
When The Filter Is The Attack Surface
Frederick Lowe - March 14, 2026
I recently completed a content moderation system for an enterprise social media platform. It would be understatement to say some of its users have incompatible worldviews.
Frederick Lowe - March 14, 2026
I recently completed a content moderation system for an enterprise social media platform. It would be understatement to say some of its users have incompatible worldviews.
Frederick Lowe - December 16, 2025
It's an understatement to say "AI" (really, Transformer-based LLMs) are actively remaking the world. Yes, other forms of AI exist. Diffusion models and RL still matter, and depending on application domain, can matter more than Transformers.
Frederick Lowe - Dec 8, 2025
Over the last two years, I built an industry-killer: an AI-powered platform for high-quality article generation. The "Eureka!" moment wasn't discovering that Generative AI can be used to do this. It was realizing that the REAL problem is creating editin...
Frederick Lowe - October 13, 2025
In November 2024, the CEO of a global digital advertising company called me about a problem potentially costing his firm millions: lost Requests For Proposal (RFP).
Frederick Lowe - July 21, 2025
Complex prompts often produce good results up to a point, after which additional refinements can reduce output quality. This happens because additional instructions create competing attention patterns in the transformer's processing.
Frederick Lowe - Mar 10, 2025
For the past two years, I've served as the Lead Architect (and frequent bench engineer) on several applied projects integrating various Large Language Models (LLMs).
Frederick Lowe - Jan 8, 2025
Most of what folks call "AI" now revolves around Generative Pre-trained Transformers (GPTs). A key concept in understanding how GPTs work is to understand the concept of "Attention".
Frederick Lowe - Oct 14, 2024
Clarity in verbal and written communication is among the most critical skills we learn as human beings. But using complex vocabulary, while efficient in expert audiences with shared knowledge, is often the wrong approach.






