[The View]AI Cannot Create Its Own Trust
Moving Beyond Creation: The Age of Verification
Meeting Standards Once Does Not Guarantee Continued Reliability
From Big Tech to Law Firms, All Competing to Be “Watchdogs”
A New Infrastructure of Trust Is Needed
As companies begin to adopt artificial intelligence (AI) in earnest, an unexpected problem has arisen. Companies are finding it increasingly difficult to clearly explain what their AI does and on what basis it makes decisions. Workday, known for its human resources management systems, faced a nationwide class-action lawsuit last year because its AI recruitment system was accused of discriminating against applicants based on age and race.
The European Union (EU)’s AI Act has mandated external audits for high-risk AI systems. Boards of directors and shareholders have also started demanding answers about how the AI deployed by the company is being controlled. Creating AI is no longer the only issue—verifying it has become another important challenge.
An industry aimed at solving this problem is now emerging from several directions in the United States. Specialized startups like Credo AI, Fiddler, and Arthur AI provide tools that can monitor AI system bias, performance, and regulatory compliance in real time.
Amazon, Microsoft (MS), and Google have started embedding governance features to automatically monitor the AI distributed on their cloud services. Law firms are turning AI regulatory advisory services and contract reviews into new revenue streams. Consulting firms are expanding business divisions related to AI verification. Different industries are attempting to tackle the challenge of AI validation in their own ways.
What makes this trend intriguing to me is the very way in which the industry is being formed. Until now, industries that build trust have usually taken a singular form: a qualified expert verifies something at a set time according to set criteria and certifies it. This model applies to automobile safety inspections, food hygiene certifications, and financial audits. However, this approach doesn’t work well with AI, because AI continues to evolve and change.
For example, auto inspections check the state of a vehicle at a specific point in time. If the standards are met at that moment, certification is granted. The same goes for financial audits—they certify the financial state at a fixed time. However, AI systems are constantly being updated with new data, models are retrained, and outputs change according to user behavior. A validation that passes today cannot be guaranteed valid tomorrow. Therefore, AI verification should not be a one-off event at a single point in time but should be a continuous process that persists throughout the life of the system. Instead of being a stamp once and for all, it is more like an ongoing monitoring system.
This is also why so many different industries are simultaneously entering this market. No single industry is positioned to address every aspect of the problem on its own. Each sector holds a different piece of the puzzle: consulting firms bring regulatory expertise, startups have real-time technology monitoring capabilities, major cloud providers offer system accessibility, and law firms have experience handling legal accountability. How these pieces come together will determine the shape of this industry in the future.
And this raises a more fundamental question: does AI verification truly restore trust? Any act of validation ultimately relies on the trustworthiness of the validator. However, if the tools monitoring AI are themselves made by other AI, on what basis do we trust the monitor? There is no answer to this question yet, and there may not be one for some time. Nevertheless, serious attempts to answer this question will shape the future form of the industry.
Ultimately, what we need to build now is the trust infrastructure for the AI era. Depending on how this infrastructure is finalized, it will determine which types of AI survive in the market and which companies are qualified to deploy AI. Until now, we have mainly focused on the industry that creates AI, but going forward, the more important story will come from the industry that validates AI.
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Professor Yoonseok Son, University of Notre Dame, USA
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.