Should Generative AI Lower the Bar for “A Person Having Ordinary Skill in the Art”?

By /

Imagine this scenario: you have a rough idea for an invention—call it “The Widget.” You know it needs elements A, B, and C, functioning as D to produce result E. However, you’re not entirely sure how to build B. In times past, patent law required you, as an inventor, to fully enable your invention in the specification so that a “person having ordinary skill in the art” (commonly shortened to PHOSITA) could make and use the invention without undue experimentation. But what if the PHOSITA now includes someone who is adept at tapping into advanced large language models (LLMs)—like ChatGPT, Bard, or “o1 pro”—to fill in the missing details? A simple prompt like “I’ve invented A, B, and C, functioning as D to produce result E, but I don’t know how to build B. Please describe how to build B” might be able to provide the missing data points.

Should the definition of a PHOSITA should be broadened to encompass the skill of using AI tools to fill in technical gaps left by the patent specification?

The concept is analogous to GPS systems. Having a sense of direction and ability to read maps were once prerequisites to becoming a taxi driver — i.e. one having ordinary skill in the art of driving a taxi. With the introduction of GPS, those elements are no longer something that a taxi driver having ordinary skill in the art is required to have. The same is now true of programming. While in the past, a person building a device with an embedded chip might be required to have the ability to write the program for that chip in order to be a PHOSITA, an AI can now easily take over that task, meaning that a PHOSITA need not possess that skill in light of AI.

Setting the Stage

The PHOSITA Standard

In patent law, the concept of “enablement” (35 U.S.C. § 112 in the United States) requires that the inventor disclose enough information for a PHOSITA to replicate the invention without undue experimentation. Traditionally, the PHOSITA is a hypothetical individual with average skill in the relevant technical field—someone who knows the common methods, tools, and knowledge bases of that field.

Enter SANDMAN Generative AI

Generative AI models (e.g., GPT-4, ChatGPT, Bard) have made it radically easier to perform research, find solutions, or generate novel configurations of ideas. Could an inventor now claim: “Look, I’ve disclosed enough so that a PHOSITA—who has access to AI—can figure out the missing details. Therefore, my invention is fully enabled”?

This raises a fascinating question: Should the threshold for enablement shift if one presumes that a PHOSITA now knows how to leverage these powerful AI systems?

Arguing in Favor of an Expanded PHOSITA

  1. Realistic Reflection of Modern Research Capabilities
    Today, a technically skilled person doesn’t stop at textbooks or standard library research; they almost always go online, run simulations, or consult specialized databases. The next logical step is tapping into advanced AI tools. If these tools are widely accessible and commonly used in the field, it’s arguably fair to incorporate their capabilities into the PHOSITA’s skillset.
  2. Efficiency and Innovation
    By recognizing that AI can fill in certain technical gaps, the law could incentivize broader, more conceptual disclosures. This might encourage inventors to share groundbreaking ideas earlier, trusting that the details can be worked out by a PHOSITA (assisted by AI). Proponents argue that this accelerates innovation by reducing the time and effort spent on drafting intricate, step-by-step mechanical or algorithmic details.
  3. Reflecting the Real World
    In many industries, especially high-tech or biotech, large parts of the process involve using specialized software and data-driven research tools. An expert in the field is already adept at these tools. Generative AI is simply the next evolution. If the actual, real-world person of “ordinary skill” is using AI, then the legal definition should mirror that reality.
  4. Consistent with the Doctrine of Equivalents
    Patent law already recognizes that minor variations or equivalents shouldn’t necessarily circumvent a patent’s scope. If AI can determine the best “equivalent” approach (e.g., “use a clip instead of a screw here”), that’s still within the technical reach of a skilled person. Therefore, including AI-based knowledge within a PHOSITA’s toolkit would be consistent with how patent law already handles variations and equivalents.
  5. Narrows what Patents Protect But Also Reduces Enablement Problems
    As an inventor, the narrowing side effect is undesirable. But overall it is probably good public policy. If an AI can come up with an implementation, and an AI is a common tool in the PHOSITA’s toolbox, the definition of what is considered routine matter with no patentable weight is broadened. So if my invention of widget a needs elements A, B, and C, functioning in some way D to produce a result E, and an AI can come up with C if given A, B, D and E, there is an argument that C doesn’t carry patentable weight. At the same time, if the combination of A, B, D and E is all I disclosed in my patent application, but C is required to make it function and C is within the skillset of a PHOSITA using AI, my omission of C from the specification should not give rise to patent invalidity under section 112.

Arguing Against Including AI in PHOSITA

  1. The Undue Experimentation Barrier
    The core requirement is that the inventor must provide enough detail so the invention can be replicated without undue experimentation. If key elements (like specific hardware choices or circuit designs) are missing, a patent examiner or a court might conclude that the inventor is effectively shifting the burden of invention onto the skilled person (or their AI). The current standard doesn’t allow an applicant to rely on future or external “invention” in order to enable their disclosure.
  2. Access and Uniformity Concerns
    While AI tools are becoming more prevalent, not everyone has equal access or ability to use them effectively. Patent law tries to envision a consistent standard across an entire field. If we assume that everyone has advanced AI, we may distort the uniform application of the law. Some might question: is it truly “ordinary skill” or is it “extraordinary skill plus AI?”
  3. Risk of Overly Broad Patents
    Patents that rely on AI to fill in critical gaps in the disclosure might become unreasonably broad. This poses the risk of stifling innovation by blocking competitors from related but distinct implementations. A patent that just names “elements A, B, C, achieving result E” without a robust specification might be too vague and grant overly broad rights, undermining the purpose of clarity in patent documents.
  4. Practical Legal Uncertainty
    Incorporating AI usage into the definition of a PHOSITA would require courts and examiners to assess how such AI might have been used. This would open up a new class of legal questions. Should an examiner ask for proof of how AI would fill in the missing details? Who’s to say the AI wouldn’t produce an incorrect or non-enabling solution in some scenarios? The unpredictability (and occasional hallucinations) of LLMs could complicate the enablement analysis.

Finding the Middle Ground

Patent law often evolves with technology. While generative AI is undoubtedly a powerful research tool, it’s uncertain how quickly or thoroughly courts and patent offices will adapt the PHOSITA standard. The ultimate balance might lie between:

  • Requiring a reasonable level of detail in the specification (so the invention is comprehensible).
  • Acknowledging common tools available to those of ordinary skill (potentially including advanced AI).

A legislative middleground (because current law does not allow this in the USA) would be to limit the patent term of any patent that requires AI to fill in a gap in the specification. Instead of 20 years from filing, such a patent might get 12 years from filing. This reduced incentive to invent would match the reduced cost of inventing through the use of AI. A similar change might be appropriate for drugs created using quantum computing.

In practice, we might see examiners and courts demand that key inventive aspects be disclosed, while allowing some less critical details to be left to routine experimentation—potentially aided by AI.

Conclusion

The idea that a “person having ordinary skill in the art” might include someone adept at using ChatGPT or other LLMs provokes compelling questions about the future of patent drafting and prosecution. While AI is clearly playing an increasingly prominent role in R&D, the legal requirements of enablement and written description remain foundational.

In favor of recognizing AI-assisted PHOSITAs: it mirrors the reality of modern research, stimulates more open sharing of core concepts, and reflects the next step in scientific tools.
Against it: it may encourage inadequate disclosure, raise fairness concerns, and lead to overbroad claim drafting.

Whichever side ultimately prevails, one thing is certain: the integration of generative AI in the patent process will continue to spark debate and shape our legal landscape for years to come.

POSTSCRIPT

Just for fun, let’s see what ChatGPT o-1 Pro can come up with. I often observe that the saying “don’t re-invent the wheel” is perhaps the worst advice an inventor can get. The wheel is one of the most valuable inventions. Everybody uses them. Even a small breakthrough in improving the wheel would be enormously valuable. So let’s check out AI “creativity”. And good news, since the invention is now published, it is prior art to any human-generated invention that comes after.

Prompt: You are an inventor. You are writing a patent disclosure for your lawyer. It should fully enable the invention you come up with, and should describe it in enough detail that a person having ordinary skill in the art can build it. The invention is an improved wheel. You want to find a novel, non-obvious improvement to a wheel. It can be any kind of wheel — from a wheel on a toy car to a wagon wheel to a wheel on a mars rover. You want this to be the most valuable possible improvement that is also patentable. Please write the disclosure.

Title:
Adaptive, Shape-Memory Wheel Assembly with Integrated Real-Time Terrain Response

BACKGROUND

Wheels have been in continuous use and development for millennia, ranging from simple wooden wagon wheels to sophisticated pneumatic tires on modern automobiles. Improvements typically focus on materials, treads, suspension interfaces, and internal supports. Current wheels, however, often rely on passive mechanisms (e.g., fixed tread patterns, unchanging sidewalls) that do not adapt in real time to changing terrain conditions, load distributions, or environmental factors such as temperature and humidity.

Certain applications, including off-road vehicles, Mars rovers, and specialized robotic systems, face dynamic and unpredictable terrains. Conventional wheels or tires used in these environments may experience:

  • Excessive vibrations and instability on rough or rocky terrain.
  • Inefficient energy usage due to imperfect contact patch or slip.
  • Wear and tear from repeated shape deformation (e.g., punctures, sidewall damage).
  • Limitations on traction in extreme conditions (ice, sand, wet surfaces).

There is thus a long-felt need for a wheel capable of dynamically adjusting its characteristics—such as diameter, tread pattern, and stiffness—to optimize traction, comfort, stability, and longevity in real time.

SUMMARY OF THE INVENTION

The present disclosure describes an Adaptive, Shape-Memory Wheel Assembly (the “Adaptive Wheel”) configured to detect terrain and operational conditions and alter its geometry on-the-fly. The Adaptive Wheel incorporates:

  1. Shape-Memory Materials for selective expansion, contraction, and stiffness modulation.
  2. Embedded Sensor Arrays that detect load distribution, temperature, and traction feedback.
  3. Controller and Actuation System that interpret sensor data and activate localized shape changes to optimize performance.
  4. Reinforced, Self-Healing Elastomer Layer or composite tread that repairs minor abrasions and punctures over time.

By continuously adjusting contact patch size, radial stiffness, and traction properties, the Adaptive Wheel reduces wear, enhances stability, and improves efficiency across diverse use cases.

BRIEF DESCRIPTION OF THE DRAWINGS

While no drawings are attached to this disclosure, the following figure descriptions are provided to further enable someone of ordinary skill in the art to practice the invention:

  • FIG. 1: A perspective view of the Adaptive Wheel showing the layered construction of the rim, spokes, shape-memory segments, and outer elastomeric tread.
  • FIG. 2: A cross-sectional view illustrating the embedded sensor and actuator arrangement, including the controller module and energy source.
  • FIG. 3: An exploded view detailing the shape-memory segments, the self-healing elastomer layer, and the communication lines running between the sensor arrays and the central controller.
  • FIG. 4: A schematic diagram of the control logic used to dynamically modulate wheel geometry in response to terrain feedback.

DETAILED DESCRIPTION OF EMBODIMENTS

1. Structural Overview

  1. Wheel Hub (10): The hub is configured to interface with a vehicle’s axle or drive shaft. This hub may include standard bearings, but can also be adapted for specialized robotics or rover drives.
  2. Spoke Assembly (20): The wheel may use conventional spokes or a disc-like structure. In one embodiment, each spoke is connected to a shape-memory segment (30) that can change length or tension.
  3. Shape-Memory Segments (30):
    • Composition: A shape-memory alloy (e.g., Nitinol) or shape-memory polymer.
    • Function: Upon receiving electrical or thermal stimuli, each segment can expand, contract, or otherwise deform.
    • Arrangement: The segments are circumferentially spaced around the rim. A central controller orchestrates each segment’s activation for localized or global deformation.
  4. Rim (40):
    • Material: Composites such as carbon fiber, aluminum alloys, or advanced polymers to provide structural integrity while maintaining low weight.
    • Mounting: Integrates mechanical or chemical fasteners that secure the shape-memory segments between the rim and the spoke assembly.
  5. Self-Healing Elastomer Layer (50):
    • Outer Tread: A tough, abrasion-resistant elastomer with embedded microcapsules of healing agent. When small cracks or punctures occur, the microcapsules rupture, release healing fluid, and bond with the surrounding material to seal the damage.
    • Optional Embedded Fiber Reinforcement: Aramid or carbon fiber strands can be added to further strengthen the tread and minimize blowouts under high load.

2. Sensor and Actuation System

  1. Sensor Array (60):
    • Load Sensors: Strain gauges, piezoelectric sensors, or MEMS-based stress monitors integrated within the rim or shape-memory segments.
    • Traction/Slip Sensors: Optical or accelerometer-based devices that measure wheel slip relative to ground speed.
    • Temperature and Humidity Sensors: For adjusting the shape-memory alloy’s activation (heat needed to switch states) or the polymer’s properties in cold or humid conditions.
  2. Controller Module (70):
    • Microcontroller or FPGA: Interprets sensor signals and coordinates actuation signals.
    • Power Management: A battery, capacitor bank, or direct connection to the vehicle’s power system provides energy to the shape-memory segments.
    • Feedback Loop: Real-time software (or firmware) that compares desired wheel performance to actual sensor data, adjusting shape and stiffness as conditions change.
  3. Actuation Method:
    • Thermal Activation: If shape-memory alloy is used, electric current is passed through the alloy to produce the heat required to cause a phase transformation.
    • Electrical or Electromagnetic Stimulation: For shape-memory polymers, localized heating elements or electromagnetic fields can be used to alter the polymer state in targeted regions.

3. Operation and Control Logic

  1. Initialization: When the vehicle (or rover) starts up, the controller performs a diagnostic test of each shape-memory segment and sensor.
  2. Data Collection: As the wheel rotates, sensors continually read load, slip, and environmental data.
  3. Decision Algorithm:
    • Compare: The sensor readings are compared to a set of predetermined thresholds (e.g., maximum allowable slip ratio, optimal contact patch size).
    • Compute: The controller computes the magnitude and location where shape adjustment is needed.
    • Command: The relevant shape-memory segments are energized to either expand or contract, adjusting the wheel’s local curvature, diameter, or stiffness.
  4. Self-Healing Activation: If small punctures or tears are detected (e.g., by rapid pressure or thickness changes), the controller can engage mild compression to encourage contact between the damaged surfaces and release healing agent from microcapsules.
  5. Continuous Loop: This sensing and adjustment cycle repeats continuously to adapt the wheel to immediate terrain conditions in real time.

4. Alternative Embodiments

  1. Passive Return to Default Shape: Once de-energized, certain shape-memory materials can revert to a baseline shape. Alternatively, the wheel could maintain its last shape until actively reset.
  2. Multi-Zone Tread Design: Different tread patterns (e.g., all-terrain vs. highway) embedded in discrete segments that can be exposed or retracted as needed.
  3. In-Wheel Energy Harvesting: Incorporating electromagnetic induction (via rotation) or piezoelectric harvesters to partially power the shape-memory elements.
  4. Airless or Pneumatic Hybrid: The wheel can be partially pneumatic for shock absorption while also using shape-memory ribs or beams within the tire structure to maintain shape in the event of pressure loss.

MANUFACTURING CONSIDERATIONS

  1. Shape-Memory Alloy/Polymer Fabrication:
    • Requires precise thermal treatment to achieve desired transition temperatures and mechanical properties.
    • Pre-programming the shape-memory elements involves fixturing them in a “trained” shape during a heat treatment.
  2. Assembly:
    • The shape-memory segments are mounted between the rim and each spoke with mechanical connectors or adhesive bonding.
    • Internal wiring is routed through channels in the rim or spokes to protect against wear and environmental elements.
  3. Elastomer Layer Application:
    • The self-healing elastomer can be molded or extruded around the rim.
    • Microcapsules containing the healing agent must be evenly distributed to ensure consistent performance.
  4. Quality Control:
    • Each finished wheel undergoes functional testing to verify sensor calibration, shape change accuracy, and self-healing efficacy under simulated punctures.

ADVANTAGES AND INDUSTRIAL APPLICABILITY

  1. Enhanced Traction Control: Actively adjusting contact patch size and tread texture improves handling on uneven or slippery surfaces.
  2. Reduced Wear and Tear: The wheel’s ability to adapt stiffness and shape can dissipate shocks more efficiently, prolonging service life.
  3. Energy Efficiency: By optimizing the wheel’s geometry for minimal rolling resistance during steady travel, energy consumption (fuel or battery) is reduced.
  4. Self-Healing Properties: Routine minor damages are sealed automatically, decreasing downtime and replacement costs.
  5. Versatile Applications: The Adaptive Wheel can be used on terrestrial vehicles, off-road equipment, robotic platforms, or planetary rovers.

CONCLUSION

This disclosure presents an Adaptive, Shape-Memory Wheel Assembly designed to tackle the longstanding challenges of varying terrain, shock absorption, traction management, and tire longevity. By integrating shape-memory materials with embedded sensors, real-time controller logic, and a self-healing elastomer tread, the wheel is capable of maintaining optimal performance under diverse and changing conditions.

The described invention is believed to be both novel and non-obvious over existing wheel technologies. It addresses a need in multiple industries—particularly those requiring precise control of wheel performance in unpredictable or extreme environments. A person of ordinary skill in the art, armed with the details provided herein, can readily build, modify, and implement the disclosed Adaptive Wheel.

This disclosure should not be interpreted as limiting. Instead, it is intended to cover all variations and alternatives falling within the scope of the invention as defined by the appended claims (which may be drafted in a later phase). All technical or functional equivalents are considered part of this disclosure.

1 thought on “Should Generative AI Lower the Bar for “A Person Having Ordinary Skill in the Art”?”

  1. Pingback: How to Patent an Idea (the Right Way)

Leave a Comment

Scroll to Top