1. Meaning of “Algorithmically Created Smart Materials”

These are materials developed using:

• Machine learning models
• Computational chemistry simulations
• Generative design algorithms
• AI-driven molecular discovery tools

Examples:

• Self-healing concrete
• AI-designed graphene composites
• Shape-memory polymers for aerospace
• Adaptive thermal insulation materials

Legal complexity:

Such materials raise IP questions in:

• Patent law (main protection route)
• Trade secrets
• Copyright (software models, not materials)
• Data rights (training datasets)

2. What exactly is protected?

(A) Patent protection (primary)

• Composition of matter
• Manufacturing process
• Functional properties (e.g., conductivity, elasticity)

(B) Trade secrets

• Training datasets
• AI models used for material discovery
• Optimization parameters

(C) Software/IP

• Algorithms used for molecular prediction

3. Core Legal Issue

The key legal challenge is:

Can something discovered or designed by an algorithm be “invented” under patent law?

Courts and patent offices generally say:

• AI can assist invention
• But human inventorship is required (in most jurisdictions)

4. Important Case Laws (Detailed Explanation)

4.1 Diamond v Chakrabarty (1980, US Supreme Court)

Facts:

• Scientist engineered genetically modified bacteria to break down oil spills.

Held:

• Living organisms can be patented if human-made
• Allowed patent on “man-made microorganism”

Principle:

“Anything under the sun made by humans is patentable subject matter.”

Relevance to smart materials:

• AI-designed materials (like nano-structured polymers) are patentable if:
  • They are not natural
  • They show human-directed innovation

Key takeaway:

Even if AI assists, the end material is patentable if human-directed invention exists.

4.2 Mayo Collaborative Services v Prometheus (2012, US Supreme Court)

Facts:

• Patent on a medical diagnostic method using natural correlations.

Held:

• Laws of nature, natural phenomena cannot be patented
• Mere application of natural law is not enough

Principle:

“You cannot patent natural laws just by applying them.”

Relevance:

If an AI discovers:

• A natural property of a material (e.g., conductivity pattern in graphene)

Then:

• That discovery alone is NOT patentable
• Only practical application in engineered material is patentable

4.3 Association for Molecular Pathology v Myriad Genetics (2013, US Supreme Court)

Facts:

• Patents on isolated human DNA sequences.

Held:

• Naturally occurring DNA is not patentable
• cDNA (synthetic DNA) is patentable

Principle:

Isolation of natural material alone is not invention.

Relevance:

For AI-designed materials:

• If AI identifies a naturally existing compound:
  • Not patentable
• If AI creates synthetic variation or modified structure:
  • Patentable

4.4 Alice Corp. v CLS Bank (2014, US Supreme Court)

Facts:

• Patent on computer-implemented financial system.

Held:

• Abstract ideas implemented on a computer are not patentable

Principle:

“Abstract idea + generic computer = not patentable invention”

Relevance:

If AI system merely:

• Optimizes known material formulas

Then:

• It may be considered abstract computational process
• Not patentable unless it produces:
  • Concrete material transformation

4.5 Thaler v Vidal (2022, US Federal Circuit)

Facts:

• Inventor listed AI system (“DABUS”) as sole inventor.

Held:

• Only natural persons can be inventors under patent law

Principle:

AI cannot be legally recognized as an inventor

Relevance:

For smart materials:

• Even if AI fully designs the material:
  • A human must be listed as inventor
• Companies must document:
  • Human contribution (selection, validation, optimization)

4.6 UKIPO “DABUS” Patent Cases (2020–2023 decisions)

Facts:

• Patent applications filed naming AI as inventor.

Held:

• Rejected because inventor must be human

Principle:

AI-generated inventions require human inventor attribution

Relevance:

In industrial smart materials:

• AI-generated polymer design is patentable only if:
  • Human is identified as inventor
  • AI is treated as tool

4.7 Enfish LLC v Microsoft (2016, US Federal Circuit)

Facts:

• Patent on self-referential database system.

Held:

• Software improving computer functionality is patent-eligible

Principle:

Technical improvement to system = patentable

Relevance:

AI-designed smart materials:

• If algorithm improves:
  • Material strength
  • Thermal response
  • Conductivity
    Then:
• The invention is considered technical improvement → patentable

4.8 KSR International v Teleflex (2007, US Supreme Court)

Facts:

• Patent on combination of known mechanical elements.

Held:

• Obvious combinations are not patentable

Principle:

Mere combination of known elements = obvious invention

Relevance:

AI often recombines known materials:

• If AI merely mixes existing compounds:
  • Not patentable (obviousness rejection)
• Must show:
  • Unexpected technical effect

5. Legal Principles Derived from Case Law

(A) AI is a tool, not inventor

From Thaler + UK DABUS cases:

• Human inventorship required

(B) Natural discovery is not invention

From Mayo + Myriad:

• AI discovering natural properties ≠ patent

(C) Technical transformation is key

From Alice + Enfish:

• Must show real-world material improvement

(D) Non-obviousness is essential

From KSR:

• AI-generated combinations must not be trivial

6. Application to Smart Materials

Example 1: AI-designed self-healing polymer

✔ Patentable if:

• New molecular structure created
• Improved healing rate proven
• Human inventor identified

Example 2: AI identifies natural heat-resistant mineral

✘ Not patentable (discovery of nature)

Example 3: AI optimizes carbon fiber composite

✔ Patentable if:

• New composition or process exists
• Demonstrates unexpected strength increase

Example 4: AI generates thousands of polymer candidates

✔ Only selected human-validated candidate is patentable

7. Key Legal Challenges

7.1 Inventorship crisis

Who is the inventor when AI contributes 90% of design?

7.2 Disclosure problem

Companies may not want to reveal:

• AI models
• training datasets

7.3 Obviousness explosion

AI can generate millions of combinations → raises bar for “non-obviousness”

7.4 Ownership ambiguity

• Employer vs developer vs AI tool provider

8. Conclusion

Protection of algorithmically created smart materials is primarily governed by patent law, but shaped heavily by judicial principles:

Core rule from all major cases:

AI may assist discovery, but legal protection requires human-directed inventive contribution producing a technical, non-obvious material transformation.