Academics who aren't bought (at least yet) "The GSM8K benchmark is widely used to assess the mathematical reasoning of models…it remains unclear whether their mathematical reasoning capabilities have genuinely advanced, raising questions about the reliability of the
RESEARCH
-
OpenAI o1 Model Enhances LLM Reasoning Through Inference Compute
By
–
HuggingFace: "OpenAI’s o1 model showed that when LLMs are trained to do the same—by using more compute during inference—they get significantly better at solving reasoning tasks like mathematics, coding, and logic."
-
AI Safety as Power Centralization: Geopolitical Control Mechanisms
By
–
We've discussed how so-called "AI Safety" is a smoke screen to centralize power & guzzle resources without accountability. See our paper where we also discuss how they use the whole "US has to protect ourselves from China." Predictable, right on cue. https://
firstmonday.org/ojs/index.php/
fm/article/view/13636/11599
… -
US Government Collaboration Secures Advanced AI Model Capabilities
By
–
…including a careful process for which frontier capabilities to include in released models…it is…important that we are working closely with the US government to best protect the most capable models from efforts by adversaries and competitors to take US technology.”
-

TinyZero: Affordable RL Finetuning Under $30
By
–
TinyZero reproduction of R1-Zero
"experience the Ahah moment yourself for < $30" Given a base model, the RL finetuning can be relatively very cheap and quite accessible. -
Building Diverse RL Environments for LLM Cognitive Strategy Development
By
–
For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.
-

UK Government Releases First International AI Safety Report
By
–
The UK gov't just released the first International AI Safety Report, offering key insights on general-purpose AI to guide informed policymaking. HAI leading scholars @YejinChoinka
, @percyliang
, @MarietjeSchaake
, and @RishiBommasani contributed to the report, which will shape -
Jevons’ Paradox in AI Environmental Debate and Rebound Effects
By
–
🌎Full paper here: "From Efficiency Gains to Rebound Effects: The Problem of Jevons’ Paradox in AI’s Polarized Environmental Debate" arxiv.org/pdf/2501.16548
→ View original post on X — @katecrawford, 2025-01-29 18:22 UTC
-
AI Progress Cannot Reduce Climate Impact Without Curbing Consumption
By
–
Conclusion: AI advances can't reduce climate impacts if they fuel exponential consumption
📌 Energy efficiency alone isn't enough
📌 Integrate total lifecycle assessments + socioeconomic analysis for climate interventions
📌 Address AI-driven economic & behavioral shifts [Translated from EN to English]→ View original post on X — @katecrawford, 2025-01-29 18:21 UTC
-
AI Environmental Impact Beyond Hardware Efficiency and Emissions
By
–
AI’s impact is shaped by market incentives, regulatory gaps, and social behaviors, not just hardware efficiency. Focusing only on direct emissions ignores system-wide feedback loops that drive AI’s expanding environmental footprint 📈
→ View original post on X — @katecrawford, 2025-01-29 18:21 UTC