PROTECTING VIRTUAL THINGS: PATENTABILITY OF ARTIFICIAL .
328PROTECTING VIRTUAL THINGS:PATENTABILITY OF ARTIFICIALINTELLIGENCE TECHNOLOGYFOR THE INTERNET OF THINGSANASTASIA GREENBERG*ABSTRACTThe Internet of Things (IoT) welcomes physicaleveryday objects into the connected digital world. ExistingIoT devices include virtual assistants, smart thermostats,and fitness trackers. IoT innovation is booming, bringingto market increasingly sophisticated devices with immensepotential for improving human well-being in the areas ofsmart cities, efficient manufacturing, and personalizedhealthcare. The true engine behind the IoT revolution isartificial intelligence (AI) which uses computing power tolearn from big data generated by IoT sensors to deliversmart solutions and accurate predictions—furnishing IoTdevices with their value. To gain a bird’s eye perspectiveon the future development of AI applications for IoT(termed AI-IoT in this Article), one important considerationis whether such technology can enjoy intellectual propertyprotection in the form of patents, and what theconsequences are of such patents on the AI-IoT innovationlandscape. Part I of this Article introduces the concepts ofAI and machine learning and describes criteria for*Ph.D.; Associate in Intellectual Property, WilmerHale, Boston. Igraduated from McGill University’s Faculty of Law and hold a Ph.D. inNeuroscience from the University of Alberta. The content of thisArticle is my own view, was developed and submitted for publicationwhile I was enrolled at McGill University’s Faculty of Law, and doesnot represent the views of any of the organizations that I am currentlyaffiliated with. Correspondence: anastasia.greenberg@mail.mcgill.ca.Volume 60 – Number 2
Protecting Virtual Things329obtaining a patent under United States intellectual propertylaw.Part II of this Article covers the historicalbackground of the subject matter eligibility of softwarepatents through jurisprudential and policy developmentswith a focus on implications for the patentability of AI-IoT.Part III of this Article addresses innovation policyconsequences of proliferation of AI-IoT patents. TheArticle finds that AI-IoT patents present a unique set oftangible inventions that may circumvent the “abstractidea” obstacle to subject matter eligibility faced by manysoftware patents. However, current evidence is ambiguousas to whether the growth of such patents would stimulate ordampen AI-IoT innovation. In any case, AI-IoT patentsshould be welcomed as current patent law does not have aclear legal test for exempting such patents and atechnologically-neutral approach to intellectual propertyshould be embraced.Abstract . 328I.Introduction . 330A.Artificial Intelligence for IoTDevices (AI-IoT) . 331B.Artificial Intelligence and Machine Learning . 333C.What is a Patent? . 335II.Are AI-IoT Inventions Patentable SubjectMatter? . 336A.The Alice Decision . 338B.The Aftermath of the Alice Decision . 339C.Patentability Challenge and Promisefor AI-IoT. 340Volume 60 – Number 2
330 IDEA – The Law Review of the Franklin Pierce Center for Intellectual PropertyIII. Innovation Policy Perspective onPatenting AI-IoT . 344A.Evidence for Negative Impact ofSoftware/AI Patents on Innovation . 345B.Evidence for the Positive Impact ofSoftware/AI Patents on Innovation . 348IV. Conclusion . 349I.INTRODUCTIONThe Internet gave rise to a virtual world, and theInternet of Things (IoT) will merge that virtual world withthe physical one. IoT consists of networks of physicaldevices connected to the Internet which gather data fromtheir environment using sensors, share information acrossthe network, and allow for intelligent data analysis.1Digitization of the physical world through IoT is expectedto drive the fourth industrial revolution.2 Bain estimatedthat the global IoT market will grow from 235 billion in2017 to 520 billion by 2021.3 The main areas ofapplication of IoT technology include smart cities formanaging traffic and other public infrastructure,autonomous vehicles, worksite infrastructure for predictive1Amy JC Trappey et al., A Review of Essential Standards and PatentLandscapes for the Internet of Things: A Key Enabler for Industry 4.0,33 ADVANCED ENGINEERING INFORMATICS 208, 208 (2017).2Jean-Marc Frangos, The Internet of Things will Power the FourthIndustrial Revolution. Here’s How, WORLD ECON. FORUM (June volution [https://perma.cc/Q2C2-UWV5].3Ann Bosche et al., Unlocking Opportunities in the Internet of Things,BAIN (Aug. 7, 2018), ies-in-the-internet-of-things [https://perma.cc/VN5H-HPBX].60 IDEA 328 (2020)
Protecting Virtual Things331maintenance, security, precision farming, and connectedhealth through wearables.4 As a practical example of asmart city IoT application, the cities of Doha, Sao Paulo,and Beijing use sensors attached to water infrastructure tomonitor and mitigate water loss.5 Since 2014, there aremore IoT devices in use than the world’s humanpopulation.6A.Artificial Intelligence for IoT Devices (AIIoT)Despite the significant promise of IoT for botheconomic and social benefit, the full potential of IoTremains unrealized. IoT devices gather massive amounts ofcomplex data, with only a small portion of that data beinganalyzed for practical ends. For example, McKinseyGlobal Institute claimed that less than one percent of thedata being collected by thirty thousand sensors on aspecific oil rig are used in decision-making.7 The key toextracting the maximum value from such “big data” (i.e.,large datasets with many sources and variables) is throughintelligent data processing and analysis using artificialintelligence (AI).8 AI is defined as “the capability of amachine to imitate intelligent human behavior [and4LexInnova, Internet of Things: Patent Landscape Analysis, WIPO 1,2–3 (2014), https://www.wipo.int/edocs/plrdocs/en/internet of things.pdf [https://perma.cc/K4CF-LL48].5Id.6See Trappey et al., supra note 1, at 209.7James Manyika et al., Unlocking the Potential of the Internet ofThings, MCKINSEY GLOBAL INSTITUTE (June 2015), -value-of-digitizing-the-physical-world [https://perma.cc/7DVY-NJTF].8Mohammad Saeid Mahdavinejad et al., Machine Learning for Internetof Things Data Analysis: A Survey, 4 DIGITAL COMM. & NETWORKS161, 161 (2017).Volume 60 – Number 2
332 IDEA – The Law Review of the Franklin Pierce Center for Intellectual Propertyintuition].”9 AI algorithms are applied to big data to extractmeaning from these data by categorizing information,finding patterns, and making predictions.To trulyappreciate the power of AI for analyzing big data, it isimportant to understand that while AI imitates “humanintuition,” unlike AI, human intuition fails at extractingrelevant patterns from big data and drawing accurateconclusions based on these patterns. Most AI technologyfinds its applications in analyzing big data on the Internet.10Researchers were able to predict flu trends using dataobtained from Twitter, Facebook uses AI for facialrecognition of users’ image posts, and Netflix uses AI tomake personally catered movie/show recommendations tosubscribers.11 Such AI-based analysis is arguably the mostcritical component of IoT. AI is essential for trainingautonomous vehicles to make decisions, predicting healthissues from data obtained by wearable devices, andregulating congestion from traffic data. It is therefore thecombination of IoT and AI that marks the entry point to thenext industrial revolution. This Article examines thecurrent and future innovation landscape for IoT technology,with a specific focus on AI software development for IoT(AI-IoT).Longstanding legal theory suggests thatintellectual property rights are essential for incentivizingcreation by giving creators/inventors a time-limitedmonopoly on the fruits of their labor in exchange for publicdissemination of knowledge. Under this framework, theArticle asks whether AI-IoT inventions can enjoy patent9Artificial Intelligence, MERRIAM-WEBSTER, al intelligence [https://perma.cc/24UHMR3Z] (last visited Nov. 10, 2019).10Hidemichi Fujii & Shunsuke Managi, Trends and Priority Shifts inArtificial Intelligence Technology Invention: A Global Patent Analysis,58 ECON. ANALYSIS & POL’Y 60 (2018).11Hyunjong Ryan Jin, Think Big! The Need for Patent Rights in theEra of Big Data and Machine Learning, 7 N.Y.U. J. INTELL. PROP. &ENT. L. 78, 102 (2018).60 IDEA 328 (2020)
Protecting Virtual Things333protection under United States’ law. The focus is onpatents since patents protect the functional aspects of theinvention, while copyright protection is concerned with theliteral copying of software code. The Article assumes thatthe true value of AI-IoT applications is its technicalfunction which can be protected through patents. The focusis also on U.S. law since the majority of both IoT and AIpatents are filed in the US.12 The last part of the Articleturns to policy considerations discussing the advantagesand drawbacks of using patent law for incentivizinginnovation in the AI-IoT space.B.Artificial Intelligence and MachineLearningBefore delving into whether AI technologies arepatentable, it is crucial to understand in some detail how AIalgorithms work. The term AI is most often used to refer toa specific category of algorithms called machine learning(ML) that allows computers to learn from data withoutbeing explicitly programmed or “hard-coded.”13 MLalgorithms are “trained” on complex data sets and are ableto learn relevant patterns and correlations from“experience.”14 There are three main categories of ML:supervised, unsupervised, and reinforcement learning. Insupervised learning, the input data (i.e., the “training data”)is labeled with the correct response and the algorithmlearns the relationship between the data and the labels tomake predictions on new, previously unseen data.15 Anexample of supervised learning is an algorithm that istrained on many pictures labeled as either containing or notcontaining a cat, to then be able to identify a cat picture that12131415Trappey et al., supra note 1, at 219.Mahdavinejad et al., supra note 8, at 165.Jin, supra note 11, at 88.Jin, supra note 11, at 89.Volume 60 – Number 2
334 IDEA – The Law Review of the Franklin Pierce Center for Intellectual Propertyit has not previously seen. In unsupervised learning, thealgorithm is fed a complex data set, but without any labels.The algorithm finds interesting patterns in the data withoutbeing shown any correct solutions. An example of this isan algorithm that is given a compilation of news articlesand the algorithm learns to group all the articles about thesame news event into one cluster.16 In the context of AIIoT, an algorithm fed heart rhythm data from a wearablecould be trained to recognize abnormal heart activity eitherthrough supervised learning by being shown previouslylabeled examples of abnormal heart signals, or throughunsupervised methods by using the data to categorizedifferent heart activity patterns into groups without labels(a person will then decide which group contains theabnormal heart rhythms). Reinforcement learning involvesalgorithms learning sequences of actions to be taken for agiven situation in order to maximize payoff, such astraining a robot to make a series of complex decisions whenplaying soccer.17The steps involved in developing a ML algorithmare (1) obtaining high-quality data for training thealgorithm, such as data acquired by sensors on IoT devices;(2) pre-processing data including cleaning data byremoving outliers or reducing dimensions; (3) training aML algorithm on the data (the ML algorithm is either anoff-the-shelf algorithm commonly used or a newlydeveloped one); and (4) obtaining a final trained algorithm(i.e., the model) which gives output data (solutions) whenshown new input data.18 While it is true that MLalgorithms learn from the data and come up with a finalmodel spontaneously, the developer’s ingenuity still plays amajor role. Many human decisions need to be made during161718Jin, supra note 11, at 89.Mahdavinejad et al., supra note 8, at 165.Jin, supra note 11, at 92.60 IDEA 328 (2020)
Protecting Virtual Things335the development process including choosing which MLmethod(s) to employ for a given problem, how to curate thetraining data, which algorithm parameters to select, andhow to test the model for accuracy. When considering apatent for an AI/ML invention, an inventor may seekprotection for either a single development step, or morecommonly, a series of these steps presented as a whole.C.What is a Patent?Section 101 of the Patent Act allows for four typesof inventions to receive patent protection: (1) processes; (2)machines; (3) manufactures; and (4) compositions ofmatter.19 AI-IoT would fall under either process (e.g., stepsin algorithm implementation) or machine (e.g., AIcombined with a physical IoT device). The term of a patentis twenty years from the date on which the application forthe patent was filed in the United States.20 In order toobtain a patent under one of these categories, an inventionneeds to meet all of the following criteria: it must be (1)patent-eligible subject matter; (2) useful; (3) novel; and (4)non-obvious.21 The courts have interpreted patentablesubject matter to mean that almost any invention is eligibleexcept for laws of nature, natural phenomena, and abstractideas.22 Useful means that the invention has a practicalapplication, not merely a theoretical application in thefuture. Novel means that the invention does not repeat“prior art” (i.e., previously patented inventions), has notbeen publicly disclosed, and is not generally known. Non1935 U.S.C. § 101 (2018).35 U.S.C. § 154(a)(2) (2018).2135 U.S.C. § 101 (patentable subject matter and utility); 35 U.S.C. §102(a) (2018) (novelty); 35 U.S.C. § 103 (2018) (non-obviousness).22Allen Clark Zoracki, When Is an Algorithm Invented? The Need for aNew Paradigm for Evaluating an Algorithm for Intellectual PropertyProtection, 15 ALB. L.J. SCI. & TECH. 579, 589 (2004).20Volume 60 – Number 2
336 IDEA – The Law Review of the Franklin Pierce Center for Intellectual Propertyobvious inventions are those that would not have beenobvious to others “skilled in the art” (i.e., experts in thefield). Therefore, requiring some kind of creative /knowledge in a standard way is a prerequisite tobeing patentable.For AI-related inventions, eligiblesubject matter represents the largest hurdle to overcomesince algorithms and mathematical formulae have beentraditionally considered to be abstract ideas.23II.ARE AI-IOT INVENTIONS PATENTABLE SUBJECTMATTER?The U.S. Supreme Court has stated that“phenomena of nature, though just discovered, mentalprocesses, and abstract intellectual concepts are notpatentable, as they are the basic tools of scientific andtechnological work.”24 The courts are against issuingpatents for such work since “monopolization of those toolsthrough the grant of a patent might tend to impedeinnovation more than it would tend to promote it.”25Although advanced AI technology, and its use in IoT, isrelatively new, the courts have been grappling with thepatentability of computer software inventions for almosthalf a century.Prior to the 1980s, it was generally accepted thatsoftware represented abstract mathematical concepts andremained unpatentable subject matter.26 In 1981, theSupreme Court ruled in Diamond v. Diehr that a formula,implemented on a digital computer, for curing rubber was23Id. at 588.Gottschalk v. Benson, 409 U.S. 63, 67 (1972).25See Mayo Collaborative Servs. v. Prometheus Labs., Inc., 132 S. Ct.1289, 1293 (2012).26See Gottschalk, 409 U.S. at 67.2460 IDEA 328 (2020)
Protecting Virtual Things337patentable.27 The patent included the process of “installingrubber in a press, closing the mold, constantly determiningthe temperature of the mold, constantly recalculating theappropriate cure time through the use of the formula and adigital computer, and automatically opening the press at theproper time.”28 Given that the computer software improveda physical-industrial process as a whole, the Court did notview the invention as abstract, and therefore it was upheldas valid.29 Following this decision, the United States Courtof Appeals for the Federal Circuit became more open toaccepting that software can be patent-eligible subjectmatter. In Alappat, the Federal Circuit accepted thevalidity of a software patent that processed data to allow forthe display of a smooth waveform on a digitaloscilloscope.30 The court viewed the patent as creating amachine in which a general-purpose computer was turnedinto a special-purpose computer when running the softwareto digitize the waveforms.31In State Street Bank and Trust Company v.Signature Financial Group, Inc., the Federal Circuit foundthat an algorithm used to calculate a share price was patenteligible since it constituted a practical and tangibleapplication of a mathematical formula that could be usedfor recording and reporting purposes.32 In State StreetBank, the court also articulated the “machine-ortransformation” test which provided that an algorithm orsoftware is patentable if (1) it is tied to a particular machineor apparatus; or (2) it transforms a particular article into a27See Diamond v. Diehr, 450 U.S. 175 (1981).Id. at 187.29Id. at 191–92.30See In re Alappat, 33 F.3d 1526, 1544 (Fed. Cir. 1994).31See id.32See State St. Bank & Trust Co. v. Signature Fin. Grp., Inc., 149 F.3d1368 (Fed. Cir. 1998).28Volume 60 – Number 2
338 IDEA – The Law Review of the Franklin Pierce Center for Intellectual Propertydifferent state or thing. Later, in Bilski v. Kappos, theSupreme Court clarified that the “machine-ortransformation” test was useful but not the sole criterion fordetermining subject matter eligibility.33 Following StateStreet Bank, the United States Patent and Trademark Office(USPTO) saw a proliferation of software- and Internetrelated patent applications.34 The decision opened a can ofworms, more than doubling the annual number of softwarepatent applications in the years following the decision,including patent applications implementing basiccalculations on a computer, as well as “business method”patents.35 The most famously known business methodpatent granted following State Street Bank is Amazon’s“one-click” patent allowing customers to make single-clickpurchases based on previously stored paymentinformation.36A.The Alice DecisionThe Supreme Court did not return to the question ofsoftware subject matter eligibility for almost two decadesuntil the case of Alice Corp. v. CLS Bank International in2014, marking a major turning point in the fate of softwarepatents.37 The patent in question involved a computerizedmethod of mitigating settlement risk by keeping track ofeach party’s account balance to prevent one party from33See Bilski v. Kappos, 561 U.S. 593 (2010).Fabio E. Marino & Teri H. P. Nguyen, From Alappat to Alice: TheEvolution of Software Patents, 9 HASTINGS SCI. & TECH. L.J. 1, 6(2017).35Christopher W. Quinn, The 20 Year War On Patents: When Will ItEnd?, LEXOLOGY https://www.lexology.com/library/detail.aspx?g 8cdd3dd7-1fb3-48dc-a7ba-b6ffb0076e8e [https
Internet of Things (IoT) will merge that virtual world with the physical one. IoT consists of networks of physical devices connected to the Internet which gather data from their environment using sensors, share information across . 2 Jean-Marc Frangos, The Internet of Things will Power the Fourth Industrial Revolution.
area due to public concern about patenting life and due to various moral concerns. However, thousands of biotechnology patents exist because patent agents have drafted claims that avoid the exclusions to patentability. Ex: Methods of treating the human body are excluded from patentability in certain jurisdictions but not « Swiss Type of claims ».
Each NETLAB remote PC or remote server runs inside of a virtual machine. VMware ESXi provides virtual CPU, virtual memory, virtual disk drives, virtual networking interface cards, and other virtual hardware for each virtual machine. ESXi also provides the concept of a virtual networking switch.
"Virtual PC Integration Components" software must be installed into each virtual machine. In a Windows host, the "Virtual PC Integration Components" software for a Windows virtual machine is located at C:\Program Files (x86)\Windows Virtual PC\Integration Components\ Multiple virtual machines can access the same target folder on the host.
Virtual Reality Racket Sports: Virtual Drills for Exercise and Training . [41,60,63], and visualization [1,20]. The use of virtual . the development of additional advanced virtual reality interfaces applicable to exercising and training in virtual reality racket sports.
“virtual birthday card” “virtual card” “virtual baby shower games” “virtual birthday party ideas” “virtual graduation” “virtual wedding” “virtual birthday card
Page 3 Installing Virtual PC Double-click the downloaded setup file and follow the prompts to install the program A folder called My Virtual Machines is created the My Documents folder. The supplied Virtual Machines must be copied into this folder Starting Virtual PC Click Start, All Programs and Microsoft Virtual PC Click the Title Bar of the Virtual PC Console and drag it to the upper right .
NCLEX Review Participant 17 Virtual-ATI Module 3. A Day in the Life of a Virtual-ATI NCLEX Review Participant. 18. Virtual-ATI Module 3. A Day in the Life of a Virtual-ATI NCLEX Review Participant. 19. Virtual-ATI Module 5. Attainment of 50% Mile-Marker. A Day in the Life of a Virtual-ATI NCLEX Review Participant. 20 .
Because children are more vulnerable than adults, and quite often are not able to make decisions for themselves, the Code protects them more than adults. The Code makes it clear that children’s interests can only be overridden if there is an exceptional reason to do so. In the Code, there are some specific clauses which protect children and
