Remember when everyone was convinced 2025 would be the year AI finally took over? Yeah, about that…
OpenAI’s Sam Altman basically promised us robot coworkers by now. He said AI agents would “materially change the output of companies” and handle all those entry-level jobs that humans find tedious. Spoiler alert: it’s not going great.
Take Klarna, the “buy now, pay later” company that went all-in on AI agents. They figured they could replace 700 human customer service reps with smart software. What could go wrong?
Everything, apparently.
Turns out even Google’s best AI agent (Gemini 2.5 Pro) fails at basic office tasks 70% of the time. OpenAI’s GPT-4? It bombs 91% of the time. Amazon’s AI? Nearly 100% failure rate. These aren’t rounding errors – they’re reality checks.
Klarna learned this the hard way. After firing their human agents, customers started flooding complaint lines about unanswered queries and unauthorized subscription changes. The AI wasn’t just bad at its job – it was actively making things worse.
Plot twist: Klarna’s CEO did a complete 180 and started desperately rehiring the humans they’d axed. Turns out replacing people is easier than finding good people to replace them with.
Here’s the thing though – this “failure” is actually setting us up for something better.
The Gartner Hype Cycle (yes, that’s a real thing) shows AI agents sitting at “Peak of Inflated Expectations” – basically the danger zone where promises outrun reality. But other AI tech is moving toward actual usefulness.
The real AI revolution isn’t about replacement – it’s about amplification. AI is already crushing it in medical diagnostics, catching cancers doctors might miss. It’s powering smartwatches that can translate languages in real-time. It’s helping teachers personalize learning and artists explore new creative territories.
This isn’t sci-fi stuff – it’s happening now, just not in the flashy “robots steal all jobs” way everyone expected.
For investors, this reality check is golden. The hype is cooling off, which means we can focus on companies actually building useful AI tools instead of chasing unicorns.
The big winners? Still the usual suspects – Microsoft (MSFT), Google (GOOGL), and Meta (META) – because they have the resources to weather the growing pains. But the real opportunities are in the supply chain: the chip makers, data center builders, and infrastructure companies powering this whole thing.
Bottom line: AI isn’t failing, it’s just growing up. And that awkward teenage phase? That’s exactly when smart money makes its move.
The creative trader wins, as they say. Just maybe don’t bet on AI replacing your job anytime soon.