AI inventions are definitively fake. Every AFF example is AI-assisted rather than AI-invented—this leaves patent law unchanged.

Notably, while raising the questions of how patent law and policy should respond to ‘autonomously generated’ inventions, none of the reviewed policy documents provides a technical definition of such inventions. For instance, the WIPO draft issues paper states that ‘it would now seem clear that inventions can be autonomously generated by AI’.24 While no explicit reference is provided in support,25 it is worth noting that, only recently, this scenario was considered by WIPO to be ‘a science fiction’.26 The World Economic Forum white paper assumes that ‘AI is no longer “just crunching numbers” but is generating works of a sort that have historically been protected as “creative” or as requiring human ingenuity’.27 However, no technical literature but only legal sources are referenced.28

Somewhat puzzlingly, the request for comments initiated by the USPTO uses the term ‘AI inventions’ to refer to both inventions that utilise AI and inventions developed by AI.29 It is assumed that both types can comprise elements such as ‘the application of AI, the structure of the database on which the AI will be trained and will act; the training of the algorithm on the data; the algorithm itself; the results of the AI invention through an automated process’.30 (This view, however, requires further precision.31 For instance, if machine learning is applied in the process of drug discovery and development, the AI technique involved in that process would not be part of the resulting drug claimed as an invention.) Further, the document deliberates that, in both cases, a natural person can contribute to the conception of an invention, including by ‘designing the algorithm and/or weighting adaptations, structuring the data on which the algorithm runs, running the AI algorithm on the data and obtaining the results’.32 This suggests that, in the USPTO’s view, the development of an invention ‘by AI’ can still involve human input, which raises a critical question as to where to draw the line between situations where AI ‘develops’ an invention and where it is used as a tool (i.e. as a problem-solving technique).

2. Examples of ‘artificial inventions’

Legal narratives of AI-generated inventions often refer to almost the same set of examples: the Oral-B toothbrush and other accomplishments of the ‘Creativity Machine’ designed by Stephen L. Thaler,33 the NASA antenna,34 achievements in the field of genetic programming reported by John Koza,35 and AI applications in drugs discovery and development.36 More recently, the project ‘Artificial Inventor’37 presented several inventions attributed to the connectionist system DABUS:38 a method for constructing and simulating artificial neural networks,39 a food container,40 and devices and methods for attracting enhanced attention.41

None of the reviewed legal sources, however, provide a technical explanation of how the computational process was set up.42 Without such understanding, it does not appear straightforward in what way the allegedly ‘AI-generated’ inventions differ from AI-aided inventions. In this regard, it is worth mentioning that AI techniques such as EAs and ANNs have been applied in various domains of science and engineering for decades,43 including in life sciences,44 molecular modelling and drug design,45 aerospace engineering,46 mechanical engineering,47 civil engineering,48 etc.49 Yet, one does not find in these accounts the language that would refer to inventions generated by ‘autonomous entities’. Rather, they depict processes of designing computational systems and applying computational approaches as instances of computer-aided problem solving, design, and engineering.50 In contrast to legal narratives claiming that computers generate inventions ‘autonomously’,51 technical literature usually uses the term ‘automated’.52

3. Automation vs. autonomy

Automation means that a task can be carried out by a device without the human participation during the performance of a function.53 The term automation can be equally used with regard to physical labour (robotics) and cognitive phenomena and functions, such as problem solving.54 Machine learning, for instance, is defined as ‘a field of computer science that studies algorithms and techniques for automating solutions to complex problems that are hard to program using conventional programing methods’.55

Automation should not be equated with autonomy. While autonomy implies self-determination or self-rule,56 it is doubtful whether computers can, at all, be autonomous from humans and perform computation ‘on their own’. The impossibility to reproduce a self-organising system is considered to be the fundamental limitation of automation.57 For any operation to be run on a computer, it needs to be programmed58 (even in the case of self-improving software59). Put figuratively: ‘Programmers are the hand that feeds AI. It’s improbable that they’ll get bitten anytime soon.’60

Admittedly, one does come across the terms ‘autonomous’ and ‘automated’ being used interchangeably in the technical literature. However, unlike in legal scholarship, ‘autonomous’ is used (rather inaccurately) as a synonym of automation ‒ i.e. referring to processes executed without the human intervention.61 (Implications of this distinction for inventorship are further discussed in Section II.5. and Part IV of this paper.)

4. No ‘genie in the machine’

While AI is sometimes portrayed as ‘the genie in the machine’ capable of fulfilling human wishes for new inventions,62 this allegory does not correspond to reality (yet). The design of systems, which would respond to commands given in a high-level language without any instructions as to how to perform a task, has been a research aspiration in the field of genetic programming.63 Yet, researchers acknowledge that such aspiration is ‘unrealistic, at least in the foreseeable future’,64 and that ‘[t]he mere possibility of recursively self-improving software remains unproven’.65

Thus, ‘machine intelligence’ appears to be the function of designing computational systems and programming computers and, hence, an expression of human intelligence.66 More so, the fundamental limitation of computational systems to perform cognitive operations is seen in that such operations can be reproduced through computer programs to the extent, to which human designers of computational systems understand the underlying intellectual mechanisms.67