#Canadian #financialliteracy & #careers information from trusted provincial & federal partners.
@chatterhigh.bsky.social gamifies online information gathering and leaves students knowing where to go for more: resources.chatterhigh.com/competitions
Free & easy to access.
#cdned #onted #bced #abed
"whoever wins the race won’t just have the next great engine of world-saving innovation. They’ll also have the greatest code-breaking machine in history."
www.wired.com/story/q-day-...
"If sustained #AI use erodes the motivation and persistence that drive long-term learning"
#onted #cdned #education
arxiv.org/pdf/2604.04721
"AI Assistance Reduces Persistence and Hurts Independent Performance"
arxiv.org/abs/2604.04721
Just what we're aiming for in #education: reduced persistence and dependence on AI to perform.
"What becomes of intuition and effort when a generative, artificial partner stands ready to answer?"
papers.ssrn.com/sol3/papers....
We don't know, but we'll let our kids find out?
Considering the emerging research on human cognition & #AI, I find it amazing that #education organizations want integrate it into teaching & learning without any hesitation - when we don't even fully understand the implications. #onted #cdned
Quantum computers are coming sooner than expected. Now's a good time to check if your encrypted messenger has moved to post-quantum cryptography. www.eff.org/deeplinks/2...
#AI damage to human intelligence part 6:
"the opacity, fluency, and low-friction interaction patterns of LLMs obscure the boundary between human and machine contribution, leading users to infer competence from outputs rather than from the processes that generate them."
arxiv.org/pdf/2604.14807
Quantum Sensing is an advanced sensor technology that vastly improves the accuracy of how we measure, navigate, study, explore, see, and interact with the world around us
www.baesystems.com/en-us/defini...
I do love a good web based UI:
"A useful, error-corrected #quantumcomputer is within reach for humanity. There's still a lot of work to do, but we think we might be able to accomplish our goal within the decade. Learn more about what it will take."
quantumai.google/learn/map
If you wouldn’t let an IDE vendor stream your keystrokes and project context to their server, why let this one?
"advocated for “right to repair.” The Ocean situation “is a canary in the coal mine for an economy populated by software-driven, connected things. How do we keep this from happening all over the place? How do we keep this from becoming an epidemic?”
www.wired.com/story/the-ri...
"We can see more evidence of cost-cutting in OpenAI's decision to toggle reasoning on and off by default automatically, depending on the complexity of the prompt."
www.theregister.com/2025/08/13/g...
...or perhaps because they're running out of compute.
"As AI agents become autonomous consumers of web content, the threat of environmental manipulation through #AI Agent Traps emerges as a critical security challenge."
"AI Agent Traps", i.e. adversarial content designed to manipulate, deceive, or exploit visiting agents
papers.ssrn.com/sol3/papers....
The 3 lead researchers/writers:
Grace Liu: PhD student in the Machine Learning Department at Carnegie Mellon
Brian Christian: DPhil (PhD) candidate at Oxford, acclaimed author & researcher
Tsvetomira Dumbalska: Postdoctoral research fellow & stipendiary lecturer in Experimental Psych at Oxford
#AI systems become dangerous if they “develop the ability to evade oversight, execute long‑term plans, and resist attempts to shut them down.”
internationalaisafetyreport.org/publication/...
Welcome to the future.
With autonomous agents now outnumbering human operators in critical sectors by 82:1, simple manual testing is no longer sufficient. Modern risks include “Retrieval Sycophancy” and infinite API loops that can drain resources
www.mexc.com/news/663289
A critical flaw is when the agent’s Reasoning Steps Don’t Match the Final Output. The structured reasoning trace is inconsistent with the final recommendation, signaling that the agent’s mental model is drifting.
dac.digital/ai-hallucina...
"People often optimize for long-term goals in collaboration - in contrast, current AI systems are fundamentally short-sighted collaborators - optimized for providing instant and complete responses, without ever saying no" #onted #cdned
arxiv.org/abs/2604.04721
Newer research coming out now:
Within‑LLM variance accounts for roughly 10–34% of total output variability: the same model and prompt can yield outputs spanning the full range from bottom‑quartile to top‑quartile scores
arxiv.org/html/2601.21...
The performance of AI in many cases is at the mercy of the data - which is seldom consistent: ai.jmir.org/2026/1/e83640
Perhaps research on the variability of the AI systems themselves (a more recent enshittification?) isn't out yet?
"the team tested various AI models on this data, they found that their performance was often overestimated by about 20% on average. “We propose that this overestimation is due to data acquisition bias"
www.uhnresearch.ca/news/9-09-20...
The inconsistency problem has many dimensions it seems.
My best guess:
Temporal quality degradation in AI models
pmc.ncbi.nlm.nih.gov/articles/PMC...
"AI models deployed in production often exhibit non‑trivial, heterogeneous degradation over time."
adam.holter.com/model-degrad...
"Claims that AI models degrade over time are not just anecdotal. There’s a real pattern, especially with Anthropic’s Claude Opus 4.6, suggesting a significant dip in performance. A study looking at 6,800 user sessions found a 67% reduction in thinking depth"
"low expectations result in a celebration of far too many mediocre projects. The quality of such projects is often marred by apathy or being rushed. Quality work takes time." #onted
reggio.constructingmodernknowledge.com/computationp...
It sure did: sites.google.com/ugcloud.ca/c...
Growth in Canadian inventorship without matching growth in per‑capita patent ownership lines up with funding architecture that treats universities as open feeders into global AI ecosystems rather than as anchors for Canadian firms.
A familiar & unimaginative solution when we should be reimagining.
#Canada’s Deteriorating #AI Position: A Comparative IP Perspective
balsilliepapers.ca/bsia-paper/c...
Jim went off at the '25 IP Conference:
"Canada used to be recognized as a vibrant AI hub; it is no longer."
Academic funding doesn't stay here.
'AI for all' needs to be Canadian.
Universities ‘absolutely vital’ to Canada’s AI strategy, minister says
universityaffairs.ca/news/ai-mini...
They've fumbled it the past decade - a surprisingly unimaginative approach from a #cdnpoli I thought more imaginative @mark-carney.bsky.social
Tweet from the 2023 ISED IP Conference ↓
Quite the switch in a year. Could this have anything to do with AI blowback so he's jumping on a new bandwagon?
NVIDIA Launches Ising, the World’s First Open AI Models to Accelerate The Path to Useful Quantum Computers
thequantuminsider.com/2026/04/14/n...
Hard to tell between foresight & greed.