Blog

The promise of nuclear fusion—clean, virtually limitless energy—has implications far beyond the lab. If commercial fusion becomes viable, it could profoundly impact global energy markets, particularly in regions whose economies are heavily reliant on fossil fuels. Among those, the Middle East stands at the forefront.

For decades, many Middle Eastern countries, especially members of OPEC such as Saudi Arabia, the UAE, Kuwait, and Iraq, have built their wealth and geopolitical leverage around oil exports. Energy exports fund their national budgets, drive infrastructure projects, and underpin their global political influence. A shift away from fossil fuels would pose both a threat and an opportunity.

The Challenge to Oil-Dependent Economies

Nuclear fusion, if successfully deployed at scale, could displace a large portion of the world’s demand for oil and gas. Unlike nuclear fission, fusion doesn’t produce long-lived radioactive waste and carries minimal risk of meltdown, making it far more acceptable to the general public and policymakers. Fusion could be the energy source of choice for developed and developing countries alike.

If fusion were to significantly lower the cost of electricity generation, demand for oil—especially for power production and transportation—could decline sharply. This would mean lower global oil prices, potentially destabilizing economies that have not diversified beyond hydrocarbons. Countries with high dependency on oil revenues could face budget deficits, currency devaluation, and social unrest unless they act proactively.

The Push Toward Diversification

Fortunately, several Middle Eastern states have already recognized this looming reality. Saudi Arabia’s Vision 2030 and the UAE’s investments in solar energy and technology sectors are examples of national strategies aimed at reducing reliance on oil. The advent of fusion could accelerate these efforts, forcing a quicker pivot to sectors like tourism, finance, logistics, education, and clean energy technology.

Interestingly, fusion technology itself might become a new field of investment and innovation in the region. With the financial power and ambition of nations like the UAE and Saudi Arabia, it’s conceivable that these countries could become partners or even leaders in commercializing fusion technologies—investing in startups, establishing research centers, or importing reactors once the technology matures.

Geopolitical Rebalancing

Fusion could also shift the balance of power in the region. Countries less dependent on oil, or those already investing in technological transformation, may come out stronger. In contrast, nations with stagnant economies and heavy fossil fuel dependency may face deeper political and economic struggles.

Moreover, global strategic alliances might evolve. Energy-importing countries—particularly in Asia and Europe—may become less tied to Middle Eastern oil producers. This could diminish the region’s geopolitical influence but also reduce the strategic vulnerabilities associated with energy supply routes.

Conclusion

Nuclear fusion has the potential to be as disruptive to the Middle East as the discovery of oil once was in the early 20th century. The outcome, however, will depend on how individual countries respond. Those who embrace change, invest in diversification, and position themselves in the fusion-powered future stand to thrive. Those who resist may find themselves left behind in a world no longer powered by oil.

Imagine a version of English where cat, dog, and car stayed exactly the same—whether you’re talking about one or twenty. No more cats, dogs, or cars. It might sound strange at first, but many languages around the world, such as Japanese, Chinese, and Thai, don’t use plural forms in the way English does. So what would English be like without plurals?

First, our sentences would rely much more on context and quantity indicators. Instead of saying “five books,” we might say “five book”—and our brains would simply learn that the number already implies plurality. In fact, we already do this in some cases. Think about the word fish—it’s both singular and plural. You’d say “I caught a fish” or “I caught five fish,” and nobody gets confused.

Removing plurals could also make English grammar simpler for learners. Irregular plurals like children, geese, and mice cause headaches for many. Without plurals, we could eliminate these quirks entirely.

However, plural markers do provide clarity in cases where numbers aren’t explicitly mentioned. “Books are on the table” clearly signals more than one item. If we dropped plurals, we might rely more heavily on other words, like “many,” “some,” or “a lot of.” That would shift the cognitive load from morphology (word forms) to syntax (sentence structure).

Would we lose something expressive or poetic? Possibly. English often uses pluralization in clever or artistic ways—just look at how thoughts, dreams, or feelings change subtly when pluralized. The plural form can give a sense of scope, repetition, or abstraction.

In short, if English didn’t have plurals, it would likely become more streamlined and perhaps more context-dependent. It might look a little more like Japanese in structure, but it would still retain the ability to communicate plurality through numbers and descriptive words. It’s a fun linguistic thought experiment—and a reminder that language evolves in many different ways around the world.

Headline prediction (with caveats): By 2045, China will likely emerge as a far stronger peer competitor to the U.S. in certain theatres—especially the Indo-Pacific and adjacent maritime, space, and cyber domains—but it will still face significant constraints. The U.S. will retain advantages in global power projection, alliances, and military-technology depth, but the margin will be narrower. The resulting strategic landscape will be one of intense competition, deterrence, and risk of escalation, rather than clear dominance by either side.

Below is a breakdown of how I see various dimensions evolving, followed by key risks and policy implications.

(more…)

Fifty years from now, it’s quite possible that many industries will be dominated by a handful of global players, due to network effects, economies of scale, and AI-driven optimization. However, I still believe that small companies will have opportunities to grow, innovate, and even become dominant forces — though the path may be very different from what it is today.

History shows us that no empire lasts forever. Whether we’re talking about Standard Oil, Nokia, or MySpace, many once-dominant players either lost their edge or were disrupted. Technology cycles, regulatory shifts, and changing customer behavior create windows for smaller, more agile players to break through. The question is not whether it’s possible, but what conditions will allow it.

In a future with heavy monopoly or oligopoly tendencies, new players might find success in niches that the big players ignore — often because those markets are too small for giants to bother with, or because the incumbents have become too risk-averse. AI and open technologies could also be a double-edged sword: while large companies will have better access to proprietary data and training infrastructure, open-source ecosystems and cloud platforms will still empower small teams to build and scale quickly.

Moreover, trust and authenticity will continue to matter. As large corporations grow more faceless and bureaucratic, there’s room for smaller companies that connect with customers on a human level — offering not just products, but stories, values, and transparency.

Regulation may also play a role. As public sentiment swings against corporate overreach, there could be stronger antitrust efforts or incentives for decentralization, creating a more level playing field.

So yes, while the game may be harder, it won’t be over. The underdog narrative — a small company with a bold idea — will still have a place in the next fifty years. The key will be to find the cracks in the system, adapt fast, and build something people truly care about.

This is a follow-up post of the following

How LLM Progress Could Stall in a World Dominated by AI-Generated Content

In a more optimistic scenario, humans continue to post original thoughts, stories, insights, and opinions online. Blogs, research papers, social media posts, open forums, and personal websites all remain active with fresh human-generated content. In this case, large language models (LLMs) like GPT and others will have a steady stream of high-quality, real-world data to learn from. But this raises a new question:

What happens when AI becomes more “creative” than most humans?

LLMs With Endless Inspiration

When trained on an open, vibrant internet full of human creativity, LLMs can achieve astonishing breadth and depth. They can remix ideas across disciplines, generate original prose, design new concepts, and produce artwork or code that would take a human hours or days.

With access to such an expansive and ever-growing dataset, AI models can learn to imitate not just average human output—but the best of it. They can synthesize the writing style of a novelist, the logic of a philosopher, and the aesthetics of a designer, all at once. Given enough high-quality training material, models could push the boundaries of what we consider creative work.

The Paradox of Creativity at Scale

We often define creativity as the ability to connect disparate ideas in new ways. Ironically, this is exactly what LLMs excel at—statistically recombining existing ideas into novel outputs. While they don’t have “intent” or “emotion” the way humans do, they can generate content that feels emotionally resonant, intellectually fresh, and artistically valuable.

As a result, many human creators may find themselves outpaced by AI in terms of output volume, quality, or originality—at least by conventional standards. We could see a flood of competent fiction, music, blog posts, ad copy, UX designs, and even business strategies, all machine-generated and indistinguishable from expert human work.

What Happens to Human Expression?

If AI becomes the default producer of online content, there’s a risk that human voices will become quieter—not because they disappear, but because they compete with a tidal wave of fluent, optimized AI content. Readers, users, and audiences might gravitate toward AI content simply because it’s faster, cheaper, or more tailored to their preferences.

But this doesn’t mean human creativity becomes obsolete. On the contrary, the value of human authenticity, unpredictability, and imperfection may rise. People may start to seek out content that feels human—raw, personal, idiosyncratic. Platforms might begin to prioritize or even verify “human-made” content, just like we now prioritize organic or artisanal products in a world of mass production.

A Future of Collaboration

Perhaps the most realistic future isn’t about AI replacing human creativity, but amplifying it. Just as Photoshop didn’t replace artists but gave them new tools, LLMs might empower more people to express themselves, refine their ideas, and build on the shoulders of giants.

In this scenario, the internet becomes a living collaboration between human and machine intelligence. Creativity doesn’t vanish—it multiplies.

The key will be not in resisting the technology, but in learning how to partner with it—consciously, ethically, and with intention.

As large language models (LLMs) become increasingly central to digital tools, education, productivity, and entertainment, it’s important to reflect on what fuels their progress: high-quality, human-generated data. But what happens if the internet becomes dominated by AI-generated content, while original human-written content becomes scarce or locked away?

Let’s explore the implications.

The Data Diet Problem

LLMs are trained on vast datasets—books, articles, websites, forums, and other publicly available text. These sources provide not just information, but human context: tone, nuance, humor, contradiction, reasoning. If we flood the internet with AI-generated content, especially content that was itself trained on earlier AI content, we risk creating a feedback loop—what some researchers call model collapse.

In this loop, models are trained on the output of previous models, and with each generation, the subtlety and originality of the language degrade. The richness of human expression slowly disappears, replaced by increasingly predictable and derivative phrasing. It’s like making photocopies of photocopies: eventually, the image becomes unrecognizable.

Restricted Access to Human Knowledge

At the same time, if high-quality human-created content is made inaccessible—either due to content owners blocking crawlers or refusing to license material—LLMs will be cut off from the very substance that makes them useful. Publishers, academic institutions, journalists, and independent creators may choose to monetize or protect their content, rather than give it freely to AI companies.

This scenario is already happening. The New York Times and other major publishers have taken steps to prevent their content from being used for AI training. Some are even pursuing legal action. From a business perspective, it makes sense. But from an AI training perspective, it’s a serious bottleneck.

The Decline of General-Purpose Intelligence

Without access to new, high-quality human-authored material, LLMs may plateau in their capabilities. Sure, companies can fine-tune existing models, optimize inference, or add retrieval systems to supplement knowledge. But the base models—the ones that power everything else—would eventually stagnate in their understanding of emerging trends, new ideas, and evolving cultural norms.

Ironically, the more we rely on LLMs to produce content, the more we risk starving the next generation of models of the very inputs they need to improve.

What Can Be Done?

This isn’t a doom scenario, but it is a real concern. Some potential solutions include:

• Incentivizing human content creation: Platforms could reward human-authored content or label it clearly to preserve its value.
• Developing curated training datasets: Companies may invest more in licensing and curating high-quality datasets instead of scraping the open web indiscriminately.
• Transparency in AI training: Users and creators may demand clarity on what data was used, pushing companies to be more ethical and collaborative.

The future of LLMs depends not just on better algorithms, but on the availability of diverse, authentic human knowledge. If we forget that, we may find ourselves in a world where AI sounds fluent—but no longer makes sense.

If you were into music and fashion in 1990s Tokyo, especially around the Shibuya district, chances are you encountered the eclectic, sophisticated sounds of Shibuya-kei. It was more than a music genre — it was a cultural moment. Drawing inspiration from French pop, bossa nova, jazz, lounge music, and even 1960s film scores, Shibuya-kei blurred the lines between kitsch and cool. It wasn’t just Japanese pop — it was cosmopolitan, ironic, and self-aware.

Artists like Pizzicato Five, Cornelius, and Kahimi Karie led the charge, delivering albums that felt like carefully curated mixtapes for a generation raised on vinyl, café culture, and imported fashion. The genre spoke to an emerging demographic in Japan: globally minded, design-conscious young people who embraced retro aesthetics and cutting-edge sensibilities at the same time.

Fast forward to today, and we’re seeing growing signs that Shibuya-kei might be due for a revival — much like how City Pop experienced a global rediscovery thanks to YouTube algorithms, vaporwave aesthetics, and a nostalgic hunger for analog warmth.

The resurgence of vinyl records, lo-fi visual styles, and retro-futuristic design has already reignited interest in genres once considered niche. Music collectors, sample diggers, and even TikTok creators are unearthing hidden gems from Japan’s past, and Shibuya-kei has all the right ingredients to resonate: timeless melodies, international influences, quirky production, and a unique charm that feels both vintage and fresh.

Personally, I think we’re just one viral track or Netflix placement away from seeing Shibuya-kei enter the cultural conversation again. If and when that happens, I’ll be ready to dust off my old Pizzicato Five CDs and welcome the revival of one of Japan’s most stylish musical exports.

Even if you’re not familiar with ice hockey, you’ve probably heard the name Wayne Gretzky. And if you haven’t, here’s why you should: Gretzky isn’t just the best hockey player in history—he’s one of the greatest athletes of all time in any sport.

Nicknamed The Great One, Wayne Gretzky dominated the NHL (National Hockey League) from the late 1970s through the 1990s. But what made him so special?

First, the numbers are staggering. He scored 894 goals and made 1,963 assists in his career, totaling 2,857 points. No one else has even come close. In fact, even if he had never scored a single goal, his assist total alone would still make him the highest point scorer in NHL history.

But Gretzky wasn’t just a stats machine—he changed how the game was played. He had a unique ability to read the game several steps ahead, like a chess grandmaster on skates. He wasn’t the biggest or the fastest player, but his anticipation, vision, and decision-making were on another level.

He also made hockey more popular across North America. In the 1980s, when he played for the Edmonton Oilers, he helped turn the team into a dynasty. Later, when he joined the Los Angeles Kings, he played a key role in growing hockey’s fan base in the U.S., especially in non-traditional markets like California.

Gretzky’s influence goes far beyond the ice rink. He became a global ambassador for the sport, known for his humility, sportsmanship, and leadership. Ask any serious hockey fan who the GOAT (Greatest of All Time) is, and you’ll almost always hear the same answer: Wayne Gretzky.

Even if you’ve never watched a game of ice hockey, Wayne Gretzky’s legacy is worth knowing. He didn’t just play the game—he redefined it.

Few debates in football ignite as much passion as the comparison between Diego Maradona and Lionel Messi. Both are Argentine icons, both are geniuses with the ball, and both have left an indelible mark on the history of the sport. Yet, they represent two very different eras, styles, and personalities.

Maradona was the ultimate street footballer—a force of nature with raw talent, explosive dribbling, and a rebellious edge. He carried Argentina to World Cup glory in 1986 almost single-handedly, and his performance against England—both the “Hand of God” and the mesmerizing solo goal—is etched into football folklore. Off the pitch, he was chaotic and controversial, but that only added to his legend. Maradona was human and mythic at the same time.

Messi, by contrast, is the epitome of consistency and technical perfection. His vision, passing, dribbling, and goal-scoring abilities are unmatched in the modern era. Messi dominated club football with FC Barcelona, winning multiple Champions League and La Liga titles. And for years, critics said he needed a World Cup to match Maradona’s legacy—until he delivered in 2022, leading Argentina to a historic win in Qatar.

If Maradona was the fiery poet of the game, Messi is its mathematician. One danced with chaos; the other mastered precision.

Who is greater? That often depends on what you value more—emotion or efficiency, drama or dominance. But maybe the better question is: why choose? Argentina gave us two footballing masterpieces. Let’s just be grateful we’ve witnessed them both.

When designing search functionality for modern applications, we often face a key decision: should we use a full-text search engine like Elasticsearch, or opt for a vector database such as Pinecone, Weaviate, or Qdrant? Both approaches are used to search text—but the underlying mechanisms, use cases, and strengths differ significantly.

Full-Text Search: Keyword-Based Precision

Full-text search engines like Elasticsearch or OpenSearch are based on traditional information retrieval methods. They use techniques such as tokenization, stemming, stop-word removal, and inverted indexes. Queries are often matched based on exact words or phrases.

Advantages of full-text search:

• High precision for keyword-based queries (e.g., “database migration tool for Laravel”).
• Mature ecosystem, with built-in features like filters, aggregations, ranking algorithms (BM25), and autocomplete.
• Fast and scalable for structured and semi-structured data.
• Excellent tooling and documentation.

Limitations:

• Struggles with semantic understanding—synonyms, paraphrased queries, and contextual meaning may not match well.
• Requires manual tuning for synonyms, fuzzy matching, or language-specific rules.

Vector Search: Semantic Understanding

Vector search is based on embeddings—high-dimensional numeric representations of text generated using machine learning models (e.g., OpenAI, Cohere, or Sentence-BERT). Instead of matching keywords, vector search finds content that is semantically similar to the query.

Advantages of vector search:

• Context-aware search: understands meaning, even if exact words are different.
• Excellent for natural language queries, such as “how do I make my database faster?”—even if those exact words aren’t in the documents.
• Ideal for question-answering, chatbots, recommendation systems, and AI-powered search interfaces.

Limitations:

• Requires an embedding model and infrastructure to generate and store vectors.
• Slower performance than full-text search for large-scale datasets unless optimized.
• Less mature filtering and ranking capabilities compared to Elasticsearch.
• Harder to explain search results to users (“why was this matched?”).

Use Case Comparison

Use Case	Full-Text Search	Vector Search
Product catalogs (with filters)	✅ Strong	⚠️ Less ideal
FAQ or knowledge base search	⚠️ Moderate	✅ Excellent
Semantic document search	❌ Weak	✅ Strong
Codebase or log search	✅ Very strong	❌ Not suitable
Hybrid AI search with context	⚠️ Requires effort	✅ Designed for this

Hybrid Approaches

Many modern systems use both full-text and vector search in tandem. For example:

• Use full-text search for filters, sorting, and metadata.
• Use vector search for semantic relevance.
• Combine scores to rank results more intelligently.

Elasticsearch now supports vector search natively, and hybrid ranking strategies are becoming more common.

Final Thoughts

If your application relies on exact keyword matching, filters, or structured search—full-text search is still the best fit. But if you’re building AI-enabled search, semantic discovery, or conversational interfaces, vector search offers a new level of intelligence.

In many real-world scenarios, the most effective solution is a hybrid approach that combines the speed and maturity of full-text search with the intelligence of vector search.