Seattle Skeptics on AI

Tamara Weed, Mar, 23 2026

Learn how memory planning techniques like CAMELoT and Dynamic Memory Sparsification reduce OOM errors in LLM inference by 40-60% without sacrificing accuracy - and why quantization alone isn't enough for long-context tasks.

Tamara Weed, Mar, 23 2026

Memory planning techniques like CAMELoT and Dynamic Memory Sparsification let LLMs handle long contexts without OOM crashes-cutting memory use by 50% while improving accuracy. No more brute-force GPU scaling needed.

Tamara Weed, Mar, 22 2026

Moving from an LLM pilot to production requires more than technology-it demands strategy, governance, and phased rollout. Learn how top enterprises avoid costly mistakes and scale AI effectively.

Tamara Weed, Mar, 21 2026

Scientific Large Language Models are transforming research by accelerating literature review, automating experimental design, and connecting cross-disciplinary insights-but they come with serious risks. Learn how they work, where they succeed, and why human oversight is still essential.

Tamara Weed, Mar, 20 2026

Secure generative AI development requires rethinking secrets, logging, and testing. Learn how prompt injection, AI-BOMs, red-teaming, and short-lived credentials protect your models from emerging threats in 2026.

Tamara Weed, Mar, 18 2026

Databricks AI red team uncovered critical vulnerabilities in AI-generated game and parser code, revealing how prompt injection and data leakage can bypass traditional security tools. Learn how to protect your systems.

Tamara Weed, Mar, 17 2026

Ensembling generative AI models by cross-checking outputs reduces hallucinations by 15-35%, making AI safer for healthcare, finance, and legal use. Learn how majority voting, cross-validation, and model diversity cut errors-and when it’s worth the cost.

Tamara Weed, Mar, 16 2026

Sparse attention and Performer variants solve the quadratic memory problem in transformers, enabling LLMs to process sequences up to 100,000+ tokens. Learn how these efficient architectures work, where they outperform standard models, and how they're being used in healthcare, legal tech, and genomics.

Tamara Weed, Mar, 15 2026

AI is transforming database schema design by generating accurate, optimized structures from plain language. Learn how AI validates models, creates safe migrations, and prevents common errors-so you can build scalable systems faster.

Tamara Weed, Mar, 14 2026

Enterprise LLM deployments need fairness evaluation frameworks to catch hidden bias before it harms users or violates regulations. Tools like FairEval and LangFair help organizations test for demographic and personality-based bias in real-world scenarios.

Tamara Weed, Mar, 12 2026

Large language models show measurable gender and racial bias in hiring and decision-making, favoring white women while penalizing Black men. Real-world testing reveals persistent, intersectional bias that current debiasing methods fail to fix.

Tamara Weed, Mar, 11 2026

Agent-oriented large language models go beyond answering questions-they plan, use tools, and act autonomously. Learn how they work, where they excel, and the risks you can't ignore.