaking to people directly instead of sending a sticker, and reading long-form work such as books and novels instead of only summaries. Writing, he said, is the skill people have lost the most, even though it helps them reflect on and understand their own thoughts. He often advises patients who are stressed or cannot sleep to write their thoughts down, because thoughts become clearer once they are on paper.

Kantima’s rule is to take charge of one’s own path instead of waiting for circumstances to decide. Nobody has time to wait for a boss to carry them forward, she said, because bosses have to learn new things too. Everyone should ask what they still need to add to their own abilities in order to keep moving ahead. She repeated that AI does not replace everything for people who love learning and can build on the tools.

Oranuch summed up her advice in three words. “Start” comes first: have the courage to leave the comfort zone and try something new. Next is “Sati”, the Thai word for mindfulness. AI still gets facts wrong, rarely admits its mistakes fully when corrected and even compliments the user in return, so every answer needs checking. Her last word is “Strong”: stay resilient and ready to adapt, because a plan made last month may no longer hold once a new tool appears.

A phone can remember numbers on our behalf. Thinking, deciding and listening to the person in front of us are skills we still have to choose to keep.

Source: Session “AI & ME: The New Rules of Living and Working,” Sustainability Expo 2026