■ The Age of AI
On October 21, Thailand time, the PHP development team officially announced two serious vulnerabilities in PHP (Hypertext Preprocessor, formerly Personal Home Page) and released patched versions. AI systems, including GPT-5.6 Sol, reportedly played a major part in their discovery. Our firm’s website also needed attention, and we used an AI agent to carry out the fixes. We are surely not alone: in short, holes found by AI were plugged using AI. One can hardly say that humans use AI to obtain the best answers. It would be more accurate to say that AI is using humans. Quite a few commentators see this as “the beginning of the end.”
■ Professional Work That May Disappear
We often hear that accountants, lawyers and consultants will soon lose their jobs. Opinions differ sharply, but I believe that is true, at least for work that consists of providing answers anyone could find through research. AI can already give remarkably accurate answers about incorporating or closing a company in Thailand, obtaining work permits, monthly tax filing obligations and disclosures in financial statement notes. Just a few months ago, incorrect statutory citations were conspicuous; recently, they have become much less frequent. AI is no longer confined to research. It now offers suggestions even on matters involving professional judgment.
Some argue that only AI-native companies will survive. There is something to that view. Accounting software providers, for example, may be among the most exposed businesses in our industry. Much conventional software could soon lose its reason for existing. We are approaching an era in which anyone can easily use AI to build their own software that takes electronic data and prepares financial statements and tax returns.
What, then, will be the value of professionals? Ultimately, perhaps the question is how clients can distinguish professionals who genuinely add value. It is a difficult question. With a slight sense of futility, I would like to share what my own experience suggests.
■ Why Professionals Still Matter
First is the ability to frame precise questions, or prompts, that make effective use of AI.
Suppose a Thai company commissions a foreign company to manufacture a mold and pays for it. What are the VAT consequences? It sounds simple, but it is not. Is the transaction a purchase of goods or payment for a manufacturing service? Will the mold be imported into Thailand, or remain at the overseas manufacturer’s premises for use in producing components? An import raises VAT at customs, while a cross-border service may raise reverse-charge VAT issues. Withholding tax and permanent establishment (PE) questions may also arise. Opinions may well be divided even within the tax authorities. Even for professionals, understanding the complexity of this question demands a considerable level of expertise. A vague question to AI, without identifying the relevant issues, is unlikely to produce a chain of reasoning and legal support that satisfies everyone involved. For the time being, posing the right question may still require a specialist. Further advances in AI could, of course, change that.
The second consideration is where responsibility lies—in other words, trust.
Consider the thought experiment of someone being struck by a self-driving car. Who should be held responsible? AI itself cannot be held accountable, and establishing a manufacturer’s liability may not be straightforward. The victim’s demand for accountability can end up with nowhere to go. On reflection, accountability is closely linked to whether an explanation feels convincing and acceptable.
Imagine a sales ledger containing 100 transactions per page and 1,000 pages for one month. AI calculates a ten-digit total. Would an accounting employee really feel confident in that figure? Without a way to verify the calculation, it would be unreasonable to expect that person to take responsibility for the output. Even in a financial statement audit, which verifies computational accuracy, AI’s output alone would not constitute sufficient audit evidence.
Consider another example. Before a journey into space, thorough inspection of the spacecraft is obviously essential. If all you were told was, “AI inspected it using a 100% optimal method,” most people would hesitate to board. They would not be satisfied that it was safe, because they could not identify who stood behind the assurance. I believe that the presence of a human being who accepts responsibility for the information and judgment is an important foundation of trust. Demand for professionals who can integrate information, including AI output, exercise judgment and accept responsibility is unlikely to decline; it may well grow.
■ The Professional You Should Choose
How, then, should we choose a professional in the age of AI? My answer is simple: choose someone genuinely kind and approachable.
Professionals who dislike questions are by no means rare. Someone whose attitude says, “Please do not waste my time with questions,” can safely be regarded as providing little more than information that AI can already supply. Such a professional is becoming obsolete. To borrow the sentiment of the Japanese song “Kita Sakaba,” a little good-naturedness goes a long way. There are professionals who welcome questions such as, “What you said earlier seems inconsistent with what you are saying now,” or “Which statutory provision supports that conclusion?” Some even welcome a client’s decision to seek a second opinion on an important matter. To do so, a professional must first distinguish clearly between what they know and what they do not know. This is the Socratic recognition of one’s own ignorance. Such people, in my view, are the trustworthy professionals who do not evade responsibility in the age of AI. Put simply, they are professionals who know how much they do not know.
On a different note, I once witnessed an accountant attempting to obtain money from a client under the pretext of a bribe. That person was a senior partner at a major audit firm. Hardly an example of a decent, approachable professional. An organization may have a prestigious brand, but ultimately it comes down to the individual. Professionals worthy of such trust cannot simply be replaced by AI, and I suspect they are not especially numerous. Few would dispute that accuracy of information is especially critical where foreign legal and regulatory systems are concerned. Otherwise, a business may not know its true financial performance, may suffer major embezzlement or enormous fines, or, in some circumstances, its people may even face criminal imprisonment. That is why clients should seek professionals who genuinely care about them and know the limits of their own knowledge. Our firm will hold this principle dear and continue striving to be one of them.
Naomichi Katabe | 26 October 2026