Artificial intelligence is reshaping the landscape of knowledge-based professions. Its use now spans financial consulting, auditing and legal advice. Individuals have begun entrusting asset management to AI, while companies are demanding fee cuts from professionals in proportion to the work AI now handles. Yet as more people use AI, its shortcomings are becoming just as clear.
More than half of U.K. investors think it is acceptable to entrust their own money matters to AI, Bloomberg reported on the 2nd. In a survey released the same day by Britain's Financial Conduct Authority, more than half of respondents said they trust AI tools for help with financial matters. That exceeded the share who said they trust television and radio presenters, and was more than double the share who said they trust influencers.
About 9% of U.K. adults receive professional advice on pensions or investments. Among the remaining 7 million people holding more than 10,000 pounds (about 20 million won) in cash, 24% said they do not invest because they do not know enough, and 12% said they feel overwhelmed by having too many options.
Against that backdrop, getting financial guidance from AI clearly helps these people. Bloomberg columnist Merryn Somerset Webb said most people do not need a complex asset structure and do not need tailored advice. Andrew Lo, a professor at the MIT Sloan School of Management, said AI is quite good at providing reliable advice, and that the latest models deliver something close to individualized advice along with learned empathy.
The drawbacks are equally clear. Because large language models learn from data that exists in the world, they are weak on taxes, mathematical precision and nuanced judgment.
There is also a concentration problem. Asked how to invest 100,000 pounds, AI often recommends putting 60% to 70% of the equity allocation into an MSCI World exchange-traded fund. Following that advice would push the U.S. share of a portfolio to 70% and leave it heavily tilted toward technology stocks.
AJ Bell, a platform operator, said everyone's circumstances differ and there is a risk that AI systems build assumptions based on broad, generic situations. Risk appetite and investment horizons can differ completely from person to person, the firm said.
Somerset Webb stressed that it is essential to ask a large language model for counterarguments, ask what assumptions it made, cross-check across multiple models and then verify the answers independently.
The fact that users' financial information may be used to train future models also raises the prospect of privacy disputes. If AI uses sensitive personal financial information as training data, the consequences include data leakage, misuse and infringement of data sovereignty. Half of the respondents in the FCA survey believed that information generated by AI is subject to regulation, but in reality it is not.
Audit Fee Growth Stalls as Clients Demand AI Discounts
Growth in the audit fees companies pay accounting firms has slowed markedly, according to the Financial Times. Fees had been rising 4% to 5% a year, but last year's increase came in below 2%, effectively in line with inflation. The FT concluded that the era of steadily rising audit fees is drawing to a close.
The reason is that clients have demanded firm price cuts matching the share of work now handled by AI. Accounting firms are doing the same themselves. KPMG International, one of the Big Four, cut the fees it pays its own auditor, Grant Thornton UK, by 14%, arguing that AI had lowered costs and it should therefore pay less.
Still, it may take time for AI's effects to show up in actual numbers. The burden grows especially during periods when new systems run in parallel with existing ones. It also remains the case that "human intervention" is still essential in high-risk areas such as catching errors in corporate figures. The FT noted that Microsoft co-founder Bill Gates has pointed out it took 20 years for the effects of the personal computer to appear in the workplace, and said that while AI will be much faster, it is still at an experimental stage for most users. Even so, the direction of audit fees is clearly downward. The FT projected that AI will make it harder for audit firms to improve their earnings.
※ Subscribe to [Park Si-jin's Global Pick] for the fastest look at global trends.
