Home

AI-generated content is becoming harder to distinguish from human-created work, increasing the need for reliable ways to verify where content comes from and whether AI was involved. Watermarking and provenance systems can provide useful evidence, but they are not foolproof, and relying too heavily on AI detectors can lead to false accusations, bias, and misplaced suspicion.

https://news.engineering.asu.edu/2026/09/the-rise-of-ai-suspicion/

Public trust in artificial intelligence has plummeted sharply even as everyday reliance on the technology reaches new highs, with growing public anxiety shifting from near-term disruptions like job displacement to long-term, once-fictional doomsday scenarios that once felt confined to science fiction. Leading experts across top academic institutions have weighed in on these rising fears, pushing back against unsubstantiated extreme doomsday claims while acknowledging that unregulated, breakneck AI development carries real, underaddressed risks to public safety and human control. Rather than pushing for a full, blanket pause on all AI progress, most informed voices are calling for balanced, expert-led guardrails that prioritize safety testing and responsible governance, rejecting both unthinking tech utopianism and ungrounded alarmism to steer the technology toward more sustainable, public-focused advancement.

https://poll.qu.edu/poll-release?releaseid=3969

froGQL is a lightweight, open-source graph database that implements the ISO #GQL standard while keeping the graph, schema, and indexes in a single file. It speeds up complex graph queries using worst-case optimal joins and can use static type checking to reject impossible queries before reading the graph, avoiding potentially expensive computation

https://sferrada.com/talk/frogql-worst-case-optimal-joins-and-type-driven-optimization-for-lightweight-gql/

Generative AI can support learning and critical thinking, but its benefits depend on students using it actively and reflectively rather than simply accepting its outputs. Effective prompt writing is linked to stronger critical-thinking awareness, yet students often overestimate their prompting abilities and may not recognize inaccurate or low-quality AI responses. This suggests that students need explicit instruction in strategic prompting, verification, reflection, and evaluation rather than basic training in how to use AI tools. One assessment approach made AI use visible by requiring students to document their prompts, explain changes to AI-generated content, and show how they verified information and reflected on what they learned. This approach helped make responsible AI use and fact-checking assessable, although unclear guidance sometimes led students to use AI in very different ways and raised questions about how well the essay itself measured their writing and argument-development skills. Overall, assessment should make students’ AI-supported learning processes visible while ensuring that critical thinking, writing, independent judgment, and accountability remain central.

https://pure.st-andrews.ac.uk/ws/portalfiles/portal/337643373/Alanquodi-et-al-2026-Critical-thinking-and-prompt-engineering-EDUCON2026-AAM-CCBY.pdf

Artificial superintelligence, defined as AI that outperforms top human experts across nearly all intellectual domains, may arrive sooner than many people expect – potentially within a year, according to a UCSD researcher who previously estimated it was five years away, pointing to recent striking examples of superhuman AI performance in specialized complex fields. We have already reached the milestone of artificial general intelligence, with the most capable frontier AI systems possessing the same flexible, broad-ranging general intellectual abilities that individual humans do. This rapidly evolving landscape calls for expanded public discussion, independent rigorous evaluations of advanced AI capabilities and safeguards, and coordinated cross-disciplinary efforts especially from universities to help steer these powerful systems in ways that benefit humanity.

https://today.ucsd.edu/story/how-close-are-we-to-artificial-superintelligence