Author: lethuphuong

📌 Medical AI can propagate misinformation up to 47% when the source looks like hospital documentation, compared to only 9% if from social media. After more than 1 million tests on 20 models, the study indicates that overall, AI “believes” and propagates fabricated information in about 32% of cases, while warning that AI is more likely to trust prompts with an “authoritative” tone like “I am a senior doctor…” rather than verifying the content. OpenAI’s GPT performed as the least deceived, while some other models accepted up to 63.6% of misinformation.

📌 Yokohama (Japan) is testing a floating data center model running on 100% renewable energy, expected to operate by late March 2026, aiming for commercialization by 2030, alongside a project located at a power plant. Advantages include proximity to power sources, utilization of surplus wind energy, zero grid pressure, and cost reduction via seawater cooling. Japan’s data center market is projected to reach $32.9 billion by 2028, growing over 80% in 5 years.

📌 The lesser-known downside of AI: it boosts productivity to record levels but trades it for mental fatigue, burnout, and skill degradation. In the software industry, AI reduces code production costs but increases coordination, evaluation, and decision-making costs, burdens that fall almost entirely on humans. Engineers shift from “builders” to “reviewers,” constantly screening AI-generated results on an endless assembly line. Humans face pressure to constantly track updates from OpenAI, Anthropic, and other AI firms, even using weekends to test new tools.

📌 AI is changing work silently but profoundly: less elimination, more restructuring. 25% of jobs are heavily impacted and under 1% of skills can be fully automated, while 46% of skills are likely to change task-by-task rather than disappear, shifting focus from “doing it yourself” to “directing AI to do it.” Competitive advantage will belong to organizations that know how to redesign roles, invest in people, and combine AI as an assistant. The real race is not humans vs. AI, but the speed at which an aging workforce adapts to the productivity brought by AI.

📌 The 2026 study exposes a significant gap in the public sector AI race: Europe, especially France, is falling behind despite heavy investment and much talk about strategy. Conversely, Singapore, Saudi Arabia, and India show that AI only delivers value when widely empowered, with clear guidance and substantive training. The data indicates that the future of public service lies not in ambitious AI plans, but in the ability to turn AI into a daily work tool for civil servants.

📌 Research shows that generative AI causes workers to work faster, take on a wider scope of tasks, and extend working hours into the evening or early morning, even though they were not asked to. The boundary between work and rest is blurred because AI makes starting work too easy, leading employees to squeeze in “a little more work.” AI also increases multitasking: running multiple agents in parallel, doing manual work while waiting for AI to process, leading to constant attention switching and high cognitive load. In the long term, this easily leads to workload creep, mental fatigue, burnout, reduced…

📌 The US Federal Trade Commission (FTC) is shifting to a “selective and pragmatic” approach to AI regulation: reducing intervention in the technology itself to avoid stifling innovation, but tightening control over false advertising claims. This is in implementation of President Trump’s Executive Order in the July 2025 AI Action Plan requiring the FTC to review and suspend investigations or penalties that “impede AI innovation.” The cancellation of the investigation order against Rytr shows that the FTC is sharply reducing enforcement related to the capabilities of AI products themselves, even when there is a risk of misuse. Previously, the FTC…

📌 Data from ASML shows that the AI infrastructure boom is still accelerating, with €13 billion in new orders in just one quarter and €32.7 billion in revenue. Despite long-term risks and uncertainties, chip manufacturers are acting as if AI demand will last for many years. This indicates that the construction of data centers and large-scale AI supply chains is unlikely to stall in the near future.

📌 A Harvard article emphasizes that bias in AI is not just a technical issue but a human behavioral issue. As AI becomes deeply involved in critical decisions, how we ask, evaluate, and act on AI results can silently amplify deviations. By slowing down, engaging in intentional reflection, and designing systems that encourage critical thinking, AI can become a better decision-making partner rather than a “megaphone” for our own prejudices.

📌 Singapore is deploying a $10 million support package comprising government grants and upfront bank funding to promote AI adoption in micro and small businesses (under 30 employees). The package marks a significant step for Singapore in bringing AI to the business group most vulnerable to being left behind. By combining subsidies, bank credit, and collaboration with schools, the government and industry associations are reducing technology investment risks for micro-enterprises. This approach not only boosts AI adoption but also strengthens an inclusive digital economic foundation where innovation is not reserved solely for large corporations.