AI's Human Problem

· The Fluency Briefing

Welcome back to your essential weekly

This Week in AI

Hey there — this week the AI world held up a mirror, and the reflection was... us. Researchers coined 'cognitive surrender' to describe our blind trust in AI, five top AI models disagreed on two-thirds of fact-checks, and Canva gave 5,000 employees a full week to learn AI only to discover the biggest obstacle was human habits. Meanwhile, graduates literally booed pro-AI commencement speakers. The machines aren't the problem. We are. Let's dig in.

Weekly Theme

📰 The Big Story

Here's the thing about the AI revolution nobody wants to say out loud: the technology isn't failing us — we're failing to think critically about it.

New research has put a name to something you've probably felt: "cognitive surrender" — the tendency to blindly accept AI outputs even when they're wrong whytryai.com, May 28. You ask ChatGPT a question, it answers with confident authority, and your brain just... stops questioning. It's not laziness. It's a cognitive shortcut baked into how we process confident-sounding information. And it's happening at scale.

The timing of this concept couldn't be sharper. A study testing five frontier LLMs on 1,000 real-world fact-checking claims found the models disagreed with each other on 67% of them lenz.io, May 28. Let that sink in: the AI tools we're surrendering our judgment to can't even agree with each other two-thirds of the time. The confidence-accuracy gap is real, and it's widening.

Then there's the cultural backlash. Graduates across the U.S. booed commencement speakers who praised AI, with students calling it tone-deaf given anxieties about their career prospects theguardian.com, May 26. This isn't a fringe reaction. MIT Technology Review's AI Hype Index captured the mood perfectly: telling the class of 2026 that AI will change the world doesn't land well when they're worried it'll change them out of a job technologyreview.com, May 28.

Why does this matter going forward? Because adoption isn't a technology problem anymore. It's a trust problem. And trust, once eroded, takes far longer to rebuild than any model takes to train.

Reaction

📋 5 Stories That Shaped the Week

Beyond the headlines, here's what shaped the week...

Canva ran a fascinating experiment: give 5,000 employees an entire week to do nothing but learn AI. The expectation was transformation. The reality? The biggest blockers weren't technical — they were human. Habits, resistance to change, and uncertainty about where to start outweighed any tool limitations fortune.com, May 28. This is the adoption story playing out everywhere, and Canva deserves credit for being honest about it.

While companies wrestled with adoption, the trust conversation took a surreal turn. The Vatican released Pope Leo XIV's encyclical on AI ethics — calling for AI to serve humanity and not concentrate power engadget.com, May 25 — but analysis suggests portions of the document were themselves AI-generated lesswrong.com, May 26. The irony writes itself. If the institution calling for AI transparency can't model it, who will?

On the security front, Anthropic's Claude Mythos Preview model found over 10,000 software vulnerabilities in its first month through Project Glasswing engadget.com, May 23. The AI bug-hunting arms race is accelerating fast, with attackers and defenders now locked in an AI-powered escalation wired.com, May 25. Here's the implication chain: better AI security tools mean faster patching. Faster patching means attackers need more sophisticated AI. Which means security costs go up for everyone.

Meanwhile, the economic reality check continued. Uber's president said AI spending is getting "harder to justify" theverge.com, May 26, echoing a veteran CEO's frustration that the hype cycle has detached from business value fortune.com, May 27. And firms across the UK are performing what PR executives call "yoga-level stretches" to rebrand basic automation as AI theguardian.com, May 24 — a sign the label matters more than the substance right now.

🔗 The Pattern We Noticed

Connecting the dots...

The thread running through this week? The gap between AI's projected confidence and actual reliability — and how that gap is infecting every layer of adoption, from individual cognition to corporate strategy to cultural trust.

Here's the implication chain. First order: AI outputs sound authoritative, so people accept them uncritically. Second order: organizations build processes on that uncritical acceptance, which means decisions get made on shaky foundations. Third order: when the cracks show — conflicting fact-checks, AI-written ethics documents, spending that's "hard to justify" — the backlash isn't proportional. It's amplified by the broken trust.

Why now? Because we've exited the honeymoon phase. The early adopters already integrated AI. Now the mainstream is arriving, and they're asking harder questions with less patience for hand-waving.

For you, this means one thing: the competitive advantage isn't using AI. It's using AI critically. The businesses that build verification habits, question outputs, and maintain human judgment in the loop will be the ones still standing when the dust settles.

Meme

🔮 On the Horizon

These stories are still unfolding — here's what to track:

📚 Term of the Week

Term illustration

Going deeper on one concept that shaped this week's AI conversation.

"Cognitive Surrender"

What it is: Cognitive surrender describes the human tendency to stop thinking critically when AI presents information confidently. Unlike simple automation bias, it's specifically about yielding your judgment — not just your tasks — to an AI system. The term captures how authoritative AI outputs can short-circuit the questioning instincts you'd normally apply to information from any other source.

Why it matters this week: New research formalized this concept just as studies showed AI models disagree with each other 67% of the time — making blind trust especially dangerous.

The bigger picture: As AI becomes embedded in more decisions — medical, legal, financial — cognitive surrender could compound errors at scale. The antidote isn't less AI usage, but deliberate verification habits that keep human judgment active.

Try this: Ask ChatGPT a factual question about a topic you know well. Before reading the answer, write down what you believe is correct. Compare the two and notice how you feel when they differ.

📬 That's a Wrap

That's a wrap on this week — and what a week for the mirror. The biggest AI story right now isn't about what machines can do. It's about what we stop doing when we hand them the reins.

Your move: Pick one AI-generated output you relied on this week — an email draft, a summary, a data analysis — and fact-check it manually. Notice what you find. That habit is worth more than any prompt template.

Fluently yours, The My AI Fluency Team

What We're Working On

✨ Founding Cohort Special - 60% Off! — Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now

✨ Free 30-Minute AI Consultation — Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call

✨ How AI-Fluent Are You? — Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz

💬 Community | 📞 Book a Consultation | 🌐 Website

My AI Fluency