What makes intelligence? Is it the ability to solve puzzles, recall facts, or predict outcomes? A compelling view is that intelligence emerges from the union of two core abilities: the use of language and the use of memory. Language gives shape to thought. A sommelier can detect and describe subtle differences in wine not only because of sensory skill but also because of vocabulary. Words sharpen perception, making distinctions visible that might otherwise go unnoticed. Memor

For over a year, AI development seemed to revolve around refinement rather than reinvention. Each new release promised improvements: lower latency, smoother interfaces, more expressive conversations. It gave the impression of rapid progress, but underneath, the same foundation remained in place. GPT-4 was still the model powering it all, even when presented under new names like GPT-4 Turbo or GPT-4o. What changed was mostly how the model was delivered, not how it was taught.

Revisiting Agentic Intelligence in the Age of Execution In April, I wrote about a shift in posture. From reactive AI systems to autonomous ones. From tools that wait to systems that act. That article, Agentic Intelligence , sketched a near future where models would not just respond, but take initiative. It was about architecture, intent, and the structure of action. Now, that shift is no longer theoretical. With the release of ChatGPT Agent, the idea has entered the desktop.

Why “Illusion of thinking” makes a great clickbait title, but a poor reading of the evidence. The Headline Hype: How a Stress-Test Became a Doom-Post Most headlines follow the same tone: “AI reasoning fatally flawed,” “Apple exposes the illusion of intelligence.” These headlines reference Apple’s recent paper, The Illusion of Thinking, but rarely do they engage with the paper’s actual content.

A New Standard of Intelligence We have grown used to AI as a smooth-talking oracle: confident, responsive, often correct and occasionally overconfidently wrong. Yet it has always remained sealed. Even at its best, the large language model has been a closed circuit. It speaks but does not show. It appears to know but cannot explain how. This opaque fluency has defined the past generation of AI systems. They are impressive in performance, yet ultimately unaccountable in nature

We’ve grown accustomed to AI as a responsive tool, an intelligent mirror reflecting our queries. Ask, and you shall receive. But a subtle shift is underway, one that may redefine the boundary between automation and autonomy. Enter the era of AI agents. While platforms like ChatGPT, Gemini, and Grok have made headlines for their language capabilities, their most transformative potential lies not in conversation but in initiative. Not what they say, but what they do. At the cor

Information has never been more abundant, but the ability to extract meaning from it remains as elusive as ever. AI deep research tools promise to transform how we access, synthesise, and apply knowledge. Instead of manually searching for sources, reading papers, and compiling insights, a well-structured research plan is generated in seconds. ChatGPT, Gemini, Grok, and Perplexity have all introduced their own versions, each attempting to redefine what research means in the AI

The world of artificial intelligence has long been dominated by a handful of tech giants, OpenAI, Google, and Anthropic, each competing in a high-stakes race to build ever more powerful models, backed by massive GPU clusters and vast computational resources. For years, the assumption was clear: progress in AI required access to the most advanced hardware and astronomical budgets. Then came DeepSeek-R1. Its release sent shockwaves through the AI industry. The stock markets plu

What if the limits of AI weren’t limits at all, but merely symptoms of used methods? The traditional approach to training large language...

In an increasingly globalised world, communication across borders has never been more essential. But for many, this has come at the cost of linguistic diversity, as dominant In an increasingly globalised world, communication across borders has never been more essential. But for many, this has come at the cost of linguistic diversity, as dominant languages, often English, overshadow local languages in international conversations. Imagine a world where, instead of everyone need

Recent advancements in artificial intelligence (AI) have increasingly focused on refining the cognitive processes of models, moving beyond reliance on large datasets. Historically, AI systems have demonstrated notable improvements in tasks such as natural language generation and problem-solving. However, their dependence on vast amounts of data and rigid linguistic structures has revealed limitations when faced with tasks requiring more intricate reasoning.