13 Jan ROBOTS AND THE FUTURE OF TAX
By Miguel Vasquez
For decades, the rhetoric that automation will be the precursor to the downfall of humankind has been a central theme of dystopian cinema. As such, in December, 2022, the alarm bells of the belligerent anti-automation populace would have certainly rang, as ChatGPT, a prototype dialogue-based artificial intelligence (‘AI’) chatbot, capable of understanding natural human language and generating detailed responses, was released for public use and interaction. The news surrounding ChatGPT was closely followed by another AI chatbot, DoNotPay’s “robolawyer” (the ‘RoboLawyer’), which is scheduled to provide real-time support in court hearing, with no human input.
ChatGPT and RoboLawyer’s recent attention has resurfaced areas of debate regarding the role of AI and robots in shaping our ever-evolving society, and whether (and indeed how), robots should be taxed.
Over the past century, automation and technological development have been the proponents of significant improvements in both the quality and continuation of human life. From the use of robots in hazardous and dangerous environments to reduce risk to human life, in medicine and in the military, to simple uses in lifting heavy loads and the completion of repetitive monotonous tasks with a consistency and efficiency impossible to be matched by humans, there is little room to argue against technology’s role in exponentially changing the trajectory of human development in modern times.
AI however, is considered as being a step too far, due to the risk that it may ultimately replace humans. For the most part, AI’s role has remained limited to the organisation and connection of data, and being highly task focused. However, ChatGPT has added an entirely new layer to AI’s role, at least to the public, in offering solutions to problems requiring not only an extensive knowledge base, but critical analysis and problem solving. While lawyers have been using forms of automation in their legal services for a considerable period of time, RoboLawyer’s potential use in a courtroom adds an even further practical application of AI to the real world. It purports to replace a human lawyer during a court hearing (albeit in a traffic-related matter), with the robot being the sole source of legal guidance before the judge.
ChatGPT and RoboLawyer are not the only AI systems that have been involved in ground-breaking developments, but rather, there are scores of achievements made by AI in recent times. For example, in late 2021, an AI system in South Africa has been listed as an inventor and granted a patent for designing and creating an invention of technology.
However, for all the hype around the unprecedented use of AI in a court hearing and its potential to change the provision of legal services, this does not mean that AI can necessarily and automatically replace professionals and lawyers. In fact, AI and machine learning still require considerable development before it can think, interpret, write and function like a human.
Before even reaching that stage, significant strides are required to ensure that the information generated is accurate. Indeed, users of both ChatGPT and RoboLawyer have reported numerous instances of incorrect information being provided in response to questions, ranging from inaccurate representations of the law, to wrong tax rates being provided.
Academia has also expressed concerns with the potential use of AI as it may encourage plagiarism and other dishonest practices by students, who may simply request answers from AI systems under the guise of being their own work.
While research and development departments tackle these issues with AI and automation, policymakers are tasked with determining how to confront the problem of robots replacing humans in the workforce. With respect to tax policies at least, which often shift and evolve to address societal and economic changes, policymakers are required to determine whether to slow this progression, or how to capitalize on it.
On the one hand, there is a concern that increased automation is likely to cause large scale unemployment, which can lead to a reduced personal income tax base, and ultimately less revenue generated for a country. While the true magnitude of a reduction in revenue may be unknown at this stage, the potential to destabilise the economic and social welfare within a country exists. Having regard to this risk, discussions surrounding taxing robots have arisen.
The rationale for taxing robots is that although humans and machines may perform the same role or task, tax laws apply differently to both. This however, creates the risk that companies engage in automation (while not necessarily improving efficiency) for the sole object of reducing expenditure on salaries, social security contributions (such as national insurance contributions) and the benefits necessary to retain a human workforce. Further, the use of tax policies aimed at robots is likely to perpetuate scepticism and general distrust of robots, by maintaining an artificial separation between the two, rather than a recognition that robots are to add value and support humankind.
In any event, if robots were to be taxed, what the most appropriate measure to effect the tax is, remains unknown. As indicated above, one of the main proponents for taxing robots is the reduced tax base by the replacement of humans with robots. The first hurdle to surmount is that robots have no legal capacity; as such, any likely tax would be on the use of robots. Practically, this means that the entity utilising robots will be doubly taxed; first on its corporation tax, and second, by way of the robot use tax. From an international tax perspective, the introduction of a robot tax, in the absence of an international consensus, is likely to reduce competitiveness for that jurisdiction’s residents, and to attract foreign investment.
The second hurdle is that a robot tax is likely to disincentivize further research and development of robotics and AI for use in areas that add value to humans, which would have damaging effects on certain critical areas such as medicine, engineering and the military.
Third, although automation may have been responsible for increased levels of employment in the past, this has historically been in the short term. In the long term, productivity has increased, and innovative service areas and lines have arisen.
Ironically, tax administrations across the world have been using AI in their systems for tax compliance for a considerable amount of time, whether through filing returns electronically, or using systems to input and receive information and records. If a robot tax was to be imposed, a potential middle ground may be that the revenue gained from such a tax be directly invested to fund programmes for unemployed persons and to retrain them for different roles that have not been rendered redundant by robots and AI.
For all the benefits that robotics and AI can provide however, it would still be unwise to wholeheartedly trust automated systems that continue to require improvement, as, inherent in improvement, is the recognition that what previously existed, was some semblance of imperfection.