AI in Industry

Applied AI across sectors: drug discovery, security, finance, productivity. See [[machine-learning]] for technical ML concepts and [[cybersecurity-thesis]] for the security equity thesis.

---

Novo Nordisk × OpenAI — Drug Discovery

Source: raw-sources/AI Projects/Novo Nordisk partners with OpenAI.md

The Partnership

Novo Nordisk (world's largest pharmaceutical company by market cap at peak) partnered with OpenAI to apply AI models to drug discovery and development processes.

Context: Novo Nordisk is best known for GLP-1 drugs (Ozempic, Wegovy). The AI push is about accelerating the next wave of molecules.

Build-on: Nvidia Gefion Supercomputer

The partnership extends an existing collaboration with Nvidia, which includes access to the Gefion supercomputer — a large-scale compute cluster designed for biomedical AI workloads.

Stack: Nvidia compute → OpenAI models → Novo Nordisk research workflows.

Competitive Context

Eli Lilly is Novo Nordisk's primary competitor in GLP-1 drugs (Mounjaro, Zepbound). Both companies are racing to discover next-generation metabolic disease treatments.

The AI-in-drug-discovery race is partly about speed to candidate — identifying promising molecules years faster than traditional methods.

Limitations

AI is not yet end-to-end in clinical trials. Current applications:

• Target identification (which proteins to drug)
• Molecule generation and screening
• Predicting binding affinity
• Clinical trial design optimization

Human judgment still required at: regulatory strategy, ethical review, clinical endpoint selection, patient stratification.

---

Applied AI — Broader Landscape

Source: raw-sources/AI Projects/List of Different ML Domains.md (Applied AI section)

High-Impact Industry Applications

Sector	AI Application
Healthcare	Drug discovery, diagnostic imaging, patient risk stratification
Finance	Fraud detection, algorithmic trading, credit scoring, FP&A automation
Logistics	Route optimization, demand forecasting, warehouse automation
Energy	Grid management, predictive maintenance, solar/wind output forecasting
Legal	Document review, contract analysis, case outcome prediction
Manufacturing	Quality control (computer vision), predictive maintenance
Retail	Inventory forecasting, personalization, dynamic pricing

Key Tension: Expert Task Automation

From [[economics-and-scarcity]]: AI is automating expert tasks (legal analysis, medical diagnosis, financial modeling) that were previously considered automation-resistant.

This creates structural unemployment risk in high-skill professions, while simultaneously creating demand for:

• AI deployment specialists
• Prompt engineers / AI workflow designers
• Domain experts who can validate and supervise AI outputs
• Cybersecurity (see [[cybersecurity-thesis]])

---

AI in Finance (Personal Application)

Hakyun's FP&A observation from raw-sources/AI Projects/Project ideas.md:

AI is entering FP&A (Financial Planning & Analysis) workflows, automating:

• Variance analysis
• Budget vs actual reporting
• Forecast model maintenance
• Board deck preparation

Implication for career: The finance roles that survive will emphasize judgment, communication, and strategic interpretation — not data manipulation. The modeling itself becomes a commodity.

Related: [[hakyun-ryu]] career direction | [[economics-and-scarcity]] structural change theory

---

AI as Infrastructure

From raw-sources/AI Projects/Research & ML/MD files - What they are.md:

John Lilly's 1975 prophecy (from [[philosophy#John C. Lilly]]) — that sufficiently complex computing systems would develop internal models and goals — is now live.

The design question is no longer "can AI do this?" — it's "what kind of AI behavior do we want, and how do we encode it?"

Constitutional AI (Anthropic's approach) is one answer: encode values into training, not just rules into prompts.

---

Related Pages

[[cybersecurity-thesis]] | [[machine-learning]] | [[economics-and-scarcity]] | [[active-projects]] | [[hakyun-ryu]]