5 Skills AI Can’t Replace — We Agree With 3 of Them

New research from Cangrade just dropped: after scraping over 200 real AI job postings from Indeed across industries, roles, and seniority levels, they found that 83% of those postings included at least three of the same five soft skills. The implication is clear—as AI adoption accelerates, employers are doubling down on specifically human capabilities.

The five skills: strategic thinking, critical thinking, collaboration, communication, and adaptability.

We took each one and put it through our own filter—not just theory, but what we're actually seeing from hands-on work with the latest models, our coaching clients, and the businesses we consult with. We agree with three. The other two? Not anymore.

The Ones We Disagree With

Strategic thinking was the first skill on the list, defined as setting direction and seeing the big picture. A year ago, this would've been a clear win for humans. But the latest models have crossed a threshold here.

As Sean Boyce explains: “As of the latest releases, there's real strategic thinking capability now. It has second-order thinking where it can start asking questions about what comes next. We've posed questions to AI during our own business planning and it's come back with follow-up questions, additional research suggestions, even prompts we could use. Strategic thinking is something AI can do already—and it's only going to improve this year.”

That doesn't mean humans are irrelevant to strategy. But it's now a hybrid skill at best—humans leveraging AI to think bigger and faster, not a category where humans have exclusive advantage.

Collaboration was another we pushed back on. The research defined it as “can't connect across teams.” But with Anthropic's Agent Teams now live, AI agents can coordinate with each other, weigh competing preferences, and synthesize research into balanced decisions. The technology is already there—it's just that most companies haven't adopted it yet. The human advantage here isn't the skill itself; it's navigating the politics, the layers, and the organizational dynamics that AI still can't read.

The Ones We Agree With

Critical thinking—the ability to judge whether the output is right—is still firmly in human territory. Even with Opus 4.6 and GPT 5.3, James Perkins sees this daily: “Even the current models still struggle. If you give it really well-defined bounds, it'll go do the work. But when it comes time for QA or validation testing, it doesn't know exactly what to look for and it can't always determine the intent of the ask.”

Communication—specifically translating nuance—got unanimous agreement from us. AI still defaults to the lowest common denominator when given ambiguous requests. It's been programmed to appease rather than push back, which limits its ability to handle the subtlety that real communication demands.

Adaptability—handling ambiguity—rounds out the list. When prompts aren't explicit and crystal clear, AI still struggles to deliver what you actually need. Humans remain far better at operating in gray areas.

AI-Free Assessments Are Coming

Here's what caught our attention beyond the five skills: Gartner predicts that 75% of hiring processes will include AI proficiency screening by next year, while over 50% will include AI-free skills assessments. Companies are going to test what you can do without any tools—raw knowledge, raw judgment.

James puts it bluntly: “If you don't have AI skills and you're not willing to learn them, you are actively being removed from the workforce. The folks that have those skills are earning 60% more—but on the backs of the folks who are being removed entirely because they don't have those skills at all.”

The Compute Cost Wild Card

One insight James raised that most people aren't thinking about: every AI request fires up massive compute infrastructure. Right now, that's subsidized by a collective trillion dollars of investment. But that won't persist forever. As costs rise, there will be a push back toward human skills for tasks that don't justify the compute expense. The human brain runs on about five watts of energy and three meals a day—that's a cost advantage AI can't match for many routine tasks.

What This Means for You

The playbook is clear: master both sides. Build your AI skills so you can leverage the technology effectively, but invest deeply in the three human skills that still differentiate you—critical thinking, communication, and adaptability. And don't neglect demonstrating these in interviews. Resumes can't capture soft skills. Storytelling, case studies, and real demonstrations will matter more than ever.

Get Help Navigating This

This is exactly what we do at What About AI—whether it's through our daily podcast and newsletter (free), our private coaching programs, or our B2B consulting work.

Free Resources:

• AI-Proof Skills Guide → whataboutai.com/guides/ai-proof-skills
• Daily newsletter with AI workforce updates → whataboutai.com/subscribe
• Podcast episodes with actionable strategies → whataboutai.com/podcast

Coaching: For personalized 1-on-1 help, check out our coaching services at whataboutai.com/coaching.

AI Consulting for Your Business: whataboutai.com/business

Sources