Enterprise & Work

95 Percent of Organizations See No Measurable AI Gains, University Leaders Warn

University leaders from Korea, the US and Vietnam told a Seoul summit that 95 percent of organizations have yet to see measurable gains from AI, and urged hands-on training to close the gap.

University leaders call for practical AI training, commercialization
University leaders call for practical AI training, commercializationnaassomz1 / pixabay
By Marcus Bennett5 min read

Why it matters

  • Yap Peng Tan cited reports suggesting 95 percent of organizations have yet to see measurable economic benefits from their AI investments.
  • VinUniversity has launched a program to train 20,000 AI professionals for Vietnam, ending with six weeks inside a company.
  • Taejae University opened in Seoul in 2023, teaches online in English, and uses an AI platform to track faculty speaking time and student skills.
  • The SMaRT algorithm, built with the World Bank for Kenya's judiciary, assigns mediators by skill, license and caseload, with a randomized controlled trial being prepared.
  • Tan noted the steam engine, electricity and the internet took about 60, 40 and 25 years to deliver productivity gains, while AI has been public for about four years.

Ninety-five percent of organizations have yet to see measurable economic benefits from their AI investments, according to reports cited by Yap Peng Tan, provost of VinUniversity in Hanoi, at a summit session in Seoul on Wednesday. The figure anchored a blunt message from higher education leaders from Korea, the United States and Vietnam: universities must push artificial intelligence out of the lab and into the marketplace, and the incentives that govern academic careers are standing in the way.

The three panelists spoke during a session on AI commercialization at the Innovation & Impact Summit 2026, hosted by Korea University in Seoul and organized by Times Higher Education, the London-based publisher behind the World University Rankings. The stakes extend beyond academia. Tan's 95 percent figure, if accurate, describes a broad gap between AI spending and AI returns across the economy, and the panelists argued that universities are one of the places where that gap either narrows or widens.

What did the Korean panelist propose?

Yeom Jae-ho, president of Taejae University and a former president of Korea University, said simply passing knowledge on to students "is not enough in the 21st century." His institution offers one concrete answer to that claim.

Taejae University opened in Seoul in 2023 and teaches its classes online in English. Its students spend semesters abroad in cities including Tokyo and New York, then return to Korea in their fourth year for a final project. The structure is deliberately international and project-based rather than lecture-based.

Yeom said he tells professors, "Don't teach anything at all," and has them lead discussions instead. An AI platform tracks how long faculty members speak and how students demonstrate skills such as critical thinking. The platform gives the university instrumented feedback on classroom practice: speaking time for faculty, observable skill demonstration for students.

What is the American research project?

Sanmay Das, professor and associate director of AI for social impact at Virginia Tech's Sanghani Center for Artificial Intelligence and Data Analytics, described a project conducted with the World Bank and the Kenyan judiciary. The project targets a specific procedural bottleneck in Kenya's legal system.

Kenya requires many civil cases to go through mediation before they reach a judge. Das's team built an algorithm called SMaRT, short for "Selecting Mediators that are Right for the Task," which assigns mediators to cases based on their skill, their licenses and how many cases they already have.

Das said the team is preparing a randomized controlled trial of the algorithm in the courts. He added that the trial would have been impossible without the World Bank's infrastructure. The reason, in his telling, is structural.

"It's because the incentives of academia have not sustained projects like this," Das said. He noted that such projects take years, while papers remain the main currency of academic careers. He compared the gap to the "valley of death," the stretch between discovery and commercialization where many new drugs stall.

The analogy carries weight for research policy. If AI deployments stall the way drug candidates stall, the failure point is not the underlying science but the institutional stretch between proof of concept and sustained real-world operation. Das's argument targets that stretch directly.

Why does the 95 percent figure matter?

Tan, whose VinUniversity is a private institution in Hanoi founded by Vietnamese conglomerate Vingroup, framed the current moment with historical comparisons. The steam engine, electricity and the internet took about 60, 40 and 25 years respectively to deliver meaningful productivity gains, he noted, while AI has been in public hands for about four years.

The comparison cuts two ways. It suggests patience is warranted, and it suggests the productivity gap Tan's 95 percent figure describes may be a timing problem rather than a permanent condition. But Tan did not rest on patience.

The real shortage, he said, is people who combine knowledge of a field with practical AI skills and experience putting AI to work. That claim defined his prescription for universities: produce graduates who can apply AI inside domains such as medicine, industry and public services, rather than graduates who merely understand the technology in the abstract.

How is VinUniversity responding?

Tan said VinUniversity is redesigning its courses around real problems from hospitals, companies and communities. It has also launched a program to train 20,000 AI professionals for Vietnam. The program ends with six weeks spent inside a company solving its problems.

The design is deliberate. Six weeks of embedded company work forces contact between classroom learning and operational reality, the same stretch of territory Das described as the "valley of death" in research. VinUniversity is attempting to cross it at the level of student training rather than technology transfer.

"AI will keep evolving faster than ever and the university must evolve with it," Tan said.

What connects the three arguments?

The panelists converged on a single diagnosis from three different positions: a university president in Seoul, an AI-for-social-impact researcher in Virginia, and a provost in Hanoi. All three argued that the traditional academic model — knowledge transmission, publication, credentialing — does not match what AI deployment now demands.

Yeom's answer is pedagogical: discussion-led teaching with AI-based tracking of student skills. Das's answer is infrastructural and incentive-based: projects like SMaRT need external partners such as the World Bank because academic incentives do not sustain multi-year deployments. Tan's answer is curricular: courses built around problems from hospitals and companies, plus a national-scale training program with mandatory industry immersion.

The common thread is practical AI competence — training people and sustaining projects that operate in the world rather than in the literature. Tan's 95 percent figure gives that agenda its economic framing: the gap between AI investment and measurable benefit is, on his reading, largely a workforce and deployment problem.

The SMaRT trial in Kenyan courts will offer one early test of whether the deployment model the panelists favor can survive the institutional conditions they described. If the randomized controlled trial proceeds, it will measure an algorithm's effect on mediator assignment in a working judiciary — a concrete, years-long engagement of exactly the kind Das said academic incentives fail to sustain.

Source: feed.koreatimes.co.kr

Share this article:

More from Marcus Bennett

Marcus Bennett

Show full bio

Senior reporter covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.

718 articles

Related articles

  1. Over 300 Delegates From 40 Countries Gather in Seoul for THE Innovation & Impact Summit 2026
  2. Elsevier's Chi urges focus and trust as AI pressures universities
  3. Universities Must Become 'Neutral Brokers' as AI Forces Mission Rethink
  4. Google DeepMind Partners with Korea to Bring Frontier AI to Science
  5. Korea Builds Five-Group Task Force to Tackle AI-Driven Inequality

« Previous articleNext article »