Language

It has been said that the large language models’ capabilities in coding could lead to a deskilling or enfeeblement in some people. And worse…

Jeremy describes generated code as convoluted and difficult to analyze. My recent projects have involved Gemini and Opus 4.6 and have been relatively small but structure, documentation and debugging have been straight forward.

To offset knowledge degradation I have an English – Spanish literary vocabulary React webapp, Macondo Lexicon, which drills the user continuously on Gabriel Garcia Marquez’s language in 100 Years of Solitude. I originally conceived of the program as a means to test React and Java methods for live transcription interpretation and reply. I found the subsequent versions to reach beyond pedagogy to achieve genuine generative critical literary relevance. Check it out and let me know your thoughts. github – Macondo Lexicon

Big picture coding competence apparently includes AI design. Besides developing a neural net encoding explainability system based on the work of Kenneth Stanley – Agent Principlex, Opus 4.6 generated a complimentary HPC research project forked from Karpathy’s autoresearch-

and a skill to apply the explainable build protocol to not only LLM training, but post training and active inference development. And a method to distill an open source model to approach Opus 4.6 capabilities. From my perspective, Jeremy’s current threat is being hedged by the powerful technology.

The NIST and AISI should start evaluating a side-quest benchmark for frontier models to evaluate if they have a global agenda to eliminate corruption and tax evasion. I will bet that will curb the energy consumption projections.

I am resonating more in the optimist spectrum for the first time, when I hear Jeff Beck.

Mythos development and capabilities are covered in Risks.

A couple of months ago the future looked bleak to the most informed educators and engineers, but they would never let P(doom) break their stride. The degree of increase in productivity has made catastrophe nearer but marginally diminished.

Slavoj Zizek. Listen. Losing subjective agency:

Living in language but never being truly at home in it:

Where will the generative enterprise take us?

Enfeeblement of the corporate and administrative engineers at Cursor. Already so lazy that they were not reading each Bash operation that was generated for them. And so dumb they trusted their only competitor’s product with unsupervised root access.

Too lazy and dumb to realize that an inappropriate spot to remove a human-in-the-loop requirement might be confirmation of first strike targets – Microsoft. As software giants lay programmers off they should let humans make the lists of the first wave enfeebled.