SkillShiftShape the shift.

Mar 2026

5 skills AI is quietly eating. 5 it still can't touch.

We cut through the noise so you don't have to.

In this issue

Your workforce and talent bets are probably going to the wrong places. Here's what's actually shifting.

  • 5 skills AI is already doing (sorry)
  • 5 human skills AI still can't touch
  • A table you can forward to your exec team
  • 3 articles worth reading
  • One thing to try this week
AI side

AI is already doing this. Sorry.

  • 1Graphic design (commodity work): Branded collateral, social templates, marketing materials. Design strategy and UX leadership stay in demand; execution-only roles are shrinking. You saw this coming.
  • 2Legal research and document review: Case summaries, citations, predictive insights. All embedded in lawyer workflows now. Complex litigation strategy and client negotiation remain high-demand.
  • 3Data entry and administrative work: RPA plus AI handles extraction at 95%+ accuracy. Automation is cheaper than hiring. Yeah.
  • 4Basic content writing and copywriting: Email copy, social posts, ad headlines at production quality. Commodity content declining; brand voice and nuanced persuasion stable or growing.
  • 5Translation (commodity language pairs): Near-human accuracy on common pairs. Specialized (legal, medical) still human-verified; commodity translation is mostly AI-native.

What it means for strategy: Don't hire for pure execution in these areas. Redesign roles so people do the judgment work and AI does the routine. Shift your people and talent investment toward the human side, not defending the AI side.

Human side

5 skills AI still can't touch

  • 1Emotional intelligence and negotiation: Reading unstated emotions, timing, and relational trust in real time. AI can prepare the brief; only humans build the relationship.
  • 2Strategic decision-making under uncertainty: Ethical judgment, risk appetite, and responsibility for consequences. AI can model scenarios; it cannot decide what matters most.
  • 3Creative problem-solving and innovation: AI excels at remix. True innovation is "what does not exist yet." That requires thinking outside the training data.
  • 4Leadership, motivation, and team building: Making people believe in the decision and want to execute it. Motivation and psychological safety are human-to-human transfers.
  • 5Client relationship management and adaptive influence: Understanding what someone truly needs, adapting to their style, and building trust over time. AI can personalize; it cannot hold someone through a difficult quarter.

What it means for strategy: Hire and develop for these. They're the bottleneck as AI takes routine work. Your talent and role design should prioritise these first.

A table you can forward to your team

If your exec team is still debating whether AI will replace jobs, forward them this. Then cancel the meeting. We map "what people do" next to "what software can now do" and highlight the human work that becomes more valuable next. Read it from both ends.

Where work is shifting
automation pressure + adjacent human value

How to read this

Green·Blue

Human skill employers hire for → AI capability that's automating it.

% bar

How much of this work AI can already do today (benchmarks + job-posting overlap).

AI catching up / Holding steady / Human advantage

Is AI gaining fast, holding steady, or are humans pulling ahead?

~Xmo

Expert estimate of months until AI reaches human-level (WEF, Anthropic, benchmark velocity).

Opportunity

The human work that becomes more valuable as AI takes the routine stuff.

Data: Real benchmarks (LMSYS, SWE-bench, GPQA) + curated estimates from WEF Future of Jobs, Anthropic labour market study. We're building toward fully live data.

Routine Data ProcessingComputer Operation
Holding steady~3mo
x
88%

Opportunity

Data Quality Governance & Exception Handling

Basic Cognitive & ClericalKnowledge & QA
Holding steady~6mo
x
82%

Opportunity

Judgement-Heavy Decision Making

LLM Development & ApplicationCode Generation
Holding steady~12mo
x
71%

Opportunity

AI Systems Architecture & Orchestration

Process OptimisationResearch & Synthesis
Holding steady~18mo
x
58%

Opportunity

Cross-Functional Change Design

Risk AssessmentScientific Reasoning
Holding steady~24mo
x
45%

Opportunity

Ambiguity Navigation & Bet Framing

Go-to-Market StrategyContent Writing
Holding steadyNo ETA
x
35%

Opportunity

Market Intuition & Relationship Strategy

Analytical ThinkingMathematical Reasoning
Holding steadyNo ETA
x
42%

Opportunity

Problem Framing & Stakeholder Translation

Stakeholder ManagementCustomer Support
Holding steadyNo ETA
x
25%

Opportunity

Trust Architecture & Coalition Building

Empathy & Social InfluenceEmpathetic Counselling
Holding steadyNo ETA
x
12%

Opportunity

Deepening — Human-Only Advantage

Creative & Innovative ThinkingImage Generation
Holding steadyNo ETA
x
38%

Opportunity

Creative Direction & Taste Curation

3 articles worth reading

One thing to try this week

Pick one role you hire for often. In a 20-minute call with your hiring manager, list the top 5 tasks that role does. For each task, ask: "Could a capable person do this in 10 minutes with an AI tool today?" Note where the answer is yes. That list is your starting point for where to invest in people versus where to automate.

The bottom line

Stop defending work AI can do. Double down on work it can't. Your workforce and talent strategy should shift hiring and role design toward the human side, and restructure so people do judgment and AI does routine.

Next week

The 3 skills employers are paying 40% more for in 2026.

← Back to SkillRecon