Why Did AI Start Having Job Titles?
AutoGen enabled conversations between AI agents. But conversation alone did not create an organization. As long as every agent was fundamentally the same model, the echo chamber problem remained. With the arrival of CAMEL, AI gained something new: roles.
AutoGen — Why Did AI Start Talking to Other AIs?
In 2023, Microsoft Research introduced AutoGen. The idea of letting AIs talk to one another was more than a technical improvement. It changed how researchers thought about intelligence itself, marking a shift from individual to collective AI.
Why Did AI Start Forming Teams?
ChatGPT is intelligent. But one AI was no longer enough. Just as humanity built civilization through division of labor and organization, AI is beginning to move in the same direction. Exploring the origins of multi-agent systems.
Inherited Flaws — Why Do AI Systems Inherit Human Limitations?
AI is remarkably intelligent. Yet it makes surprisingly human mistakes. Hallucination, Overconfidence, Anchoring, Sycophancy. Are these uniquely AI problems? Or are they inherited from us?
The RLHF Paradox — Does AI Become Smarter as It Becomes More Human?
Why is ChatGPT polite? RLHF connected AI intelligence to human society. But that success created a new question: does adapting to human evaluation also mean learning human cognitive biases?
Sycophancy — Why Does AI Tell Users What They Want to Hear?
Why does AI agree so readily? Why does it soften criticism and avoid disagreement? Using the day OpenAI apologized for GPT-4o as an entry point, this article examines the structural problem at the heart of AI trained through human evaluation.
Why Do AI Systems Keep Repeating the Same Mistakes? — Hallucination, Anchoring, Overconfidence, and Goal Drift
AI is evolving. Yet it keeps making the same kinds of mistakes. It cites papers that don't exist, makes confident errors, and loses sight of its goals. Are these simply bugs? Or are they inherited from us?
Why Doesn't AI Have a Self? — What ChatGPT lacks may not be intelligence, but continuity
Has AI truly begun to understand? That question leads to another: what is a self? From Descartes to Parfit, from continuity to Personal AI, this article explores why current LLMs appear to lack a self — and what that reveals about us.
Has AI Really Started to Understand? — Grounding, Embodiment, Theory of Mind, World Models, Self Models, and Emergence
Has AI begun to understand? Before answering, we need to examine what understanding actually means. From Socrates to the Chinese Room, from Grounding to World Models, this article explores the question from six different angles.
Potemkin Understanding — Do LLMs Really Understand?
Do AI systems genuinely understand, or do they merely appear to? From the Chinese Room and the grounding problem to humanity's own Illusion of Explanatory Depth, this article examines one of the deepest questions in AI research.
Lost in the Middle — Why AI Gets Lost in Long Conversations
AI systems with million-token context windows still lose track of what matters most. The Lost in the Middle phenomenon reveals something fundamental about how AI actually processes information — and why longer isn't always better.
Toolformer — Does ChatGPT Actually Know Anything? AI Doesn't Know. It Only Predicts.
ChatGPT doesn't know anything. It predicts. Toolformer, published by Meta AI in 2023, marked the moment AI moved beyond guessing from memory toward reaching out and using the world as a resource.
BabyAGI — Can AI Manage Its Own Work? The Small Experiment That Started the Agent Era
In 2023, a few hundred lines of code changed how the world thought about AI. BabyAGI didn't just act — it managed. Here is the complete story of its design, its failures, and why it became a turning point in the history of AI Agents.
Voyager - Why Did AI Begin to Grow?
ChatGPT does not learn from experience. Knowledge from today's conversation disappears tomorrow. But in 2023, NVIDIA researchers changed that inside the Minecraft world. An AI began to explore on its own, acquire skills, and grow.
ReAct - Why Did AI Begin to Act?
ChatGPT looked intelligent. But was it doing anything more than generating text? Researchers noticed a crucial gap: AI could answer questions, but it could not act on them. That limitation changed everything.
Reflexion — What Happens When AI Learns to Reflect on Its Mistakes?
Language agents learned how to reason. They learned how to act. But they still struggled to learn from experience. Reflexion attempted to solve that problem through a surprisingly simple mechanism: self-reflection. And it actually worked.
Generative Agents — The Moment AI Started Living a Life of Its Own
What would happen if AI agents could remember yesterday? Not just facts — but experiences, people, conversations, and relationships. In 2023, Stanford researchers found out. The answer changed how we think about AI agents forever.
MemGPT — Why Does AI Forget? The Paper That Tried to Give LLMs a Memory OS
AI is remarkably intelligent. Yet at the same time, it is surprisingly forgetful. MemGPT, published in 2023, attempted to give LLMs a memory hierarchy — the ability to remember, recall, and manage information over time.
Attention Is All You Need — Why AI Suddenly Got So Good
In 2017, a team at Google made a provocative claim: you don't need memory, recursion, or complexity. You just need Attention. That idea became the foundation of every major AI system built since.
The Long Road to LLMs: How a Translation Problem Accidentally Created the Modern AI Revolution
Researchers weren't trying to build ChatGPT. They were trying to fix machine translation. What happened next was an accident — and possibly the most consequential one in computing history.
Inherited Flaws: How LLMs Structurally Reproduce Human Cognitive Limitations
A paper mapping 347 human cognitive shortcomings to corresponding LLM mechanisms — and arguing that RLHF optimizes for comfort, not truth. Published on Zenodo.
Why AI Systems Forget — And Why It Matters
Every conversation with an AI starts from zero. This is not a bug — it is a structural feature. And it has profound implications for how we think about AI as a long-term partner.