Extract
[...] To achieve this, concept proofs or pilots were developed and deployed to determine what works best in different scenarios. These works have enabled a progression of the level of accuracy of the categories selected by taxpayers themselves from 70% to 97% for requests for information falling under the upper category. On average, automatic classification has allowed for a reduction of over 24 hours in the delivery time to experts in the field. (OECD, 2023) In line with Ireland's 2021 national AI strategy, the Irish Revenue Commissioners launched a pilot project in 2018 to examine whether natural language processing (NLP) technologies based on AI could be used to offer improved customer service, reduce costs, and increase efficiency. [...]
[...] Also, rapid data analysis and decision-making support. - Challenges and risks of AI in the Geneva tax administration The integration of artificial intelligence in tax administrations involves certain potential risks, particularly with regard to data protection, transparency, and fairness. Here are some risks to consider: Data confidentiality: The use of AI involves the handling of large amounts of sensitive tax data, raising concerns about confidentiality and security. Reduction of tax employees and Task replacement: The automation of certain repetitive tax tasks by AI could lead to a reduction in demand for certain types of work, which could impact traditional jobs in the tax field. [...]
[...] Employees using AI in their personal life think that artificial intelligence has had a positive impact on their daily life. AI will alleviate their workload concerning repetitive tasks and those that take a lot of time, such as taxpayer document requests, case status, fiscal legal research, writing, project management, case analysis, and research of law and link with jurisprudence, data analysis, and information verification. 7 out of 8 questioned employees, or more than think that AI will enable the rapid and effective detection of fiscal fraud, and they also think that the processing time of tax declarations for taxpayers will be positively impacted, in the sense that the duration of processing will be less than the current duration. [...]
[...] Skatti handles around 15,000 chats per month and has so far answered around 200,000 requests with individual customer treatment. (skatteverket, 2023) Sweden wants to use AI to facilitate contact with clients, but also to prevent taxpayers from evading their tax obligations and to combat crime. In other words, it aims to secure tax money to finance our general well-being. One of the important contributions of AI is to create the necessary conditions to do more of what requires human intelligence. For Sweden, it is about strengthening human skills through AI. It's man and machine in cooperation. [...]
[...] This analysis will include a qualitative analysis supported by a web documentary analysis and a remote interview to find trends and confirm perspectives. 1. Evolution of the situation abroad Explain how AI is used in tax administrations abroad and particularly in Sweden and Ireland The evolution of fiscal intelligence in Europe has been marked by the increasing adoption of advanced technologies, including artificial intelligence to strengthen the capabilities of tax administrations. Fiscal intelligence in Europe has evolved towards increased use of AI to enhance tax fraud detection, automate tax processes, improve tax compliance, and invest in the skills and resources needed to take advantage of these technological advancements. [...]
