AI Across Every Industry
AI Across Every Industry
AI disruption is not uniform across industries. The sectors most exposed are those with high volumes of routine knowledge work and well-documented processes. The sectors most resistant are those requiring physical dexterity, novel judgment, human relationships, and contextual adaptability that AI currently cannot replicate at the required reliability level.
Healthcare
What AI is automating now: Medical imaging analysis (radiology, pathology) is the most advanced application — AI systems match or exceed radiologist accuracy on specific image-reading tasks. Clinical documentation (nurses and doctors dictating notes, AI transcribing and structuring them) is widely deployed. Diagnostic support tools suggest differentials based on patient records.
What AI cannot yet reliably do: Full diagnostic responsibility (liability and reliability constraints), complex treatment planning requiring integration of patient context and values, patient communication and therapeutic relationships, surgical dexterity at human level (though surgical robots with AI assistance are advancing).
The net effect: Healthcare AI is reducing the administrative burden on clinicians — a primary driver of clinician burnout. It is not reducing the number of clinicians needed; population health demands are growing faster than AI efficiency gains.
Law
What AI is automating: Contract review and drafting, legal research (finding relevant precedents), document discovery in litigation, due diligence for M&A transactions, and basic legal FAQ response. Law firms use AI to process in hours what previously took paralegal teams weeks.
What AI cannot yet reliably do: Court appearances requiring advocacy and judgment, complex strategy in novel legal situations, client counseling requiring emotional intelligence, and anything requiring regulatory approval or professional licensing.
The disruption pattern: Junior associate roles that primarily involved document review and research are most threatened. Senior partner roles requiring relationship management, courtroom performance, and strategic judgment are less exposed — but the partner-to-associate ratio is shifting as AI handles more junior work.
Finance
What AI is automating: Algorithmic trading (already largely automated pre-LLM), financial report generation, fraud detection, credit scoring, customer service (chatbots handling routine banking queries), and investment research summarization.
What AI cannot reliably do: Complex credit underwriting for novel situations, relationship banking, M&A advisory requiring deep client relationships, and investment decisions in genuinely novel market regimes not represented in training data.
Notable impact: Hedge funds and quantitative trading firms have been using AI and machine learning for over a decade. The new wave of generative AI primarily affects the front-office knowledge work — analyst reports, presentations, client communications.
Education
What AI is automating: Content generation (lesson plans, explanatory content, practice problems), personalized tutoring for well-defined subjects (mathematics, coding, language learning), administrative tasks, and basic assessment.
What AI cannot reliably do: Mentoring, motivation, social-emotional learning, and the unpredictable human judgment required in a classroom of 30 diverse students.
The disruption: Traditional testing and essay-based assessment is fundamentally challenged — if AI can write competent essays on any topic, grading those essays tests who is better at prompting AI, not student learning. Education systems are grappling with assessment redesign.
Logistics and Operations
What AI is automating: Route optimization, demand forecasting, inventory management, warehouse picking (in combination with robotics), quality control vision systems.
What AI cannot yet do: The full physical dexterous work of loading, unloading, and handling non-standardized goods — and last-mile delivery in complex urban environments with unpredictable obstacles.
Marketing and Creative
What AI is already doing at scale: Copywriting, image generation, video generation, personalized content at scale, A/B test variant generation, social media content, and ad creative.
What remains human-intensive: Brand strategy, genuine creative direction, cultural intuition, and building authentic audience relationships. The creative output quality of AI tools is sufficient for most commercial applications; differentiation increasingly comes from human curation and strategy.
The Cross-Industry Pattern
The industries most disrupted are those with:
- High volumes of routine, document-based work
- Well-defined processes with clear inputs and outputs
- High labor costs creating strong economic incentive for automation
- Digital-native operations with structured data AI can train on
The industries most resistant are those requiring:
- Complex physical dexterity in unstructured environments
- Human relationships as the core value (therapy, care)
- High-stakes judgment in novel situations with legal liability
- Licensed professional accountability