AI hasn’t changed anything in our society yet

The predictions that never landed

Remember all those dramatic headlines about AI taking over our jobs—or worse, humanity itself? Yeah, they were hard to miss. But here we are, and the robots haven’t replaced us yet. Sure, AI has found its way into some industries, but the sweeping disruption everyone was bracing for? It’s more of a drizzle than a storm.

Even Bill Gates recently jumped into the prediction game, claiming AI could bring us a two-day workweek by 2035. Sounds great, right? But let’s not forget Keynes said something similar 100 years ago about technology giving us 15-hour workweeks by now. Spoiler alert: we’re still grinding away at our desks five days a week (if not more).

The truth is, technological advancements—whether it was the computer back in the day or AI now—don’t seem to reduce how much we work. They just shift what we work on. More screens, more compliance, more digital “paperwork.” AI isn’t rewriting the rules of society; it’s just tweaking them slightly while we keep running on the same treadmill.

The LinkedIn hype machine

Let’s talk about where AI has had an impact: LinkedIn posts. You’ve seen them—the self-proclaimed thought leaders bragging about how they’ve “fully integrated AI into their workflows.” But when you dig deeper (or ask for specifics), there are rarely any concrete examples. Often, it turns out their big “integration” is just having ChatGPT open in another tab to help them write emails, brainstorm ideas, summarise meetings or plan communication campaigns. Groundbreaking stuff. Enough to be stamped “AI-powered” on the website.

AI has become more of a buzzword than a tool and it damaged its image. It’s trendy to say you’re using it, even if all you’re doing is asking it to rephrase sentences or come up with catchy LinkedIn captions. And while there’s nothing wrong with using AI for these tasks—it is helpful—it’s not exactly the game-changing innovation these influencers make it out to be. Worst, it can feel “scammy” to mention AI is integrated in a workflow, because we’ve heard it too many times.

Why AI adoption is moving at a snail’s pace

Let’s be honest: businesses aren’t exactly rushing to embrace AI with open arms. And for good reason—it’s expensive, complicated, and riddled with potential headaches. Most companies are taking a cautious “wait and see” approach, and honestly, who can blame them?

First off, integrating AI isn’t just a matter of flipping a switch. It often involves overhauling entire workflows, retraining employees, and navigating a minefield of data privacy concerns. Plus, with so many different AI tools and platforms out there (most of which are proprietary and lock you into their ecosystem), businesses are understandably hesitant to commit until it’s clear which technologies will actually stick around. Nobody wants to invest millions in an AI system only to find out it’s the Betamax of its time.

The paradox is that AI is evolving exponentially fast, it’s not worth to integrate it now since it could be outdated in 6 months.

This cautious adoption is why AI hasn’t made much of a dent in big-picture metrics like job growth or wages. Sure, it’s automating some tasks here and there—data entry, customer service chatbots—but these changes are mostly happening in niche areas. Take retail, for example: self-checkout systems have reduced the need for some cashier jobs, but this hasn’t exactly triggered mass unemployment across the economy. The impact has been more of a ripple than a wave.

And here’s the kicker: while AI might eliminate some jobs, it’s also creating new ones. Roles like AI trainers (yes, people who teach AI systems how to behave), human-AI collaboration specialists (fancy title for making sure humans and machines play nice), and ethics officers (because someone has to keep the robots from going rogue) are popping up to fill the gaps. These emerging positions are helping offset job losses caused by automation, keeping the labor market more balanced than you’d expect from all those doomsday predictions.

As for the economy? Don’t hold your breath for an AI-fueled boom anytime soon. Goldman Sachs is betting on productivity gains starting around 2027 and really kicking in during the 2030s. So far, though, most industries are still figuring out how to integrate AI without breaking everything else in their operations. It’s like trying to install a turbocharger on an old car—you might get more speed eventually, but not without some serious tinkering first.

What does it take to really integrate AI into a workflow?

Let’s get real: integrating AI into a workflow isn’t like slapping a “Powered by AI” sticker on your app. Most of the time, it requires rewriting entire codebases or building new software from scratch with AI at its core. And guess what? It’s expensive as hell.

Developing custom AI solutions can cost anywhere from $10,000 for basic chatbots to $10 million+ for enterprise-grade systems. Even mid-sized projects like predictive analytics or natural language processing tools run between $100k and $500k. And that’s before you factor in the dependency on third-party APIs—most AI tools aren’t open-source, so you’re stuck relying on companies like OpenAI or Anthropic to keep their servers running and APIs accessible. Imagine building your business on a tool that could vanish overnight if the provider hikes prices or goes bankrupt.

This is why big corporations like Adobe or Figma can roll out “AI-powered” features like “expand the background” or “auto-rename layers.” They’ve got the funds to throw at in-house teams or pricey outsourcing. Meanwhile, startups claiming to “revolutionise industries with AI” are often just repackaging ChatGPT integrations behind flashy UIs—and investors are eating it up.

Speaking of which: the AI bubble is real. Companies like OpenAI ($150B valuation) and Anthropic ($60B) are trading at valuations 30-50x their actual revenue. It’s like the dot-com crash all over again, where hype outpaces reality. The SEC has already fined firms like Delphia and Global Predictions for lying about their AI capabilities, and analysts warn that 28% of tech stocks are wildly overvalued heading into 2025. Most of these “AI startups” are selling vaporware, banking on FOMO to keep investors hooked until the music stops.

What AI actually does (so far)

Let’s give credit where it’s due: AI is useful. Tools like ChatGPT and others have made things like summarising text, rewriting drafts, and generating ideas faster and easier. And in 2024, we started seeing more software integrate basic AI features—think buttons that say “Enhance this text” or “Summarise this page.” Handy? Absolutely – the very text you’re reading now was made with the assistance of Perplexity to find sources and rewrite it into a well structured article. But is revolutionary? Not quite – it’s only completing a task faster, while still being operated and reviewed by a human.

The problem is that most people don’t really know what AI can do beyond chatbots – and it’s normal. Because the softwares we have access to now rarely do much than generating content and visuals for now. There’s this vague idea that AI is everywhere these days, but in reality, its integration into everyday tools is still in its infancy. For most of us, it’s just an assistant—a helpful one, sure—but not the life-altering force it’s been hyped up to be.

The recent release of ChatGPT 4o with visual capabilities marks a significant step towards more advanced AI applications in design. This iteration can now rework existing images by understanding and manipulating their style, offering a more intuitive tool for designers and opening the era of “vibe designing“. While it still requires human operation and prompt refinement, the improved comprehension of user intent makes it accessible to a broader audience. However, it’s important to note that the current capabilities are limited to generating static text and visuals. The system doesn’t yet offer interactive layered designs where individual elements can be edited through prompts, nor does it produce vector graphics suitable for high-resolution print work. Mostly because the feature is still integrated into a chatbot interface, but also because the technology for an extended use like describe above isn’t ready yet.

This advancement has sparked discussions about potential job displacement, particularly in fields like YouTube thumbnail design. However, it’s worth considering that many of these roles were already being outsourced to lower-cost markets through platforms like Upwork. While AI tools like ChatGPT 4o may indeed replace some of these jobs, it’s important to recognise that these positions often weren’t the highest-paid or most value-adding in the creative industry. The true impact of AI in design is likely to be in augmenting human creativity rather than wholesale replacement of designers. Allowing us to work faster and spend more time on things we love: actually designing and solving problems.

A useful tool, not a revolution (yet)

So where does that leave us? AI isn’t useless—it’s made some tasks faster and easier—but it also hasn’t lived up to the massive hype surrounding it. The predictions of job takeovers and two-day workweeks feel more like marketing stunts than realistic forecasts. And while LinkedIn influencers might want you to believe they’re riding the cutting edge of an AI revolution, most of them are just using ChatGPT to save time on emails.

For now, AI is best seen as what it actually is: a tool. A good one! But not a miracle worker or society-shifting force—at least not yet. So let’s keep using it where it makes sense and don’t re-adjust the whole prism of reality around it. AI is still at its very early stages, and I predict that tools like ChatGPT will soon become much pricier given the enormous resources it takes to run these models (despite their constant optimisations), making future versions of AI – the real dealbreaker ones – available to much fewer people, like professionals only.