201 articles
Science Advances research found AI gives writers 9% more useful stories but reduces collective narrative diversity when everyone uses the same AI. Here is what generative AI has learned about
A randomised trial found AI tools made experienced developers 19% slower. Harvard Business Review's 2026 research confirmed 88% of heavy AI users report burnout. Here is what the data actually
69% of developers kept using AI tools after a trial proved they were 19% slower with them. Here is the psychology behind why we keep returning to AI despite knowing
542,000 industrial robots installed in 2024 — second highest in history. The global market value hit an all-time high of $16.7bn. Here is how the robotics industry actually works, beyond
The ambient invisible AI market is growing at 24.5% annually. The technology is already in smart homes, healthcare, and manufacturing. Here is what it actually is, what privacy it costs
AI's carbon footprint could equal New York City's in 2025. Data centre energy use doubles to 945 TWh by 2030. But Google cut Gemini's energy per prompt by 33x in
One AI video uses as much energy as 200,000 spam classifications. Google cut Gemini energy use 33x but total AI energy is rising. The Jevons Paradox explains why smarter AI
EU AI Act full enforcement arrived August 2026, fines up to €35 million or 7% of revenue. 64% of workers use shadow AI bypassing corporate controls. Here is what AI
A 2025 study found that greater AI knowledge correlates with LESS trust in police facial recognition. 8 Americans wrongfully arrested from misidentification. Here is the honest evidence on AI in
China's $37bn Wuhu data island is the flagship of its plan to close a 60-point gap in global AI compute against the US. Stuart Kerr maps the strategy, the obstacles,
Albania appointed AI system Diella as world's first cabinet-level AI minister in September 2025. Stuart Kerr examines what the experiment means for democracy, accountability, and AI in government.
AI-generated phishing is now fluent, personalised, and capable of agentic attacks. Stuart Kerr examines the prompt injection threat, vibe hacking, and what organisations and individuals can do now.
MIT's VaxSeer uses AI to predict which flu strains will dominate each season, outperforming WHO selections in retrospective tests. Stuart Kerr examines what it means for the future of vaccine
Switzerland's EPFL and ETH Zurich launched Apertus in September 2025, a fully open LLM with published weights, architecture, and training data. Stuart Kerr examines what it means for Europe's AI
Small language models running entirely on smartphones are no longer research experiments. Stuart Kerr examines what Phi-3 Mini, Apple Intelligence, and the edge AI shift mean for privacy, latency, and
AI apocalypse narratives are dominating public discourse, but researchers warn they distort priorities and crowd out present harms. Stuart Kerr examines the surge in doom forecasting and its consequences.
Politicians are dismissing authentic evidence as AI-generated fabrication, exploiting the 'liar's dividend'. Stuart Kerr examines the research, the mechanism, and what it means for democratic accountability.
DCN data shows median 10% YoY Google traffic decline for publishers in 2025, with non-news brands down 14%. Stuart Kerr examines how AI Overviews are restructuring the economics of online
Evolutionary biologist David Krakauer calls today's AI 'fake intelligent': fluent without comprehension. Stuart Kerr examines the argument, what genuine intelligence would require, and why the distinction matters.
A UK Government report projects AI data centre water demand could reach 4.2-6.6bn cubic metres annually by 2027. Stuart Kerr examines the hidden water costs of AI training that carbon
Central banks including the ECB and Czech National Bank are deploying AI for inflation nowcasting. Stuart Kerr examines what the results show, the limitations that matter, and what it means
Nine in ten UK councils are already using or exploring AI. Stuart Kerr examines what the experiments actually look like, who is benefiting, and what accountability questions remain unanswered.
AI is approving and denying health insurance claims in seconds, and pre-emptive profiling is moving upstream into underwriting. Stuart Kerr examines the algorithmic insurance gamble and what it means for
AI models degrade over time as their training data ages, and recursive training on synthetic content accelerates the problem. Stuart Kerr examines model drift, the curse of recursion, and what
From precision farming to personalised nutrition, AI is quietly reshaping every link in the food chain. Stuart Kerr examines what algorithmic food systems mean for access, choice, and the environment.
AI tools are already reconstructing crime scenes and analysing digital evidence in legal investigations. Legal scholars and judicial panels are now debating what happens when those reconstructions become so detailed
Nations are increasingly deploying state-backed AI models as tools of geopolitical influence, embedding their values and strategic interests into the systems running government ministries, universities, and healthcare networks across the
A new Google Cloud and Harris Poll survey finds 87 percent of game developers are already using AI agents in their workflows, with more than a third deploying it for
AI is transforming what and how we teach, but the risks of bias, inequality, and narrow curriculum design demand human oversight at every stage of educational reform.
AI systems built on majority-language data risk erasing linguistic diversity worldwide. Minority languages, dialects, and oral traditions face digital marginalisation unless urgent action is taken.