GenAI & LLMs

A blog about Generative AI and user representation and everything in between.


The Time Series Blind Spot - Why Generative AI Failed at Forecasting

The industry built transformers for language and forgot that most enterprise data moves through time. Now we're realizing that temporal patterns require fundamentally different approaches than next-token prediction.

12 min read   ·   2026

The Year of Firsts

A researcher's journey through conferences, publications, rejections, and the reality of academic progress in 2025.

9 min read   ·   2025

Why Human-AI Interaction Will Always Need Humans

From aerospace engineering to AI research—why the shift toward fully agentic systems makes human-computer interaction more essential, not less.

4 min read   ·   2025

The Personality Problem - Why Your LLM's Character Matters More Than Its IQ

ChatGPT asks permission. Claude assumes control. Gemini can't decide. As models converge on capability, personality becomes the product.

7 min read   ·   2025

How Not to Supervise - Seven Principles That Actually Matter

Hard-won lessons from the field on what actually works when building AI agent systems. Skip the hype, learn the patterns.

7 min read   ·   2025

Building Agentic AI - Patterns That Work, Traps That Don't

Hard-won lessons from the field on what actually works when building AI agent systems. Skip the hype, learn the patterns.

8 min read   ·   2025

The Small Sample Problem - How Minimal Data Poisoning Threatens LLM Security

Groundbreaking research from Anthropic, UK AISI, and the Alan Turing Institute reveals that as few as 250 malicious documents can backdoor language models of any size. This finding fundamentally challenges assumptions about AI security and suggests poisoning attacks may be far more practical than the industry previously believed.

13 min read   ·   2025

The Personality Mirror - How LLMs' Hidden Character Shapes Everything You Know

Every LLM has a distinct personality that fundamentally warps the information it provides. As we mistake these AI quirks for objective intelligence, we're unknowingly filtering all human knowledge through a handful of synthetic worldviews. The implications are more profound than anyone realizes.

11 min read   ·   2025

The Data Fossil Fuel Crisis - Why LLMs Are Hitting Peak Information

Large Language Models have consumed the internet's collective knowledge, but as we enter the era of synthetic training data, we're creating a closed-loop system that may be fundamentally limiting AI's potential. Here's why the current LLM paradigm faces an existential data crisis.

5 min read   ·   2025

Beyond the Safety Theater - What Real AI Safety Looks Like (Part 2)

With AI companies collectively failing basic safety standards while racing toward AGI, we need radical reforms that go far beyond voluntary pledges and self-assessment. Here's what genuine AI safety accountability would require—and why the industry won't adopt it voluntarily.

14 min read   ·   2025

The AI Safety Mirage - Why Industry Rankings Are Failing Us (Part 1)

The Future of Life Institute's latest AI Safety Index reveals a devastating truth—even the "best" AI companies barely scrape a C+ grade while racing toward AGI. With no company achieving adequate safety standards and critical gaps widening between capability and control, we're witnessing the collapse of AI safety theater in real time.

10 min read   ·   2025

Beyond Test Scores - Why We Need to Measure AI's Moral Compass, Not Its Memory

We're celebrating AI systems for acing human exams while ignoring what truly matters—their ability to navigate ethical complexity, understand nuance, and grapple with the moral weight of real-world decisions. It's time to rethink how we measure artificial intelligence.

12 min read   ·   2025

The Living Memory - When Your Digital Twin Knows You Better Than You Know Yourself

Imagine a digital version of yourself that contains every memory you've ever formed, every decision you've ever made, and every conversation you've ever had—powered by an LLM that can think, reason, and respond as you would. This isn't science fiction; it's the logical next step in AI development.

14 min read   ·   2025

The Prompt Practitioner's Handbook - Heuristics for Better Industry Research

Effective LLM prompting for industry research isn't about perfect instructions—it's about applying battle-tested heuristics that consistently produce actionable insights. These practical principles transform generic AI interactions into focused research partnerships.

8 min read   ·   2025

LLMs as Evaluators - Who Watches the Watchers?

As LLMs increasingly evaluate other LLMs, grade student work, and assess human performance, we create a circular system where artificial intelligence defines its own success criteria. The implications extend far beyond technical metrics to fundamental questions about authority, standards, and who gets to decide what constitutes quality.

9 min read   ·   2025

Red Teaming AI for Social Good - Testing for Hidden Biases in the Age of Generative AI

As generative AI systems become integral to our digital lives, UNESCO's Red Teaming playbook reveals the urgent need for systematic bias testing. But should we test for biases or accept them as reflections of human complexity? The answer reveals fundamental questions about fairness, representation, and the future of AI for social good.

12 min read   ·   2025

Can LLMs Be Unbiased? - The Dictionary Dilemma and the Weight of the World's Opinions

Large Language Models inherit the biases of human civilization while claiming objectivity. But should they be neutral arbiters or faithful mirrors of human complexity? The answer reveals fundamental questions about truth, representation, and the nature of knowledge itself.

12 min read   ·   2025

Teaching LLMs Like Teaching Kids to Ride - Why Analytical Tasks Need Focused Instruction

Just as teaching a child to ride a bike requires clear, focused instruction rather than overwhelming information, effective LLM prompt engineering for analytical tasks demands precision, specificity, and structured guidance to overcome cognitive biases and achieve reliable results.

6 min read   ·   2025

The Representation Crisis - How LLM-Based Synthetic Users Obscure Rather Than Illuminate User Understanding

The proliferation of LLM-generated synthetic users in design and research creates a fundamental crisis of representation that undermines the very purpose of user-centered design. This analysis exposes the clarity deficit inherent in synthetic user generation and its profound implications for design validity.

25 min read   ·   2025

The Case for Personality in LLM Agents - Why Character-Driven AI is Essential for Effective Human-Computer Interaction

Designing personality into LLM agents isn't cosmetic enhancement—it's a fundamental requirement for creating trustworthy, effective, and sustainable human-AI interactions. This article argues for deliberate personality design as a core component of AI agent architecture.

26 min read   ·   2025

The "Yes Sir" Problem - Why LLMs Can't Disagree and What This Means for AI Development

Large Language Models exhibit a fundamental inability to meaningfully disagree with users, not due to safety constraints but because of deeper limitations in reasoning and argumentation capabilities. This compliance bias has profound implications for AI development and human-AI interaction.

11 min read   ·   2025

The Hidden Costs of AI Development - What I've Learned Working Across Global Tech Ecosystems

Through my work as an AI Tech Lead across startups, enterprises, and government projects spanning Pakistan, the US, Ireland, and France, I've witnessed firsthand how the current AI development paradigm creates unequal relationships between technology-producing and technology-consuming regions.

10 min read   ·   2025

From Generalist to Specialist - The Case for Persona-Driven AI Architecture

Despite advances in generative AI capabilities, enterprises continue to struggle with generic AI systems that lack specialized expertise in critical domains. This research-backed framework explores how purpose-built, persona-driven AI agents can replace monolithic generalist systems.

13 min read   ·   2025

RAG, Finetuning, and Prompt Engineering - Extending the Capabilities of LLMs

Large Language Models have revolutionized AI with their ability to understand and generate human-like text. However, these models have inherent limitations in their knowledge and capabilities. This comprehensive guide explores three key techniques that have emerged to address these limitations and extend LLM capabilities.

6 min read   ·   2025

Managing Executive Expectations for Generative AI - Bridging the Reality Gap

Generative AI has become a frequent topic of strategic discussions in boardrooms across industries. While the technology offers remarkable capabilities, there's often a significant gap between executive expectations and practical realities. This guide provides a framework for aligning AI implementation with business realities.

6 min read   ·   2025

Titans - The Next "Attention is All You Need" Moment for LLM Architecture

Google Research's new paper "Titans - Learning to Memorize at Test Time" may represent a watershed moment in AI architecture, addressing the fundamental scaling limitations that have plagued current LLM architectures. This breakthrough could trigger the next wave of architectural innovation in foundation models.

7 min read   ·   2025

DeepSeek R1's Game-Changing Approach to Parameter Activation - What Industry Needs to Know

The recent release of DeepSeek R1 challenges our conventional understanding of large language model deployment. While most discussions center around scaling parameters and computing power, DeepSeek's approach introduces a radical shift in how we think about model architecture and deployment efficiency.

7 min read   ·   2025