Recent content on Kenneth's Blog
The feudal lords of AI 1.0 built their moats on GPU scarcity. But every squeeze contains the seeds of its own destruction. As the infrastructure glut arrives and efficiency gains compound, the era of renting intelligence gives way to sovereign ownership. This is not about better models-it's about who owns them, and who has options.
The first in a three-part series dissecting transformer architecture and LLM inference optimization. Part 1 explores why language models are inherently slow, examining the dual-highway information flow that creates both their power and their performance bottleneck. Understanding the architecture reveals why your LLM is always the bottleneck - and what can actually be done about it.
This essay dissects the widening gap between AI hype and reality, arguing that large language models have hit a plateau - the “S-curve” - despite industry claims of imminent superintelligence. It contrasts bold predictions and massive investments with underwhelming flagship releases, framing today’s AI era as less about building godlike intelligence and more about integrating imperfect tools into…
Project Overview This entry marks the beginning of documentation for what I’m calling the Progressive State Transformer (PST) - a hierarchical narrative generation system designed to create coherent, structured stories using LLMs that I’ve been working on (on and off) for the past months. The core innovation is maintaining persistent state across multiple narrative layers, allowing for…
This essay explores how monopoly-seeking behavior creates intellectual hyper-liquidity, fostering innovation while making the unprecedented access to valuable technology an unsustainable yet transformative phenomenon.
In an era where AI solutions often chase complexity, this insight illuminates a fundamental truth: Large Language Models excel primarily at word-level pattern recognition and manipulation. Rather than pursuing broad, unbounded applications that will rapidly become obsolete, organizations should focus on implementing LLMs in constrained, specific tasks that leverage their core text processing…
WebAssembly transcends the traditional dichotomy between development simplicity and deployment flexibility. Its profound insight lies in recognizing that the complexity of distributed systems stems not from their inherent nature, but from prematurely fixed architectural decisions. By elevating deployment topology to a runtime concern while preserving a unified development model, WebAssembly…
Hi there, I’m Kenneth A scientist who transitioned to software development, combining 4 years of biomedical research with 4 years of professional software engineering. My experience spans software consultancy, financial sector development, and freelance work for startups. Currently focusing on integrating my cross-disciplinary background in Data Science, Software Engineering, and Product…
Website Operator (Angaben gemäß § 5 TMG) Kenneth Wolters Responsible for Content (Verantwortlich für den Inhalt nach § 55 Abs. 2 RStV) Kenneth Wolters Contact Information Email: kenneth.wolters@hey.com Disclaimer Liability for Content The contents of this website have been created with the utmost care. However, I cannot guarantee the contents’ accuracy, completeness, or topicality. According…
Data Collection and Usage This blog collects minimal data required for core functionality: Server logs with IP addresses and browser information, retained for 30 days Contact form submissions if you choose to reach out Analytics are implemented through privacy-focused Plausible Analytics, which collects only aggregate page views without personal identifiers. No Cookie Usage This site operates…