AI Practicalist

  • Disruption, Resistance, and Eventual Embrace: Lessons for AI from 25 Years of Tech Shifts

    Technologies rarely arrive in the world as quietly accepted friends. More often, they stumble in like awkward interlopers; feared, mocked, and sometimes even banned. Yet over time, many of these same technologies become so ingrained in our daily lives that we wonder how we ever managed without them.

    Artificial intelligence today is somewhere in this stormy middle act. It is disrupting industries, unsettling norms, and sparking heated debates about creativity, work, privacy, and ethics. To better understand where AI might be headed, it helps to look backward. The past 25 years offer plenty of examples of innovations that were once disruptive, even vilified, before becoming widely accepted. Each one shows how society’s early discomfort gave way to normalization, and how the AI debate today seemingly echoes those patterns.

    Here are six major disruptions of the last quarter-century and how they foreshadow what may come with artificial intelligence.

    1. Online Shopping: From “Who Would Buy Shoes Without Trying Them On?” to Amazon Prime Day

    In the mid-1990s and early 2000s, online shopping looked like a novelty. Critics scoffed at the idea: Who would give their credit card number to a website? Who would buy clothes without trying them on? Surely people would stick with brick-and-mortar retail forever.

    The early e-commerce experience didn’t help. Websites were clunky. Shipping was slow and expensive. Returns were often impossible. Scams were real and rampant. Mainstream press coverage carried a tone of skepticism, at best portraying online shopping as a niche convenience for tech geeks, at worst online shopping was a dangerous risk.

    Fast forward two decades, and e-commerce is not just normal but dominant. Amazon alone now captures nearly 40% of all U.S. e-commerce sales. Online shopping is such a default that consumers assume a store must have a website.  If it doesn’t, they often look elsewhere. Retail giants like Walmart and Target now compete aggressively in online sales, and events like Black Friday have migrated into the digital realm with phenomena like Cyber Monday and Amazon Prime Day.

    Parallels to AI today:

    • Just as shoppers once doubted the practicality and safety of online transactions, there are many questions around AI’s reliability and trustworthiness.
    • Over time, improved infrastructure (faster shipping, better fraud protections, easy returns) removed barriers. AI too looks to gain broader acceptance as reliability, basic understanding of LLMs, and user safeguards improve.
    • What began as a novelty (buying books online) became an everyday utility. Similarly, today’s AI “novelties” (image generators, chatbots) are already evolving into unseen infrastructure and tools we barely notice but rely on constantly: think zero-click searching and AI coding agents.

    2. Social Media: From Time-Wasting Teen Hobby to Global Communication Infrastructure

    When MySpace, Facebook, and later Twitter (now X) first arrived, they were largely dismissed as distractions for teenagers. News coverage in the early 2000s often portrayed social networks as fads or worse. Employers and schools fretted about “time-wasting” and tried to block access.

    Yet within a decade, social media transformed politics, news, advertising, and culture. Facebook became a primary source of news for millions. LinkedIn redefined professional networking. Instagram reshaped fashion and lifestyle industries. TikTok now drives music charts and global cultural trends.

    The transition was not smooth.  Scandals around privacy, mental health, and misinformation continue to plague the platforms, but the broader arc is clear: something once trivialized as “kids posting about lunch” became essential digital infrastructure.

    Parallels to AI today:

    • The early dismissal of social media as “just for fun” mirrors the current attitude toward AI tools like art generators or novelty chatbots.
    • Over time, platforms expanded into serious domains: business, politics, healthcare. AI is already on the same path and is moving from playful experiments into customer service, medical diagnostics, and research assistance.
    • Social media showed that cultural adoption can happen faster than expected. AI adoption may follow a similarly rapid trajectory, or even faster.

    3. Smartphones: From “Why Would I Need Email in My Pocket?” to Global Necessity

    When the first iPhone launched in 2007, not everyone was convinced. Some critics called it overpriced. Others wondered who would want a touchscreen without a physical keyboard. Business professionals stuck to their BlackBerries, or more often, neither.

    Even before the iPhone, early smartphones from Palm and Nokia were often seen as niche devices for executives or geeks, not the general public. The phrase “internet in your pocket” didn’t sound exciting, it sounded unnecessary.

    Today, smartphones are not just normal, they’re practically indispensable. In many parts of the world, and for many groups of people, they are the primary computing device. They have reorganized entire industries: music, photography, journalism, gaming, finance. Billions of people worldwide rely on them as their central hub for work, communication, commerce, and entertainment.

    Parallels to AI today:

    • Early doubts about the need for smartphones echo current skepticism around AI assistants: “Why would I need that?” Over time, killer apps emerge that make the utility obvious.
    • Smartphones changed not just technology but culture and behavior. AI, too, may alter how we search, write, create, and collaborate in ways that feel strange now but natural later.
    • Integration is key.  Just as smartphones succeeded when hardware, software, and app ecosystems came together, AI will likely succeed when models, tools, and workflows converge seamlessly.

    4. Cloud Computing: From “I Don’t Trust My Data on Someone Else’s Server” to Business Default

    In the 2000s, companies were deeply skeptical of the cloud. Why trust sensitive business data to Amazon, Microsoft, or Google? Early objections were fierce: security concerns, compliance risks, performance worries. For years, many enterprises clung to on-premises servers, convinced that outsourcing infrastructure was reckless and inefficient.

    But as cloud reliability improved, and as cost savings became undeniable, adoption surged. Today, cloud computing is the default architecture for startups and increasingly for enterprises. Entire business models (Netflix, Slack, Zoom) would not exist without it. AWS, Azure, and Google Cloud now generate hundreds of billions of dollars in revenue.

    Parallels to AI today:

    • Early cloud resistance mirrors today’s enterprise resistance to AI: fears about security, control, and reliability.
    • Just as cloud eventually proved not only safe but also more scalable and cost-effective, AI will likely prove not only acceptable but also superior in speed, efficiency, and adaptability for many tasks.
    • “Hybrid cloud” approaches helped cautious companies transition. AI may follow with “human-in-the-loop” workflows where AI augments, rather than replaces, humans.

    5. Ride-Sharing and the Gig Economy: From “Illegal Taxis” to Urban Transportation Norm

    When Uber and Lyft launched, backlash was immediate and intense. Taxi companies lobbied against them. Cities banned them. Critics warned of unsafe rides, exploitation of drivers, and unregulated chaos.

    Yet riders loved the convenience, lower prices, and ease of app-based ordering. Over time, regulators adapted, companies improved safety measures, and the services became entrenched. Today, ride-hailing is standard in most major cities worldwide, and the broader gig economy model has spread into food delivery, freelancing, and other industries.

    Parallels to AI today:

    • Ride-sharing disrupted not just a technology but a labor model, much as AI is disrupting white-collar work.
    • Resistance centered on legality, safety, and fairness. Over time, compromise and regulation are shaping acceptance. These are some of the very same issues AI faces.
    • Consumers’ appetite for convenience often outweighs resistance. AI tools that save time and money may win adoption regardless of controversy.

    6. Streaming Media: From “Who Would Watch a Movie on Their Computer?” to the Death of Cable

    In the early 2000s, the idea of watching full movies or shows online seemed absurd. Bandwidth was limited and screens were small. Blockbuster and cable dominated. Streaming was dismissed as clunky and inferior.

    Then came Netflix, Hulu, and YouTube. As internet speeds improved, streaming overcame technical barriers. Consumer habits shifted rapidly, especially with the arrival of smart TVs and mobile apps. By the 2010s, streaming wasn’t just mainstream, it was the dominant model. Traditional cable subscriptions plummeted. Blockbuster filed for bankruptcy. Today, “binge-watching” is a cultural norm.

    Parallels to AI today:

    • Early dismissal of streaming as low-quality echoes dismissal of AI outputs as “cheap” or “inferior.”  AI will have to overcome “slop” as streaming has overcome “buffering”.
    • Once convenience and quality improved, consumers embraced streaming en masse. AI tools may likely follow: today’s quirks will give way to tomorrow’s smooth, reliable experiences.
    • Streaming also raised new cultural concerns: attention spans, binge habits, and content overload. AI too will bring trade-offs, requiring new cultural norms.

    The Pattern: Disruption, Resistance, Acceptance

    Across these examples, a familiar arc emerges:

    1. Novelty and skepticism – New tech looks unnecessary, dangerous, or trivial.
    2. Backlash and resistance – Regulators, industries, and skeptics push back.
    3. Improvement and integration – Usability, infrastructure, and reliability improve.
    4. Adoption and normalization – Consumers and businesses quietly make it part of daily life.
    5. Transformation – Culture and industries reorganize around the new baseline.

    AI today is somewhere between stages 2 and 3. It has moved past pure novelty (people see real uses) but faces intense resistance: ethical debates, fears of job loss, concerns about bias and misinformation. As AI improves and integrates into workflows, it may follow the path of e-commerce, smartphones, cloud, and streaming, transforming not just industries but culture itself.

    Conclusion: Why AI Feels Different And Why It’s the Same

    Skeptics often insist that AI is uniquely dangerous, unlike past technologies. To be fair, AI does raise profound questions about authorship, labor, and human identity. But so did past disruptions. When photography emerged in the 19th century, artists decried it as the death of painting. When ATMs spread in the 1970s, bank tellers feared mass layoffs. When word processors arrived, people said writing would decline.

    The lesson is not that all fears are unfounded, there are always real costs, but that disruption followed by acceptance is a recurring rhythm of technological progress. The past 25 years are filled with examples of technologies that seemed threatening or silly at first, only to become indispensable.

    Artificial intelligence today may feel overwhelming, but history suggests a future where it is not exotic, not terrifying, but perhaps, normal. We may wake up one day and realize AI is as unremarkable, and as essential, as ordering shoes online, hailing a ride, or streaming a show before bed.

  • Faster Iteration Everywhere: The Catfish Effect in AI

    Introduction: The Pond Gets Crowded

    In business folklore, the catfish effect describes what happens when you throw a catfish into a tank of lethargic codfish. Instead of drifting in circles, the cod suddenly become lively. Why? Because the catfish forces them to move or be eaten. The metaphor has long been used in sports strategy and corporate leadership, but in the artificial intelligence world, it’s more than a story — it’s daily reality.

    AI is an industry driven by iteration: new models, new platforms, new frameworks, new features. What once took years to update now happens in weeks. This is not just marketing hype; it’s the direct result of competition. Every new tool that hits the scene acts as a catfish, stirring up incumbents and forcing them to improve faster.

    For solo entrepreneurs, small businesses, and curious practitioners, this acceleration is both thrilling and overwhelming. Thrilling because tools get better constantly. Overwhelming because there’s always something new. The challenge is not keeping up with every release, but knowing how to benefit from the churn.

    This article explores the catfish effect as applied to AI, with an emphasis on how faster iteration is reshaping the landscape. We’ll trace where the effect shows up, how it benefits you, and why it matters to think like a catfish yourself.

    Section 1: The Catfish Effect Explained

    The catfish effect thrives in competitive environments. In a pond with only one kind of fish, growth slows. Innovation plateaus. Costs remain high because there’s no pressure to lower them. But the moment a scrappy competitor enters, the rules change.

    In AI, this plays out across multiple fronts:

    • Model development: OpenAI releases GPT-3, but soon Anthropic launches Claude, Google ships Gemini, and Meta puts out Llama. Each one pressures the others to roll out improvements quickly.
    • Platforms and services: First movers like Jasper AI enjoyed success in AI writing, but new tools like Copy.ai, Sudowrite, and open-source competitors forced rapid feature development.
    • Pricing and accessibility: Without competitors like RunPod or Banana.dev, cloud compute costs for small builders would remain sky-high. Their catfish presence pulled prices down.

    The catfish effect is not just about rivalry; it’s about survival. In AI, survival means relevance. Companies cannot coast on a single release for long, because someone else will deliver something sharper, cheaper, or more open.

    Section 2: Historical Precedents of Iteration

    Fast iteration isn’t new. Technology has always had its catfish moments:

    • Browsers: In the 1990s, Netscape dominated until Internet Explorer chased it, sparking rapid updates. Later, Firefox and Chrome disrupted the scene, forcing new standards for speed and usability.
    • Smartphones: Apple’s iPhone disrupted the market, but Android’s fast iteration forced Apple to release features quicker than their old cycles.
    • Digital art tools: Adobe Photoshop was untouchable until Procreate, Affinity, and free tools like GIMP made them adapt. Subscription pricing, cloud collaboration, and AI-driven features all came faster than Adobe originally planned.

    AI is following this same pattern, only faster. The difference? Cloud infrastructure and global developer communities amplify the effect. Instead of updates every few years, we now see iteration every few weeks.

    Section 3: Where Faster Iteration Shows Up in AI

    1. Large Language Models (LLMs)

    GPT-3 stunned the world in 2020. It felt untouchable. But soon came open-source models like BLOOM, Llama, and Mistral. Claude focused on safety and reasoning. Gemini emphasized multimodal capabilities. Each “catfish” forced competitors to adapt. OpenAI released GPT-4 far sooner than expected, and the rumor mill about GPT-5 reflects the same pressure.

    The result: users now enjoy richer models with better reasoning, longer context windows, and more transparent benchmarks.

    2. Text-to-Image Models

    MidJourney built its reputation on beautiful, stylized images. But Stability AI’s Stable Diffusion offered open-source flexibility, and Civitai created community-driven ecosystems. The pressure nudged MidJourney to roll out new versions faster, open up some customizations, and improve usability. DALL-E, once stagnant, suddenly added inpainting and image editing to catch up.

    Iteration moved from yearly to quarterly cycles, directly benefiting artists, creators, and entrepreneurs.

    3. No-Code and Low-Code Builders

    Platforms like Bolt.new, Lovable.dev, and Replit AI are essentially catfish in the coding pond. They show what’s possible with AI-assisted development, pushing traditional software IDEs (integrated development environments) and workflow platforms to accelerate updates.

    Faster iteration means that solo founders can now go from idea to prototype in days, not months. This speed changes the scale of who gets to build.

    4. Compute and Infrastructure

    AWS, Azure, and Google Cloud had a lock on compute pricing. But scrappy providers like RunPod and Banana.dev entered, offering bare-metal GPU access at lower prices. These catfish forced the giants to rethink pricing tiers, introduce AI-specific infrastructure, and improve accessibility for smaller clients.

    This shift benefits everyone who needs horsepower to train or fine-tune models without a corporate budget.

    5. Specialized AI Tools

    From transcription to video editing, the pattern repeats. Otter.ai had early dominance, but competition from tools like Sonix, Fireflies, and Whisper forced constant iteration. In video, Descript’s features got sharper once competitors leaned into AI editing and auto-captioning.

    In every niche, catfish appear, and the pace of updates accelerates.

    Section 4: Why Faster Iteration Matters

    1. Lower Barriers to Entry

    When tools evolve rapidly, costs drop and features spread faster. You don’t need to wait for “enterprise-ready” rollouts. Early adopters get advanced capabilities immediately.

    2. More Choices

    Competition creates variety. You can pick the tool that best fits your workflow instead of being locked into one provider.

    3. Pressure on Incumbents

    Big players cannot coast. They’re forced to lower prices, improve accessibility, and add features they might otherwise delay for years.

    4. Cultural Shift Toward Experimentation

    Rapid iteration normalizes experimentation. In AI, this mindset is critical. You don’t need to wait for a perfect tool—you can build, test, and pivot quickly.

    Section 5: Use Cases for Entrepreneurs

    Use Case 1: Rapid Prototyping

    With AI builders updating constantly, entrepreneurs can test new product ideas at a fraction of the time and cost. A small consultancy can spin up a niche app in days, see if clients respond, and pivot without massive sunk costs.

    Use Case 2: Lean Marketing

    As AI writing tools iterate, marketers can experiment with campaigns, headlines, and outreach faster. Instead of waiting weeks for polished copy, drafts are generated and tested instantly.

    Use Case 3: Community Advantage

    Communities like Hugging Face give individuals access to bleeding-edge tools before corporations adopt them. By engaging with these catfish platforms, solo builders can stay ahead of enterprise adoption curves.

    Use Case 4: Competitive Positioning

    If you adopt new tools early, you can differentiate. A freelancer using AI-accelerated video editing today will outpace competitors who stick with traditional workflows.

    Use Case 5: Cost Arbitrage

    Catfish competition almost always drives prices down. Entrepreneurs who stay alert to these shifts can cut operating costs by switching platforms at the right moment.

    Section 6: Risks of Faster Iteration

    Iteration isn’t free of downsides. Moving too fast can break trust. Tools may ship half-baked features or change pricing without warning. For businesses, relying on tools that iterate too quickly can create instability.

    The catfish effect accelerates progress, but it also creates churn. That means entrepreneurs must balance curiosity with caution. Adopt new tools, but always keep backups and maintain flexibility.

    Section 7: Thinking Like a Catfish

    There’s another angle: you don’t just benefit from the catfish effect—you can embody it. Solo entrepreneurs can act as catfish in their own markets. By innovating quickly, niching deeply, or adopting new AI capabilities first, you create movement in your industry.

    Being a catfish doesn’t mean being reckless; it means being disruptive enough to force others to move. In AI-driven niches, this often looks like:

    • Offering AI-powered services your competitors haven’t tried yet.
    • Building custom tools with open-source models.
    • Using rapid iteration as a marketing edge—showing clients that you move faster.

    In short, don’t just swim with the codfish. Be the catfish.

    Conclusion: The Pond Will Never Be Still Again

    The AI ecosystem is in constant churn because of the catfish effect. Faster iteration is not a phase; it’s the new normal. Every week brings a new release, update, or competitor that forces the rest of the field to accelerate.

    For solo entrepreneurs and small businesses, this is an invitation. Rather than fearing the pace, you can ride it. Test new tools. Compare platforms. Switch when the benefits outweigh the risks. Think like a catfish yourself, moving quickly enough to force others to adapt.

    The pond is crowded, the fish are restless, and the catfish are everywhere. That may sound chaotic, but for those willing to swim, it’s the most fertile pond we’ve ever had.