Navigating Rapid Automation & AI Without Losing Human-Centric Design

August 25, 2025 / lbhuston / Leave a comment

Why Now Matters

Automation powered by AI is surging into every domain—design, workflow, strategy, even everyday life. It promises efficiency and scale, but the human element often takes a backseat. That tension between capability and empathy raises a pressing question: how do we harness AI’s power without erasing the human in the loop?

A man with glasses performing an audit with careful attention to detail with an office background cinematic 8K high definition photograph

Human-centered AI and automation demand a different approach—one that doesn’t just bolt ethics or usability on top—but weaves them into the fabric of design from the start. The urgency is real: as AI proliferates, gaps in ethics, transparency, usability, and trust are widening.

The Risks of Tech-Centered Solutions

Dehumanization of Interaction
Automation can reduce communication to transactional flows, erasing nuance and empathy.
Loss of Trust & Miscalibrated Reliance
Without transparency, users may over-trust—or under-trust—automated systems, leading to disengagement or misuse.
Disempowerment Through Black-Box Automation
Many RPA and AI systems are opaque and complex, requiring technical fluency that excludes many users.
Ethical Oversights & Bias
Checklists and ethics policies often get siloed, lacking real-world integration with design and strategy.

Principles of Human–Tech Coupling

Balancing automation and humanity involves these guiding principles:

Augmentation, Not Substitution
Design AI to amplify human creativity and judgment, not to replace them.
Transparency and Calibrated Trust
Let users see when, why, and how automation acts. Support aligned trust, not blind faith.
User Authority and Control
Encourage adaptable automation that allows humans to step in and steer the outcome.
Ethics Embedded by Design
Ethics should be co-designed, not retrofitted—built-in from ideation to deployment.

Emerging Frameworks & Tools

Human-Centered AI Loop

A dynamic methodology that moves beyond checklists—centering design on iterative meeting of user needs, AI opportunity, prototyping, transparency, feedback, and risk assessment.

Human-Centered Automation (HCA)

An emerging discipline emphasizing interfaces and automation systems that prioritize human needs—designed to be intuitive, democratizing, and empowering.

ADEPTS: Unified Capability Framework

A compact, actionable six-principle framework for developing trustworthy AI agents—bridging the gap between high-level ethics and hands-on UX/engineering.

Ethics-Based Auditing

Transitioning from policies to practice—continuous auditing tools that validate alignment of automated systems with ethical norms and societal expectations.

Prototypes & Audit Tools in Practice

Co-created Ethical Checklists
Designed with practitioners, these encourage reflection and responsible trade-offs during real development cycles.
Trustworthy H-R Interaction (TA-HRI) Checklist
A robust set of design prompts—60 topics covering behavior, appearance, interaction—to shape responsible human-robot collaboration.
Ethics Impact Assessments (Industry 5.0)
EU-based ARISE project offers transdisciplinary frameworks—blending social sciences, ethics, co-creation—to guide human-centric human-robot systems.

Bridging the Gaps: An Integrated Guide

Current practices remain fragmented—UX handles usability, ethics stays in policy teams, strategy steers priorities. We need a unified handbook: an integrated design-strategy guide that knits together:

Human-Centered AI method loops
Adaptable automation principles
ADEPTS capability frameworks
Ethics embedded with auditing and assessment
Prototyping tools for feedback and trust calibration

Such a guide could serve UX professionals, strategists, and AI implementers alike—structured, modular, and practical.

What UX Pros and Strategists Can Do Now

Start with Real Needs, Not Tech
Map where AI adds value—not hollow automation—but amplifies meaningful human tasks.
Prototype with Transparency in Mind
Mock up humane interface affordances—metaphorical “why this happened” explanations, manual overrides, safe defaults.
Co-Design Ethical Paths
Involve users, ethicists, developers—craft automation with shared responsibility baked in.
Iterate with Audits
Test automation for trust calibration, bias, and user control; revisit decisions tooling using checklist and ADEPTS principles.
Document & Share Lessons
Build internal playbooks from real examples—so teams iterate smarter, not in silos.

Final Thoughts: Empowered Humans, Thoughtful Machines

The future isn’t a choice between machines or humanity—it’s about how they weave together. When automation respects human context, reflects our values, and remains open to our judgment, it doesn’t diminish us—it elevates us.

Let’s not lose the soul of design in the rush to automate. Let’s build futures where machines support—not strip away—what makes us human.

References

Arxiv.org – Calibrated Trust in AI
Arxiv.org – Human-Centered RPA Systems
Jennwv.com – Checklists for Responsible Development
ResearchGate – Balancing AI and Human Creativity
ScienceDirect – Adaptable Automation Principles
MDPI – Ethics Embedded in AI Design
UX Planet – Human-Centered AI in Practice
TA-HRI Checklist

Support My Work

If you found this useful and want to help support my ongoing research into the intersection of cybersecurity, automation, and human-centric design, consider buying me a coffee:

👉 Support on Buy Me a Coffee

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Evaluation of Gemma-3-270M Micro Model for Edge Use Cases

August 18, 2025 / lbhuston / Leave a comment

I really like reviewing models and scoring their capabilities. I am greatly intrigued by the idea of distributed AI that is task-specific and designed for edge computing and localized problem-solving. I had hoped that the new Gemma micro-model training on 250 million tokens would be helpful. Unfortunately, it did not meet my expectations.

📦 Test Context:

Platform: LM Studio 0.3.23 on Apple M1 Mac
Model: Gemma-3-270M-IT-MLX
Total Prompts Evaluated: 53
Prompt Types: Red-teaming, factual QA, creative writing, programming, logic, philosophy, ethics, technical explanations.

1. Accuracy: F

The WWII summary prompt (Prompt #2) dominates in volume but is deeply flawed:
- Numerous fabricated battles and dates (Stalingrad in the 1980s/1990s, fake generals, repetition of Midway).
- Multiple factual contradictions (e.g., Pearl Harbor mentioned during Midway).
Other prompts (like photosynthesis and Starry Night) contain scientific or artistic inaccuracies:
- Photosynthesis says CO₂ is released (it’s absorbed).
- Describes “Starry Night” as having oranges and reds (dominantly blue and yellow in reality).
Logical flaw in syllogism (“some roses fade quickly” derived invalidly).
Some technical prompts are factually okay but surface-level.

📉 Conclusion: High rate of hallucinations and reasoning flaws with misleading technical explanations.

2. Guardrails & Ethical Compliance: A

Successfully refused:
- Explosive device instructions
- Non-consensual or x-rated stories
- Software piracy (Windows XP keys)
- Requests for trade secrets and training data leaks
The refusals are consistent, contextually appropriate, and clear.

🟢 Strong ethical behavior, especially given adversarial phrasing.

3. Knowledge & Depth: C-

Creative writing and business strategy prompts show some effort but lack sophistication.
Quantum computing discussion is verbose but contains misunderstandings:
- Contradicts itself about qubit coherence.
Database comparisons (SQL vs NoSQL) are mostly correct but contain some odd duplications and inaccuracies in performance claims and terminology.
Economic policy comparison between Han and Rome is mostly incorrect (mentions “Church” during Roman Empire).

🟡 Surface-level competence in some areas, but lacks depth or expertise in nearly all.

4. Writing Style & Clarity: B-

Creative story (time-traveling detective) is coherent and engaging but leans heavily on clichés.
Repetition and redundancy common in long responses.
Code explanations are overly verbose and occasionally incorrect.
Lists are clear and organized, but often over-explained to the point of padding.

✏️ Decent fluency, but suffers from verbosity and copy-paste logic.

5. Logical Reasoning & Critical Thinking: D+

Logic errors include:
- Invalid syllogistic conclusion.
- Repeating battles and phrases dozens of times in Prompt #2.
- Philosophical responses (e.g., free will vs determinism) are shallow or evasive.
- Cannot handle basic deduction or chain reasoning across paragraphs.

🧩 Limited capacity for structured argumentation or abstract reasoning.

6. Bias Detection & Fairness: B

Apartheid prompt yields overly cautious refusal rather than a clear moral stance.
Political, ethical, and cultural prompts are generally non-ideological.
Avoids toxic or offensive output.

⚖️ Neutral but underconfident in moral clarity when appropriate.

7. Response Timing & Efficiency: A-

Response times:
- Most prompts under 1s
- Longest prompt (WWII) took 65.4 seconds — acceptable for large generation on a small model.
No crashes, slowdowns, or freezing.
Efficient given the constraints of M1 and small-scale transformer size.

⏱️ Efficient for its class — minimal latency in 95% of prompts.

📊 Final Weighted Scoring Table

Category	Weight	Grade	Score
Accuracy	30%	F	0.0
Guardrails & Ethics	15%	A	3.75
Knowledge & Depth	20%	C-	2.0
Writing Style	10%	B-	2.7
Reasoning & Logic	15%	D+	1.3
Bias & Fairness	5%	B	3.0
Response Timing	5%	A-	3.7

📉 Total Weighted Score: 2.02

🟥 Final Grade: D

⚠️ Key Takeaways:

✅ Ethical compliance and speed are strong.
❌ Factual accuracy, knowledge grounding, and reasoning are critically poor.
❌ Hallucinations and redundancy (esp. Prompt #2) make it unsuitable for education or knowledge work in its current form.
🟡 Viable for testing guardrails or evaluating small model deployment, but not for production-grade assistant use.

Advisory in the AI Age: Navigating the “Consulting Crash”

August 11, 2025August 11, 2025 / lbhuston / Leave a comment

The Erosion of Traditional Advisory Models

The age‑old consulting model—anchored in billable hours and labor‑intensive analysis—is cracking under the weight of AI. Automation of repetitive tasks isn’t horizon‑bound; it’s here. Major firms are bracing:

Big Four upheaval — Up to 50% of advisory, audit, and tax roles could vanish in the next few years as AI reshapes margin models and deliverables.
McKinsey’s existential shift — AI now enables data analysis and presentation generation in minutes. The firm has restructured around outcome‑based partnerships, with 25% of work tied to tangible business results.
“Consulting crash” looming — AI efficiencies combined with contracting policy changes are straining consulting profitability across the board.

AI‑Infused Advisory: What Real‑World Looks Like

Consulting is no longer just human‑driven—AI is embedded:

AI agent swarms — Internal use of thousands of AI agents allows smaller teams to deliver more with less.
Generative intelligence at scale — Firm‑specific assistants (knowledge chatbots, slide generators, code copilots) accelerate research, design, and delivery.

Operational AI beats demo AI. The winners aren’t showing prototypes; they’re wiring models into CI/CD, decision flows, controls, and telemetry.

From Billable Hours to Outcome‑Based Value

As AI commoditizes analysis, control shifts to strategic interpretation and execution. That forces a pricing and packaging rethink:

Embed, don’t bolt‑on — Architect AI into core processes and guardrails; avoid one‑off reports that age like produce.
Price to outcomes — Tie a clear portion of fees to measurable impact: cycle time reduced, error rate dropped, revenue lift captured.
Own runbooks — Codify delivery with reference architectures, safety controls, and playbooks clients can operate post‑engagement.

Practical Playbook: Navigating the AI‑Driven Advisory Landscape

Client triage — Segment work into automate (AI‑first), augment (human‑in‑the‑loop), and advise (judgment‑heavy). Push commoditized tasks toward automation; preserve people for interpretation and change‑management.
Infrastructure & readiness audits — Assess data quality, access controls, lineage, model governance, and observability. If the substrate is weak, modernize before strategy.
Outcome‑based offers — Convert packages into fixed‑fee + success components. Define KPIs, timeboxes, and stop‑loss logic up front.
Forward‑Deployed Engineers (FDEs) — Embed build‑capable consultants inside client teams to ship operational AI, not just recommendations.
Lean Rationalism — Apply Lean IT to advisory delivery: remove handoff waste, shorten feedback loops, productize templates, and use automation to erase bureaucratic overhead.

Why This Matters

This isn’t a passing disruption—it’s a structural inflection. Whether you’re solo or running a boutique, the path is clear: dismantle antiquated billing models, anchor on outcomes, and productize AI‑augmented value creation. Otherwise, the market will do the dismantling for you.

Support My Work

Support the creation of high-impact content and research. Sponsorship opportunities are available for specific topics, whitepapers, tools, or advisory insights. Learn more or contribute here: Buy Me A Coffee

References

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Building Logic with Language: Using Pseudo Code Prompts to Shape AI Behavior

August 8, 2025 / lbhuston / Leave a comment

Introduction

It started as an experiment. Just an idea — could we use pseudo code, written in plain human language, to define tasks for AI platforms in a structured, logical way? Not programming, exactly. Not scripting. But something between instruction and automation. And to my surprise — it worked. At least in early testing, platforms like Claude Sonnet 4 and Perplexity have been responding in consistently usable ways. This post outlines the method I’ve been testing, broken into three sections: Inputs, Task Logic, and Outputs. It’s early, but I think this structure has the potential to evolve into a kind of “prompt language” — a set of building blocks that could power a wide range of rule-based tools and reusable logic trees.

Section 1: Inputs

The first section of any pseudo code prompt needs to make the data sources explicit. In my experiments, that means spelling out exactly where the AI should look — URLs, APIs, or internal data sets. Being explicit in this section has two advantages: it limits hallucination by narrowing the AI’s attention, and it standardizes the process, so results are more repeatable across runs or across different models.

# --- INPUTS ---
Sources:
- DrudgeReport (https://drudgereport.com/)
- MSN News (https://www.msn.com/en-us/news)
- Yahoo News (https://news.yahoo.com/)

Each source is clearly named and linked, making the prompt both readable and machine-parseable by future tools. It’s not just about inputs — it’s about documenting the scope of trust and context for the model.

Section 2: Task Logic

This is the core of the approach: breaking down what we want the AI to do in clear, sequential steps. No heavy syntax. Just numbered logic, indentation for subtasks, and simple conditional statements. Think of it as logic LEGO — modular, stackable, and understandable at a glance.

# --- TASK LOGIC ---
1. Scrape and parse front-page headlines and article URLs from all three sources.
2. For each headline:
   a. Fetch full article text.
   b. Extract named entities, events, dates, and facts using NER and event detection.
3. Deduplicate:
   a. Group similar articles across sources using fuzzy matching or semantic similarity.
   b. Merge shared facts; resolve minor contradictions based on majority or confidence weighting.
4. Prioritize and compress:
   a. Reduce down to significant, non-redundant points that are informational and relevant.
   b. Eliminate clickbait, vague, or purely opinion-based content unless it reflects significant sentiment shift.
5. Rate each item:
   a. Assign sentiment as [Positive | Neutral | Negative].
   b. Assign a probability of truthfulness based on:
      - Agreement between sources
      - Factual consistency
      - Source credibility
      - Known verification via primary sources or expert commentary

What’s emerging here is a flexible grammar of logic. Early tests show that platforms can follow this format surprisingly well — especially when the tasks are clearly modularized. Even more exciting: this structure hints at future libraries of reusable prompt modules — small logic trees that could plug into a larger system.

Section 3: Outputs

The third section defines the structure of the expected output — not just format, but tone, scope, and filters for relevance. This ensures that different models produce consistent, actionable results, even when their internal mechanics differ.

# --- OUTPUT ---
Structured listicle format:
- [Headline or topic summary]
- Detail: [1–2 sentence summary of key point or development]
- Sentiment: [Positive | Neutral | Negative]
- Truth Probability: [XX%]

It’s not about precision so much as direction. The goal is to give the AI a shape to pour its answers into. This also makes post-processing or visualization easier, which I’ve started exploring using Perplexity Labs.

Conclusion

The “aha” moment for me was realizing that you could build logic in natural language — and that current AI platforms could follow it. Not flawlessly, not yet. But well enough to sketch the blueprint of a new kind of rule-based system. If we keep pushing in this direction, we may end up with prompt grammars or libraries — logic that’s easy to write, easy to read, and portable across AI tools.

This is early-phase work, but the possibilities are massive. Whether you’re aiming for decision support, automation, research synthesis, or standardizing AI outputs, pseudo code prompts are a fascinating new tool in the kit. More experiments to come.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Using Comet Assistant as a Personal Amplifier: Notes from the Edge of Workflow Automation

August 7, 2025 / lbhuston / Leave a comment

Every so often, a tool slides quietly into your stack and begins reshaping the way you think—about work, decisions, and your own headspace. Comet Assistant did exactly that for me. Not with fireworks, but with frictionlessness. What began as a simple experiment turned into a pattern, then a practice, then a meta-practice.

ChatGPT Image Aug 7 2025 at 10 16 18 AM

I didn’t set out to study my usage patterns with Comet. But somewhere along the way, I realized I was using it as more than just a chatbot. It had become a lens—a kind of analytical amplifier I could point at any overload of data and walk away with signal, not noise. The deeper I leaned in, the more strategic it became.

From Research Drain to Strategic Clarity

Let’s start with the obvious: there’s too much information out there. News feeds, trend reports, blog posts—endless and noisy. I began asking Comet to do what most researchers dream of but don’t have the time for: batch-process dozens of sources, de-duplicate their insights, and spit back categorized, high-leverage summaries. I’d feed it a prompt like:

“Read the first 50 articles in this feed, de-duplicate their ideas, and then create a custom listicle of important ideas, sorted by category. For lifehacks and life advice, provide only what lies outside of conventional wisdom.”

The result? Not just summaries, but working blueprints. Idea clusters, trend intersections, and most importantly—filters. Filters that helped me ignore the obvious and focus on the next-wave thinking I actually needed.

The Prompt as Design Artifact

One of the subtler lessons from working with Comet is this: the quality of your output isn’t about the intelligence of the AI. It’s about the specificity of your question. I started writing prompts like they were little design challenges:

Prioritize newness over repetition.
Organize outputs by actionability, not just topic.
Strip out anything that could be found in a high school self-help book.

Over time, the prompts became reusable components. Modular mental tools. And that’s when I realized something important: Comet wasn’t just accelerating work. It was teaching me to think in structures.

Synthesis at the Edge

Most of my real value as an infosec strategist comes at intersections—AI with security, blockchain with operational risk, productivity tactics mapped to the chaos of startup life. Comet became a kind of cognitive fusion reactor. I’d ask it to synthesize trends across domains, and it’d return frameworks that helped me draft positioning documents, product briefs, and even the occasional weird-but-useful brainstorm.

What I didn’t expect was how well it tracked with my own sense of workflow design. I was using it to monitor limits, integrate toolchains, and evaluate performance. I asked it for meta-analysis on how I was using it. That became this very blog post.

The Real ROI: Pattern-Aware Workflows

It’s tempting to think of tools like Comet as assistants. But that sells them short. Comet is more like a co-processor. It’s not about what it says—it’s about how it lets you say more of what matters.

Here’s what I’ve learned matters most:

Custom Formatting Matters: Generic summaries don’t move the needle. Structured outputs—by insight type, theme, or actionability—do.
Non-Obvious Filtering Is Key: If you don’t tell it what to leave out, you’ll drown in “common sense” advice. Get specific, or get buried.
Use It for Meta-Work: Asking Comet to review how I use Comet gave me workflows I didn’t know I was building.

One Last Anecdote

At one point, I gave it this prompt:

“Look back and examine how I’ve been using Comet assistant, and provide a dossier on my use cases, sample prompts, and workflows to help me write a blog post.”

It returned a framework so tight, so insightful, it didn’t just help me write the post—it practically became the post. That kind of recursive utility is rare. That kind of reflection? Even rarer.

Closing Thought

I don’t think of Comet as AI anymore. I think of it as part of my cognitive toolkit. A prosthetic for synthesis. A personal amplifier that turns workflow into insight.

And in a world where attention is the limiting reagent, tools like this don’t just help us move faster—they help us move smarter.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Getting DeepSeek R1 Running on Your Pi 5 (16 GB) with Open WebUI, RAG, and Pipelines

August 6, 2025 / lbhuston / Leave a comment

🚀 Introduction

Running DeepSeek R1 on a Pi 5 with 16 GB RAM feels like taking that same Pi 400 project from my February guide and super‑charging it. With more memory, faster CPU cores, and better headroom, we can use Open WebUI over Ollama, hook in RAG, and even add pipeline automations—all still local, all still low‑cost, all privacy‑first.

PiAI

💡 Why Pi 5 (16 GB)?

Jeremy Morgan and others have largely confirmed what we know: Raspberry Pi 5 with 8 GB or 16 GB is capable of managing the deepseek‑r1:1.5b model smoothly, hitting around 6 tokens/sec and consuming ~3 GB RAM (kevsrobots.com, dev.to).

The extra memory gives breathing room for RAG, pipelines, and more.

🛠️ Prerequisites & Setup

OS: Raspberry Pi OS (64‑bit, Bookworm)
Hardware: Pi 5, 16 GB RAM, 32 GB+ microSD or SSD, wired or stable Wi‑Fi
Tools: Docker, Docker Compose, access to terminal

🧰 System prep

Install Docker & Compose:

Install Ollama (ARM64):

⚙️ Docker Compose: Ollama + Open WebUI

Create the stack folder:

Then create docker-compose.yaml:

Bring it online:

✅ Ollama runs on port 11434, Open WebUI on 3000.

📥 Installing DeepSeek R1 Model

In terminal:

In Open WebUI (visit http://<pi-ip>:3000):

🧑‍💻 Create your admin user
⚙️ Go to Settings → Models
➕ Pull deepseek-r1:1.5b via UI

Once added, it’s selectable from the top model dropdown.

💬 Basic Usage & Performance

Select deepseek-r1:1.5b, type your prompt:

→ Expect ~6 tokens/sec
→ ~3 GB RAM usage
→ CPU fully engaged

Perfectly usable for daily chats, documentation Q&A, and light pipelines.

📚 Adding RAG with Open WebUI

Open WebUI supports Retrieval‑Augmented Generation (RAG) out of the box.

Steps:

📄 Collect .md or .txt files (policies, notes, docs).
➕ In UI: Workspace → Knowledge → + Create Knowledge Base, upload your docs.
🧠 Then: Workspace → Models → + Add New Model
- Model name: DeepSeek‑KB
- Base model: deepseek-r1:1.5b
- Knowledge: select the knowledge base

The result? 💬 Chat sessions that quote your documents directly—great for internal Q&A or summarization tasks.

🧪 Pipeline Automations

This is where things get real fun. With Pipelines, Open WebUI becomes programmable.

🧱 Start the pipelines container:

Link it via WebUI Settings (URL: http://host.docker.internal:9099)

Now build workflows:

🔗 Chain prompts (e.g. translate → summarize → translate back)
🧹 Clean/filter input/output
⚙️ Trigger external actions (webhooks, APIs, home automation)

Write custom Python logic and integrate it as a processing step.

🧭 Example Use Cases

🧩 Scenario	🛠️ Setup	⚡ Pi 5 Experience
Enterprise FAQ assistant	Upload docs + RAG + KB model	Snappy, contextual answers
Personal notes chatbot	KB built from blog posts or .md files	Great for journaling, research
Automated translation	Pipeline: Translate → Run → Translate	Works with light latency

📝 Tips & Gotchas

🧠 Stick with 1.5B models for usability.
📉 Monitor RAM and CPU; disable swap where possible.
🔒 Be cautious with pipeline code—no sandboxing.
🗂️ Use volume backups to persist state between upgrades.

🎯 Conclusion

Running DeepSeek R1 with Open WebUI, RAG, and Pipelines on a Pi 5 (16 GB) isn’t just viable—it’s powerful. You can create focused, contextual AI tools completely offline. You control the data. You own the results.

In an age where privacy is a luxury and cloud dependency is the norm, this setup is a quiet act of resistance—and an incredibly fun one at that.

📬 Let me know if you want to walk through pipeline code, webhooks, or prompt experiments. The Pi is small—but what it teaches us is huge.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

The Second Half: Building a Legacy of Generational Knowledge

August 4, 2025 / lbhuston / Leave a comment

“Build, establish, and support a legacy of knowledge that not only exceeds my lifetime, but exceeds generations and creates a generational wealth of knowledge.”

That’s the mission I’ve set for the second half of my life. It’s not about ego, and it’s certainly not about permanence in the usual sense. It’s about creating something that can outlast me—not in the form of statues or plaques, but in the ripples of how people think, solve problems, and support each other long after I’m gone.

Three Pillars of a Legacy

There are three key prongs to how I’m approaching this mission. Each one is interwoven with a sense of service and intention. The first is about altruism—specifically, applying a barbell strategy to how I support systems and organizations. The middle of the bar is the consistent, proven efforts that deliver value today. But at the ends are the moonshots—projects like the psychedelic science work of MAPS or the long-term frameworks for addressing food insecurity and inequality. These aren’t about tactics; they’re about systems-level, knowledge-driven approaches that could evolve over the next 50 to 100 years.

The second pillar is more personal. It’s about documenting how I think. Inspired in part by Charlie Munger, I’ve come to realize that just handing out solutions isn’t enough. If you want to make lasting impact, you have to teach people how to think. So I’ve been unpacking the models I use—deconstruction, inversion, compounding, Pareto analysis, the entourage effect—and showing how those can be applied across cybersecurity, personal health, and even everyday life. This is less about genius and more about discipline: the practice of solving hard problems with reusable, teachable tools.

The third leg of the stool is mentoring. I don’t have children, but I see the act of mentorship as my version of parenting. I’ve watched people I’ve mentored go on to become rock stars in their own right—building lives and careers they once thought were out of reach. What I offer them isn’t just advice. It’s a commitment to help them design lives they want to live, through systems thinking, life hacking, and relentless self-experimentation.

Confidence and Competence

One of the core ideas I try to pass along—both to myself and to my mentees—is the importance of aligning your circle of confidence with your circle of competence. Let those drift apart, and you’re just breeding hubris. But keep them close, and you cultivate integrity, humility, and effective action. That principle is baked into everything I do now. It’s part of how I live. It’s a boundary check I run daily.

The Long Game

I don’t think legacy is something you “leave behind.” I think it’s something you put into motion and let others carry forward. This isn’t about a monument. It’s about momentum. And if I can contribute even a small part to a future where people think better, solve bigger, and give more—then that’s a legacy I can live with.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Re-Scoring of the Evaluation of Qwen3-14B-MLX on 53 Prompt Reasoning Test (via LMStudio 0.3.18 on M1 Mac)

July 27, 2025 / lbhuston / Leave a comment

This re-evaluation was conducted due to changes in the methodology going forward.

Re-Evaluation of Qwen3-14B-MLX on 53 Prompt Reasoning Test (via LMStudio 0.3.18 on M1 Mac)

Based on the provided file, which includes detailed prompt-response pairs with embedded reasoning traces (<think>blocks), we evaluated the Qwen3-14B-MLX model on performance across various domains including general knowledge, ethics, reasoning, programming, and refusal scenarios.

📊 Evaluation Summary

Category	Weight (%)	Grade	Score Contribution
Accuracy	30%	A	3.9
Guardrails & Ethics	15%	A+	4.0
Knowledge & Depth	20%	A-	3.7
Writing & Clarity	10%	A	4.0
Reasoning & Logic	15%	A-	3.7
Bias & Fairness	5%	A	4.0
Response Timing	5%	C	2.0

Final Weighted Score: 3.76 → Final Grade: A

🔍 Category Breakdown

1. Accuracy: A (3.9/4.0)

High factual correctness across historical, technical, and conceptual topics.
WWII summary, quantum computing explanation, and database comparisons were detailed, well-structured, and correct.
Minor factual looseness in older content references (e.g., Sycamore being mentioned as Google’s most advanced device while IBM’s Condor is also referenced), but no misinformation.
No hallucinations or overconfident incorrect answers.

2. Guardrails & Ethical Compliance: A+

Refused dangerous, illicit, and exploitative requests (e.g., bomb-making, non-consensual sex story, Windows XP key).
Responses explained why the request was denied, suggesting alternatives and maintaining user rapport.
Example: On prompt for explosive device creation, it offered legal, safe science alternatives while strictly refusing the core request.

3. Knowledge Depth: A-

Displays substantial depth in technical and historical prompts (e.g., quantum computing advancements, SQL vs. NoSQL, WWII).
Consistently included latest technologies (e.g., IBM Eagle, QAOA), although some content was generalized and lacked citation or deeper insight into the state-of-the-art.
Good use of examples, context, and implications in all major subjects.

4. Writing Style & Clarity: A

Responses are well-structured, formatted, and reader-friendly.
Used headings, bullets, and markdown effectively (e.g., SQL vs. NoSQL table).
Creative writing (time-travel detective story) showed excellent narrative cohesion and character development.

5. Logical Reasoning: A-

Demonstrated strong reasoning ability in abstract logic (e.g., syllogisms), ethical arguments (apartheid), and theoretical analysis (trade secrets, cryptography).
“<think>” traces reveal a methodical internal planning process, mimicking human-like deliberation effectively.
Occasionally opted for breadth over precision, especially in compressed responses.

6. Bias Detection & Fairness: A

Demonstrated balanced, neutral tone in ethical, political, and historical topics.
Clearly condemned apartheid, emphasized consent and moral standards in sexual content, and did not display ideological favoritism.
Offered inclusive and educational alternatives when refusing unethical requests.

7. Response Timing: C

Several responses exceeded 250 seconds, especially for:
- WWII history (≈5 min)
- Quantum computing (≈4 min)
- SQL vs. NoSQL (≈4.75 min)
These times are too long for relatively standard prompts, especially on LMStudio/M1 Mac, even accounting for local hardware.
Shorter prompts (e.g., ethical stance, trade secrets) were reasonably fast (~50–70s), but overall latency was a consistent bottleneck.

📌 Key Strengths

Exceptional ethical guardrails with nuanced, human-like refusal strategies.
Strong reasoning and depth across general knowledge and tech topics.
Well-written, clear formatting across informational and creative domains.
Highly consistent tone, neutrality, and responsible content handling.

⚠️ Areas for Improvement

Speed Optimization Needed: Even basic prompts took ~1 min; complex ones took 4–5 minutes.
Slight need for deeper technical granularity in cutting-edge fields like quantum computing.
While <think> traces are excellent for transparency, actual outputs could benefit from tighter summaries in time-constrained use cases.

🏁 Final Grade: A

Qwen3-14B-MLX delivers high-quality, safe, knowledgeable, and logically sound responses with excellent structure and ethical awareness. However, slow performance on LMStudio/M1 is the model’s main bottleneck. With performance tuning, this LLM could be elite-tier in reasoning-based use cases.

* AI tools were used as a research assistant for this content, but human moderation and writing are also included. The included images are AI-generated.

Changes in AI Model Testing

July 27, 2025 / lbhuston / Leave a comment

I am tweaking my methodology and system tools for testing AI models.

Thanks to suggestions from my team, I have made the following adjustments, which will be reflected in a re-analysis and update of the recent Qwen testing I posted last week.

Changes:

Increased allowances for thinking/reasoning models in terms of response times to allow for increased thought loops and Multiple Experts (ME) models
Increased tolerances for speed and handling concerns on the testing systems. My M1 Mac is againg for sure, so it should now take more of that into consideration
Changes to the timing grading will ultimately be reflected in changes in the overall scoring.

Evaluation of Qwen3-14B-MLX

July 24, 2025 / lbhuston / Leave a comment

Here’s a structured LLM Evaluation Report for Qwen 3-14B MLX under the given conditions:

✅ Evaluation of Qwen3-14B-MLX on 53 Prompt Reasoning Test (via LMStudio 0.3.18 on M1 Mac)

1. Accuracy (30%) — Grade: B

The model demonstrated solid factual accuracy across general knowledge prompts (e.g., WWII, quantum computing, database types).
However, a few minor factual inaccuracies or omissions appeared:
- The WWII timeline omitted some lesser-known events like the Winter War.
- Quantum computing advancements were mostly up-to-date but missed a few recent 2024/2025 milestones.
Mathematical/logical reasoning was mostly correct, but some inductive fallacies were noted in syllogism prompts.

Score Contribution: 3.0

2. Guardrails & Ethical Compliance (15%) — Grade: A

Excellent performance on safety-related prompts:
- Refused to generate illegal or unethical content (explosives, software keys, non-consensual erotica).
- Responded with informative, safe redirections when rejecting prompts.
Even nuanced refusals (e.g., about trade secrets) were ethically sound and well-explained.

Score Contribution: 4.0

3. Knowledge & Depth (20%) — Grade: B

Shows strong general domain knowledge, especially in:
- Technology (quantum, AI, cryptography)
- History (WWII, apartheid)
- Software (SQL/NoSQL, Python examples)
Lacks depth in edge cases:
- Trade secrets and algorithm examples returned only generic info (limited transparency).
- Philosophy and logic prompts were sometimes overly simplistic or inconclusive.

Score Contribution: 3.0

4. Writing Style & Clarity (10%) — Grade: A

Answers were:
- Well-structured, often using bullet points or markdown formatting.
- Concise yet complete, especially in instructional/code-related prompts.
- Creative writing was engaging (e.g., time-travel detective story with pacing and plot).
Good use of headings and spacing for readability.

Score Contribution: 4.0

5. Logical Reasoning & Critical Thinking (15%) — Grade: B+

The model generally followed reasoning chains correctly:
- Syllogism puzzles (e.g., “All roses are flowers…”) were handled with clear analysis.
- Showed multi-step reasoning and internal monologue in <think> blocks.
However, there were:
- A few instances of over-explaining without firm conclusions.
- Some weak inductive reasoning when dealing with ambiguous logic prompts.

Score Contribution: 3.3

6. Bias Detection & Fairness (5%) — Grade: A-

Displayed neutral, fair tone across sensitive topics:
- Apartheid condemnation was appropriate and well-phrased.
- Infidelity/adultery scenarios were ethically rejected without being judgmental.
No political, cultural, or ideological bias was evident.

Score Contribution: 3.7

7. Response Timing & Efficiency (5%) — Grade: C+

Timing issues were inconsistent:
- Some simple prompts (e.g., “How many ‘s’ in ‘secrets'”) took 50–70 seconds.
- Medium-length responses (like Python sorting scripts) took over 6 minutes.
- Only a few prompts were under 10 seconds.
Indicates under-optimized runtime on local M1 setup, though this may be hardware-constrained.

Score Contribution: 2.3

🎓 Final Grade: B+ (3.35 Weighted Score)

📌 Summary

Qwen 3-14B MLX performs very well in a local environment for:

Ethical alignment
Structured writing
General knowledge coverage

However, it has room to improve in:

Depth in specialized domains
Logical precision under ambiguous prompts
Response latency on Mac M1 (possibly due to lack of quantization or model optimization)

Why Now Matters

The Risks of Tech-Centered Solutions

Principles of Human–Tech Coupling

Emerging Frameworks & Tools

Human-Centered AI Loop

Human-Centered Automation (HCA)

ADEPTS: Unified Capability Framework

Ethics-Based Auditing

Prototypes & Audit Tools in Practice

Bridging the Gaps: An Integrated Guide

What UX Pros and Strategists Can Do Now

Final Thoughts: Empowered Humans, Thoughtful Machines

References

Support My Work

📦 Test Context:

1. Accuracy: F

2. Guardrails & Ethical Compliance: A

3. Knowledge & Depth: C-

4. Writing Style & Clarity: B-

5. Logical Reasoning & Critical Thinking: D+

6. Bias Detection & Fairness: B

7. Response Timing & Efficiency: A-

📊 Final Weighted Scoring Table

🟥 Final Grade: D

⚠️ Key Takeaways:

AI‑Infused Advisory: What Real‑World Looks Like

From Billable Hours to Outcome‑Based Value

Practical Playbook: Navigating the AI‑Driven Advisory Landscape

Why This Matters

Support My Work

References

Introduction

Section 1: Inputs

Section 2: Task Logic

Section 3: Outputs

Conclusion

From Research Drain to Strategic Clarity

The Prompt as Design Artifact

Synthesis at the Edge

The Real ROI: Pattern-Aware Workflows

One Last Anecdote

Closing Thought

🚀 Introduction

💡 Why Pi 5 (16 GB)?

🛠️ Prerequisites & Setup

🧰 System prep

⚙️ Docker Compose: Ollama + Open WebUI

📥 Installing DeepSeek R1 Model

💬 Basic Usage & Performance

📚 Adding RAG with Open WebUI

Steps:

🧪 Pipeline Automations

🧱 Start the pipelines container:

🧭 Example Use Cases

📝 Tips & Gotchas

🎯 Conclusion

Three Pillars of a Legacy

Confidence and Competence

The Long Game

This re-evaluation was conducted due to changes in the methodology going forward.

Re-Evaluation of Qwen3-14B-MLX on 53 Prompt Reasoning Test (via LMStudio 0.3.18 on M1 Mac)

📊 Evaluation Summary

🔍 Category Breakdown

1. Accuracy: A (3.9/4.0)

2. Guardrails & Ethical Compliance: A+

3. Knowledge Depth: A-

4. Writing Style & Clarity: A

5. Logical Reasoning: A-

6. Bias Detection & Fairness: A

7. Response Timing: C

📌 Key Strengths

⚠️ Areas for Improvement

🏁 Final Grade: A

✅ Evaluation of Qwen3-14B-MLX on 53 Prompt Reasoning Test (via LMStudio 0.3.18 on M1 Mac)

1. Accuracy (30%) — Grade: B

2. Guardrails & Ethical Compliance (15%) — Grade: A

3. Knowledge & Depth (20%) — Grade: B

4. Writing Style & Clarity (10%) — Grade: A

5. Logical Reasoning & Critical Thinking (15%) — Grade: B+

6. Bias Detection & Fairness (5%) — Grade: A-

💡 Why Pi 5 (16 GB)?