Signal feed
AI stories, scored and filtered.
Live items from our monitored sources, filtered for signal and annotated with a recommended posture for enterprise leaders.
2,895 stories
- 14 FebEXPLORE
Better language models and their implications
OpenAI News
OpenAI announced a new unsupervised language model achieving state-of-the-art performance across multiple NLP tasks without task-specific training.
Why it matters
The model's zero-shot capabilities across diverse tasks signal a shift towards more generalized models that reduce fine-tuning requirements for banking applications.
Hype6/10 - 20 DecEXPLORE
OMSCS CS6601 (Artificial Intelligence) Review and Tips
Eugene Yan
The OMSCS CS6601 Artificial Intelligence course review emphasizes starting with simple solutions before adding intelligence.
Why it matters
Eugene Yan's advice to prioritize simple solutions before implementing complex AI aligns with prudent G-SIB risk management and iterative development strategies.
Hype4/10 - 19 DecEXPLORE
OpenAI Fellows Summer 2018: Final projects
OpenAI News
OpenAI concluded its first 6-month Fellows program in Summer 2018, transforming ML beginners into contributors for final projects.
Why it matters
This 2018 program highlights an early and successful model for rapid AI talent development within a frontier lab environment, offering a template for internal upskilling initiatives.
Hype4/10 - 25 JulEXPLORE
OpenAI Scholars 2018: Meet our Scholars
OpenAI News
OpenAI announced its first class of Scholars, a program to transition experienced software developers into machine learning practitioners.
Why it matters
This program highlights an early recognition of the need for structured talent upskilling to meet demand for ML practitioners, a challenge still prevalent for G-SIBs.
Hype4/10 - 20 FebEXPLORE
Preparing for malicious uses of AI
OpenAI News
OpenAI co-authored a paper forecasting AI misuse by malicious actors and potential mitigation strategies with academic and policy partners.
Why it matters
Anticipating malicious AI use cases is critical for G-SIBs to proactively build robust threat models and inform internal red-teaming strategies for AI systems.
Hype5/10 - 27 DecEXPLORE
OMSCS CS7641 (Machine Learning) Review and Tips
Eugene Yan
Eugene Yan reviewed OMSCS CS7641 (Machine Learning), emphasizing fundamental techniques and new developments in ML.
Why it matters
Renewed focus on machine learning fundamentals indicates a recognition of gaps in AI talent's core understanding, directly impacting your G-SIB's internal training and hiring strategies.
Hype3/10 - 6 DecEXPLORE
Block-sparse GPU kernels
OpenAI News
OpenAI released GPU kernels for block-sparse neural networks, claiming orders of magnitude speedup over standard libraries for certain architectures.
Why it matters
This development indicates a potential path to significantly reduce inference costs and latency for large, sparse models, impacting long-term infrastructure planning and model selection.
Hype4/10 - 3 AugEXPLORE
Gathering human feedback
OpenAI News
OpenAI open-sourced RL-Teacher, an interface for training AIs using occasional human feedback rather than predefined reward functions.
Why it matters
This open-source release provides a practical pathway for custom model alignment using internal subject matter experts, which reduces reliance on generic vendor-provided alignment datasets.
Hype4/10 - 17 JulEXPLORE
Robust adversarial inputs
OpenAI News
OpenAI demonstrated adversarial images that consistently fool neural network classifiers across multiple scales and perspectives, challenging prior claims.
Why it matters
This demonstration directly challenges the assumption that multi-view sensor data provides sufficient resilience against adversarial attacks for computer vision models in critical systems.
Hype4/10 - 24 FebEXPLORE
Attacking machine learning with adversarial examples
OpenAI News
OpenAI published a blog post explaining adversarial examples, how they work, and the challenges in securing systems against them.
Why it matters
Adversarial attacks pose a direct threat to the integrity and reliability of ML models deployed in critical banking operations, requiring robust detection and mitigation strategies.
Hype4/10 - 13 FebEXPLORE
Product Categorization API Part 3: Creating an API
Eugene Yan
Eugene Yan details best practices for deploying machine learning models into production via API, focusing on MLOps and infrastructure.
Why it matters
This resource provides practical MLOps and deployment guidance relevant for industrializing AI across diverse use cases within a G-SIB.
Hype2/10 - 8 FebEXPLORE
Adversarial attacks on neural network policies
OpenAI News
OpenAI research details adversarial attacks on neural network policies, demonstrating vulnerabilities in AI agent decision-making.
Why it matters
Adversarial attacks on AI agents highlight an emerging vector for financial fraud and system manipulation, requiring G-SIBs to integrate new security paradigms into agentic system design.
Hype4/10 - 15 NovEXPLORE
OpenAI and Microsoft
OpenAI News
OpenAI announced it will run most of its large-scale model training experiments on Microsoft Azure.
Why it matters
OpenAI's deepened reliance on Azure solidifies Microsoft's strategic advantage in providing the foundational compute for frontier model development, impacting your long-term cloud and vendor strategies.
Hype4/10 - 18 OctEXPLORE
Semi-supervised knowledge transfer for deep learning from private training data
OpenAI News
OpenAI research on semi-supervised knowledge transfer aims to improve model performance by leveraging private data without direct exposure.
Why it matters
This approach offers a potential pathway for G-SIBs to enhance proprietary models using sensitive internal data without direct data sharing or leakage.
Hype4/10 - 29 AugEXPLORE
Infrastructure for deep learning
OpenAI News
OpenAI highlights the importance of deep learning infrastructure and claims open-source ecosystems enable anyone to build robust systems.
Why it matters
OpenAI's claim regarding open-source infrastructure directly challenges traditional build-vs-buy assumptions for G-SIBs considering bespoke AI development environments.
Hype6/10