Do Algorithms Dream of Economic Endeavour?

Retrofuturism mourns the deprivation of an “enchanted world” with crystalline aims and conviction in human consciousness (Svetlana Boym). Algorithms have obscured the law into a “black box” that needs to be interpreted. Should algorithms assume a new life as an “invention” under patent protection today? Or, does this erode the fundamental principle of “manner of manufacture” from its chain novel? Do Algorithms Dream of Economic Endeavour?

‘The question is not, ‘Can they reason?’ Nor ‘Can they talk’ But, ‘Can they suffer?’-- Jeremy Bentham.¹

Algorithms have encrypted themselves within the Intellectual Property system. Although our laws are reasonably consistent with the original IP principles, the opaqueness of algorithms has raised concerns about the aims underlying the patent regime. Bentham’s inquiry, ‘Can they suffer?’, challenges whether non-human entities deserve corresponding moral or legal protection.² Transposed into the IP system, the question transforms: If conscious suffering is a prerequisite for proprietary rights, how do autonomous yet insentient algorithms fit withinthe IP framework? This article analyses the inevitable inconsistencies between algorithms and the law, underscoring the gap created by rapidly developing technologies and the transitioning orientation of our IP laws.

An algorithm is a “highly definite and ordered” step-by-step set of instructions that produces an output to solve a problem, embodied in mathematics, computer programming and manufacturing processes.³ Patents embody the union of self-interest and utilitarianism.⁴ Their purpose is to encourage innovations for the benefit of society by granting inventors the exclusive right, for 20 years, to exploit their invention.⁵ In exchange, patents promote public disclosure, ensuring that the invention enters the public domain.⁶ Inventions benefit us by adding to the scientific body of knowledge.⁷

Invention or Discovery?

Whether algorithms constitute an “invention” or mere “discovery” has been viewed through the NRDC conception of “manner of manufacture,” (MOM) which historically supported a narrow view that algorithms are inherently unpatentable.⁸ However, restrictive approaches to software patents have evolved, particularly following Aristocrat II .⁹ The decision recognised the commercial exploitation of computer-implemented inventions, demonstrating a movement away from the traditional restrictive approach. In 2019, PayPal filed a patent for an MLA created to generate recommendation scores employing scalable collaborative cross-domain filtering.¹⁰ From user information, the model generated a recommendation then used cross-domain user data for the second score.¹¹ The claim was rejected, finding the invention lacked technical contribution outside ordinary computer implementation. PayPal didn’t improve the computer. This MLA was “a scheme for processing data”. ¹² An abstract rather than a concrete, tangible invention.¹³

IBM recalibrated the orthodox test of “MOM” to confront the reality of algorithm-based inventions.¹⁴ IBM had a method for significantly improving curve coordinate control points on a computer screen, using B-spline and interpolating techniques.¹⁵ To Burchett J, curve-correcting constituted an algorithm as it solved a specific arithmetic puzzle. IBM rejected the Freeman-Walter-Abele test: to (1) “recite a [Benson] algorithm”, (2) “applied in any manner to physical elements”.¹⁶ So, IBM asked instead whether the invention includes a “commercially useful effect”.¹⁷ However, algorithms themselves remained unpatentable.¹⁸

Aristocrat paved a holistic framework for determining patentability by considering “inventive and non-inventive features”, whether the computer-algorithm produced an “artificial state of affairs” and harnessed productive economic value.¹⁹ The Apex Court deemed Aristocrat’s generic “electronic gaming machine” patentable.²⁰ Aristocrat clarified that computerising an abstract idea is unpatentable. Patentability arises only when the algorithm implemented creates an artificial state of affairs and utility.²¹ Although Aristocrat proved that the commercialisation of patenting software is possible, inventions don’t extend to mere algorithms. In isolation, the algorithm is distinct from the interdependent player interface and game controller that together comprise the invention. The invention resides within the interface highlighting the integration of configurable symbols and game features, alongside their functional interaction with the machine’s operations. ²² Thus, standalone algorithms are not patentable. The principle applies to computer-implemented inventions that translate abstract ideas into tangible artificial states of affairs with valuable implementation. Accordingly, the judgment confronts the historically narrow view of “MOM,” replacing previous strict interpretations with a progressive approach to software patentability.²³

When Laws of Nature ‘Algorithm’ met ‘Patent’ in our Promethean Age:

Originally, the patent system believed that granting monopoly rights to inventors enabled society to benefit from their ingenuity.²⁴ Its underlying theory was utilitarian as inventors were encouraged to invest in developing their inventions into “public goods”.²⁵

However, algorithmic patentability raises uncertainty through the “Resilience-Unpredictability” paradox, which causes the “foresight endangerment problem”.²⁶ As the resilience of algorithms strengthens, their product’s internal system becomes opaque and output is more unpredictable.²⁷ Edmund Kitch on Prospect Theory, suggests that patent holders are incentivised to efficiently improve their scientific inventions without fear of competition.²⁸ This ignites a private motivation to enhance inventions and retain exclusive control over subsequent improvements. Still, algorithms evolve rapidly. MLAs can detect patterns imperceptible to the human mind and develop their own internal logic based on training data. Indecipherable “Black Boxes” unleash significant unpredictability in outputs.²⁹

The opacity in algorithm innovation produces a paradox at the core of innovation: increased algorithmic resilience allows its usefulness to prosper, aligning with Prospect Theory, but simultaneously renders their outputs dangerously unpredictable, reducing transparency and foresight.³⁰ Arguably, competitive innovation drives progress, as firms in a competitive marketplace innovate in order to retain consumers and maintain a technological advantage over their rivals. However, ideas themselves cannot be “overgrazed,” since their use by one party does not diminish their accessibility to others.³¹ For instance, Microsoft holds 119,196 software-related patents worldwide,³² including more than 18,000 AI patents.³³ Microsoft’s extensive control over its software technologies may impede innovation for emerging software companies, a concern reflected in ‘cumulative innovation theory, which suggests that concentrated control or excessive rivalry in innovation can produce inefficiencies in societal progress.³⁴ Merges and Nelson contend that reasonable competition, rather than monopolistic patenting, can efficiently promote innovation over time.³⁵ Under the Prospect, Competitive and Cumulative frameworks, “innovation” has consistently been aimed toward societal progress and the expansion of science and technology in the useful arts. Although increasingly accurate algorithms have aided society in innovation, uncertainty continues to pervade their deployment. Therefore, reconsidering the traditional framework of “innovation” to restrict the unpredictability of algorithms.

For instance, Google’s algorithm in ‘AlphaGo’ developed its own distinctive play style after it was trained using data compiled in millions of games played by Go professionals.³⁶ AlphaGo refined its method using the game outcomes to calculate the optimal performance then adapt to new scenarios. While the adaptive capacity of AlphaGo represents a significant form of innovative progress, it made the system’s decision-making unpredictable for human players. This unpredictability obscures the “public good” original aim of the Patent system, assuming that innovation ultimately benefits society in a transparent and comprehensible way.³⁷ Without understanding how the algorithm functions, sharing knowledge with the public is futile, undermining a key goal of the patent system. Accordingly, a degree of restraint on algorithmic innovation may be necessary to preserve the broader societal benefits that the patent system seeks to promote.

With the rise of artificial intelligence, it has become difficult to exclude a coherent characterisation of algorithms from the IP system. A risk exists with Aristocrat II diverging from the original principle in NRDC. Algorithm patentability must be clearly defined, not left to doctrinal uncertainty.

Ultimately, Bentham’s question persists: In surrendering our autonomy to algorithmic systems, are we letting our own human consciousness drift into the dreamscape of ‘economic endeavour’?

Cases

Aristocrat Technologies Australia Pty Ltd v Commissioner of Patents [2025] FCAFC 131 (‘Aristocrat II’).

CCOM v Jiejing

Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177.

Grant v Commissioner of Patents (2006) 154 FCR 62, 73.

Integra Life Sciences v Merk KgaA, F 3d 207, 1351 (Newman J) 2002.

International Business Machines Corporation v Commissioner of Patents

National Research Development Corporation v Commissioner of Patents

Legislation

Copyright Act 1968 (Cth)

Patents Act 1990 (Cth)

Footnotes

[1] Jeremy Bentham, An Introduction to the Principles of Morals and Legislation (Claredon press, 1st ed, 1789).

[2] Rose Martin, Petko Kusev, Joseph Teal, Victoria Baranova, Bruce Rigal, ‘Moral Decision Making: From Bentham to Veil of Ignorance via Perspective Taking Accessibility’ (2021) 1;11(5):66 PubMed Central.

[3] Andrew Christie, Serena Syme, ‘Patents for Algorithms in Australia’ [1998] 20(4) Sydney Law Review 517 (‘Christie’).

[4] Dan L. Burk and Mark A. Lemley, The Patent Crisis and How the Courts can solve it, (The University of Chicago Press, 2009), 66 (‘Burk & Lemley 2009 Patent Crisis’).

[5] Patents Act 1990 (Cth) ss 13(1)-(2).

[6] Australian Law Reform Commission, Genes and Ingenuity: Gene Patenting and human health (Report 99, August 2010).

[7] Integra Life Sciences v Merk KgaA, F 3d 207, 1351 (Newman J) 2002.

[8] Stephanie Constand, ‘Patently a problem? Recent Developments in Human Gene Patenting and Their Wider Ethical And practical Implications’ (2013) 13(1) QUT Law Review 100, 107.

[9] Mark Summerfield, ‘At Last—‘Software Patent’ Appeal Upheld in Australian Court’ Patentology (Web page, 31st August 2013) < https://blog.patentology.com.au/2013/08/at-last-software-patent-appeal-upheld.html> (‘Summerfield 2013 Appeal Upheld’)

[10] James Neil, Emina Besirevic, Clayton Utz ‘From algorithms to applications: unpacking Pay Pal’s AI patent refusal’ (Web Page, 24 November 2023) < https://www.claytonutz.com/insights/2023/november/from-algorithms-to-applications-unpacking-paypals-ai-patent-refusal>.

[11] See ibid.

[12] Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177.

[13] Gilbert + Tobin, ‘AI and Patents Key Considerations’ (Web Page, 19 August 2024) <https://www.gtlaw.com.au/insights/ai-and-patents-key-considerations>

[14] Christie (n 6) [3.1].

[15] David Webber, Davies Collision Cave, ‘Intellectual Property in Internet Software’ (Web Page, 20 December 1999) < https://dcc.com/news-and-insights/intellectual-property-in-internet-software/>

[16]John V. Swinson, ‘Recent Software Patent Developments in the United States’ (1994) 5(2) Journal of Law, Information and Science 281, [1].

[17] Anne Fitzgerald, Scott Philips, ‘Patentability of software in Australia: CCOM v Jiejing’ (1994) 5(2) Journal of Law, Information and Science 296, [1] (‘Fitzgerald 1994 Patentability of Software’).

[18] Christie (n 6).

[19] Mark Summerfield, ‘High Court Backs Aristocrat on Software Patentability—It’s Time For IP Australia to Follow Suit’ Patentology (Web Page, 07 February 2026) <https://blog.patentology.com.au/2026/02/high-court-backs-aristocrat-on-software.html>.

[20] Ibid.

[21] Same Mickan, James and Wells ‘High Court Backs Aristocrat: What This Decision Means for Software Patents in Australia’, James and Wells (Web Page, 12 February 2026) < https://www.jamesandwells.com/nz/aristocrat-high-court-2026-software-patent-eligibility-australia/>.

[22] James Lawrence and Paul Mahony, ‘Patentability of Computer-Implemented Inventions Post-Commissioner of Patents v Aristocrat: A Brave New World or Dystopian Future?’ (2021) Institute of Patent and Trade Mark Attorneys of Australia 30, 37.

[23] Anthony Selleck, Chloe Franken, ‘Aristocrat and Beyond: The Future of Computer-Implemented Inventions in Australia’ (Web Page, 02 October 2025) Griffith Hack < https://www.griffithhack.com/insights/aristocrat-and-beyond-the-future-of-computer-implemented-inventions-in-australia/>.

[24] Colin Bodkin, Patent Law in Australia (Thomson Reuters, 2nd ed, 2009) 74-85.

[25] Chris Dent, ‘Opposing What? Nature, Purposes and Questions of Reform of the Opposition Decision In the Patent System’ (2010) Flinders Law Journal 1, 8.

[26] Robert Donoghue, ‘Freedom under algorithms: How unpredictable and asocial management erodes free choice’ (2025) Volume 8, Frontiers Artificial Intelligence, 1, [1]-[6] (‘Donoghue 2025 Freedom’).

[27] Ibid; Hosanagar K. A human’s guide to machine intelligence: How algorithms are shaping our lives and how we stay in control (New York, Penguin Books, 2020).

[28] Burk & Lemley 2009 Patent Crisis (n 8), 69.

[29] Katarina Foss-Solbrekk, ‘Through Routes to protecting AI systems and their algorithms under IP law: The Good, the bad and the ugly’ (2021) Vol 16 Issue 3, Journal of Intellectual Property Law & Practice, 247, 255.

[30] Carl Harrap, ‘Opinion: It’s time to discussing the patentability of computer-implemented inventions in Australia’ FPA Patent Attorneys (Web Page, 06 February 2026) < https://www.fpapatents.com/news-insights/insights/opinion-its-time-to-stop-discussing-the-patentability-of-computer-implemented-inventions-in-australia/>

[31] Burk & Lemley 2009 Patent Crisis (n 8), 72-73.

[32] Microsoft Patents—Insight & Stats, Insights By Greyb, (Web Page, 2025) < https://insights.greyb.com/microsoft-patents/>.

[33] Bao Tran, AI Patent Showdown: Google Vs. Microsoft Vs Amazon – Who Holds the most? Patent PC (Web Page, Feb 15, 2026) https://patentpc.com/blog/ai-patent-showdown-google-vs-microsoft-vs-amazon-who-holds-the-most.

[34] Jacob Turner, Robot Rules Regulating Artificial Intelligence (Palgrave Macmillan, 2019) 209.

[35] ‘Burk & Lemley 2009 Patent Crisis’ (n 8), 73

[36] ‘Donoghue 2025 Freedom’ (n 61)

[37] ‘Burk & Lemley 2009 Patent Crisis’ (n 8), 74-76.