The AI Resistance Score: A 4-Part Framework to Audit Your Job's Layoff Risk in 2026

Most "AI-proof jobs" lists are useless the moment you try to apply them to your actual career. They tell you nurses are safe and copywriters aren't, then leave you to guess where a "senior product marketing manager who also does light design work" falls. Job titles are too broad to carry the signal you actually need.

What you need instead is a way to score the tasks inside your role — because that's the level at which AI substitution is actually happening. Research groups tracking AI displacement have converged on a four-dimension framework that does exactly this, and it's specific enough that you can run your own job through it in about ten minutes. This article walks through the framework, shows you the data behind it, and gives you a step-by-step way to calculate your own score and act on it.

Why Job Titles Are the Wrong Unit of Analysis

Two people with the identical title "financial analyst" can have wildly different AI exposure. One spends 80% of their week building recurring reports and reconciling spreadsheets — high automation risk. The other spends 80% of their week in rooms with executives, making judgment calls on deals with legal and reputational consequences — low risk. Same title, same pay band, opposite trajectory.

This is why sector-level and title-level "safe jobs" lists keep failing people. Goldman Sachs' occupational exposure data illustrates the gap starkly: office and administrative support tasks are roughly 46% automatable with current-generation AI, while construction tasks sit at about 6% and installation/repair tasks at about 4%. That's not a difference in industry glamour — it's a difference in what the actual hour-to-hour work requires.

The fix is to stop asking "is my industry safe?" and start asking "what percentage of my weekly hours go to defensible, non-automatable work?"

The 4-Dimension AI Resistance Score

Score your role from 0–25 on each dimension, for a total out of 100. Be honest about hours, not job description bullet points.

1. Physical Presence (0–25)

Does the work require physically manipulating unpredictable, variable real-world environments — a different pipe configuration in every house, a different patient's vein, a different construction site?

• High score (20–25): Electricians, plumbers, surgeons, physical therapists, emergency responders
• Mid score (10–19): Retail management, hands-on manufacturing supervision, field sales
• Low score (0–9): Anything performed entirely at a keyboard

Skilled trades score roughly 91/100 on composite AI-resistance frameworks, driven almost entirely by this dimension — every job site is physically unique, which makes robotic and software substitution impractical at current cost.

2. Human Relationship (0–25)

Does the value you create depend on a trust relationship, not just information transfer? Therapy, negotiation, caregiving, and high-stakes sales all score high here — the relationship is the product, not a wrapper around it.

• High score (20–25): Therapists, nurse practitioners, executive coaches, account executives on relationship-driven enterprise deals
• Mid score (10–19): Customer support with some relationship continuity, teaching
• Low score (0–9): Transactional, one-off interactions with no relationship continuity

Mental health counseling and therapy consistently top independent AI-resistance research at 95–97 out of 100, almost entirely on the strength of this dimension — the therapeutic relationship itself is the mechanism of healing, not an efficient way to deliver information.

3. Creative Judgment (0–25)

Does the work require originating something new under ambiguous constraints, versus recombining known patterns? Generative AI is good at the second and still weak at the first when the stakes and context are genuinely novel.

• High score (20–25): Creative direction with cultural/brand judgment, strategic consulting on unprecedented situations, original research
• Mid score (10–19): Design execution within an established system, standard marketing copy
• Low score (0–9): Formulaic content production, templated deliverables

4. Ethical Accountability (0–25)

Does someone need to be legally or professionally liable for the decision? AI can draft a diagnosis, a legal opinion, or a financial recommendation — but licensing boards, courts, and regulators still require a human to sign their name to it and bear the consequences.

• High score (20–25): Physicians, attorneys, licensed financial advisors, safety engineers
• Mid score (10–19): Compliance-adjacent roles without final sign-off authority
• Low score (0–9): No accountability structure attached to the output

What Your Total Score Means

Score	Interpretation	What to Do
75–100	Structurally protected	Keep building depth; AI is a tool, not a threat, in this role
50–74	Mixed exposure	Identify and shed the low-scoring third of your tasks; double down on the rest
25–49	High exposure	Start a 90-day pivot plan now, while you still have leverage and a paycheck
0–24	Severe exposure	Treat this as urgent — build a transition plan in parallel with your current job

Run this exercise honestly and most people land in the 50–74 band: not doomed, but carrying real exposure in a third or more of their weekly hours. That's actionable information — it tells you exactly which tasks to delegate to AI tools yourself (so you're the one capturing the productivity gain) and which to protect and deepen.

How to Raise Your Score Without Changing Jobs

You don't need to become an electrician to improve your number. Most people can meaningfully shift their score within their current role in 2–3 quarters:

1. Audit your calendar for one week. Tag every meeting and task block against the four dimensions. Most people are surprised how much of their week is dimension-0 work — status updates, report formatting, routine scheduling.
2. Volunteer for the highest-accountability work available to you. Sign-off authority, client-facing judgment calls, and anything with legal or financial consequence raises your Ethical Accountability score and is hard to backfill with a tool.
3. Move toward relationship-continuity work. If your role touches customers or stakeholders, push toward the accounts and relationships that compound over time rather than one-off transactional touches.
4. Get certified or licensed where a credential exists. Licensing requirements are one of the few durable moats against AI substitution because they create a legal accountability floor that software can't cross.
5. Use AI tools yourself, visibly. Workers with strong AI fluency earn roughly 56% more than peers in the same role without it — the fastest way to raise your relative resistance score isn't avoiding AI, it's being the person operating it.
6. Re-score quarterly. AI capability is moving fast enough that scores drift. A quarterly ten-minute re-audit catches erosion before it shows up in a layoff list.

A Worked Example

Take a mid-level marketing manager at a SaaS company. Their weekly hours break down roughly as: 40% writing and editing campaign copy and reports, 25% coordinating with agencies and internal stakeholders, 20% analyzing campaign performance data, and 15% presenting strategy recommendations to leadership.

Scoring this role honestly:

• Physical Presence: 2/25. Entirely screen-based work.
• Human Relationship: 12/25. Stakeholder coordination has some relationship continuity, but most of it is transactional status-syncing rather than trust-building.
• Creative Judgment: 14/25. Campaign strategy involves real judgment, but a large share of the copy and reporting work is pattern-recombination that generative tools already handle well.
• Ethical Accountability: 6/25. No licensing, legal sign-off, or safety liability attached to the output.

Total: 34/100 — high exposure. That's a sobering number for someone who might assume "marketing manager" sounds safely mid-career. But it's also actionable: the highest-leverage move for this person isn't a career change, it's shifting hours away from copy production (which AI tools now do faster than a human first draft) and toward the strategy-presentation and stakeholder-trust work that scores higher — while using AI tools themselves to produce the first-draft copy in a fraction of the time. Do that consistently for two quarters and the same person's score can realistically move into the high-40s to low-50s without a title change.

This is the point of running the numbers instead of trusting a generic "is marketing AI-proof?" headline: the honest answer is "some of it, and here's exactly which parts."

Common Mistakes When Self-Scoring

• Scoring your job description, not your actual week. Titles inflate; calendars don't lie.
• Ignoring the trend line. A 55 today that's trending down 5 points a quarter is a different risk profile than a stable 55.
• Treating "creative" work as automatically safe. Templated content production scores low on Creative Judgment even if the job title says "creative." The dimension is about originality under ambiguity, not the industry label.
• Assuming credentials alone protect you. Ethical Accountability only scores high if you actually hold sign-off authority — a licensed professional in a support role without that authority doesn't get the full credit.

Key Takeaways

• Score your role, not your title, across four dimensions: Physical Presence, Human Relationship, Creative Judgment, and Ethical Accountability.
• Skilled trades (~91/100) and mental health/therapeutic roles (~95-97/100) top independent AI-resistance research because they max out on physical unpredictability and relationship trust, respectively.
• Office and administrative tasks are roughly 46% automatable today versus single-digit percentages for skilled trades — the exposure gap is real and measurable.
• You can raise your score inside your current job in 2-3 quarters by shifting toward accountability, relationship continuity, and licensed work — and by becoming visibly AI-fluent yourself.
• Re-score every quarter. This isn't a one-time diagnosis; AI capability is moving fast enough that yesterday's safe task mix can shift.

Next Steps

Run your own numbers with LayoffReady's assessment — it factors in your role, industry, and skill mix to generate a personalized risk score and a concrete roadmap, not just a generic checklist. Take the free assessment and get your risk breakdown in under 10 minutes.