배현영일반회원

16 articles published

배현영 is a journalist at META-X.

Recent Articles

Research Review · 2026-09-04

Hybrid Work Boosts Retention, Sustains Performance | Nature

A six-month randomized trial of 1,612 Trip.com employees found that two optional work-from-home days cut attrition by 33% and raised job satisfaction. Two-year performance reviews, promotions, and coding output showed no meaningful decline, reframing hybrid work as a retention strategy.

배현영
Research Review · 2026-09-01

AI in Education: Preferred Scaffolding & Learning Outcomes

This study compares five AI scaffolding roles in mathematical modelling and asks whether preferred support also improves performance. The findings suggest that it does not always do so, shifting attention from satisfaction to the timing, form, and gradual fading of AI support.

배현영
Research Review · 2026-08-28

Reliability of Published Research: The Replication Challenge

Researchers retested 274 positive claims from 164 social and behavioral science papers. Only 55.1% produced a significant result in the original direction. Rather than proving the rest false, the study shows why science needs replication, transparency, and cautious interpretation.

배현영
Research Review · 2026-08-25

AI Cooperation: LLMs in Repeated Games Review

Large language models are evolving from conversational tools into agents that repeatedly interact with humans and other AI systems. This study placed five LLMs, including GPT-4, in repeated strategic games to examine cooperation, defection, and coordination. The researchers further showed that Socia

배현영
Research Review · 2026-08-24

Optimal Sleep & Biological Aging: What Clocks Reveal

A 2026 Nature study examined sleep duration using 23 biological ageing clocks derived from MRI, proteomic and metabolomic data. Several measures showed a U-shaped pattern, with the lowest biological ageing around 6.4–7.8 hours of sleep. Both short and long sleep were linked to poorer health outcomes

배현영
Research Review · 2026-08-13

Generative AI Harms Math Learning, PNAS Study Finds

Generative AI improved performance during practice, but students using unrestricted GPT-4 scored 17% lower than the control group on an unaided test. A guided tutor using teacher-designed hints reduced this harm, although it did not significantly outperform the control group.

배현영
Research Review · 2026-08-11

Physical AI: Intelligent Systems for Real-World Interaction

This paper reviews the conceptual genealogy and technological evolution of Physical AI and classifies the field into three interconnected domains: Physics-Informed AI, Generative Physical AI, and Embodied AI. Physical AI is presented as an integrative paradigm in which intelligent systems understand

배현영
Research Review · 2026-08-03

Centaur AI Predicts Human Behavior in Psychological Studies

Researchers reporting in Nature developed Centaur, an AI model trained on more than 10.68 million choices made by over 60,000 participants across 160 psychological experiments. Centaur predicted human choices on new participants and previously unseen tasks more accurately than conventional cognitive

배현영
Research Review · 2026-07-31

AI Leaves the Screen: Physical AI and the Next Robotics Revo

Physical AI moves intelligence into robots that sense, decide, and act in the real world. The paper links perception, mapping, planning, control, and human–robot interaction, while highlighting digital twins, sim-to-real transfer, safety, latency, energy, security, scalable learning, and ethics.

배현영
Research Review · 2026-07-27

AI in Education's Multimodal Gap: Beyond Text-to-Text

Generative AI can create text, images, speech, music, and video, yet education research still centers on text-based LLMs. A topic-modeling review of 4,175 studies identified 38 topics and 14 themes, revealing broad adoption but limited exploration of multimodal learning.

배현영
Research Review · 2026-07-23

How Multimodal Is Generative AI Research in Education?

Generative AI can create text, images, speech, music, and video, yet education research still centers on text-based LLMs. A topic-modeling review of 4,175 studies identified 38 topics and 14 themes, revealing broad adoption but limited exploration of multimodal learning.

배현영
Insight · Analysis · 2026-01-19

[META-X] January Week 4 2026 Newsletter | META-X

META-X Newsletter Issue 56, January Week 4 2026. Global news: 'A Choice for Preservation' — CD Projekt restructures while preserving philosophy, selling GOG.

배현영
Insight · Analysis · 2026-01-12

[META-X] Newsletter: January

[Business & Insights] Samsung presents AI as daily life, not technology — The First Look 2026 at CES 2026 Las Vegas was designed not as a product showcase but as a space for attendees to directly experience AI-integrated daily life.

배현영
Insight · Analysis · 2026-01-05

META-X Newsletter: January Week 2 2026 | META-X

Larian Studios raises the accountability question. In the gaming industry, the generative AI debate is shifting from 'did you use AI or not?' to 'where and at what stage did you use it?'

배현영
배현영 — Journalist | META-X