R&D in software and AI

What Is AI? The Department for Science, Innovation and Technology (DSIT) identifies AI as one of the five critical technologies through its UK Science and Technology Framework, describing AI as “machines that perform tasks normally performed by human intelligence, especially when the machines learn from data how to do those tasks.” The most well-known system, ChatGPT, has opened the possibilities of AI to a much wider audience and has been hailed as a technological revelation. It is a generative AI chatbot that uses natural language processing, allowing it to analyse text or speech data. Users enter prompts into the tool, and the system generates responses in the form of text, images, or even videos. The tool is constantly evolving using feedback provided by users, which in turn improves future learning and responses generated, also known as machine learning. What Is ML? Machine learning is a branch of AI that uses algorithms that process data to improve the functionality and learning of the system over time. An algorithm is “a set of mathematical instructions or rules that, especially if given to a computer, will help to calculate an answer to a problem.” This is similar to the mathematical formulations studied in school but on a much larger scale. The critical element of this is the amount and quality of data provided. The more data provided, the easier it is for the algorithm to learn, understand, and analyse trends, which results in it producing more accurate predictions or information when new data is applied. However, the range and variation of data provided are also very important to ensure the model has no built-in bias and the chance of delivering incorrect responses is reduced. What Are the R&D Guidelines for Software Development Projects? You may be wondering if software development is considered R&D and if it qualifies for R&D tax credits. HMRC’s Corporate Intangibles Research and Development Manual has specific information concerning the application of the DSIT guidelines in relation to software (CIRD81960). However, the guidelines applied to these projects are no different from those in other fields. Software is a rapidly developing field, so what would be considered qualifying is also constantly changing. There is significant information available for those who want to build an AI model or ML algorithm with a wide range of open-source AI systems and libraries that have been around for several years. Simply creating an AI model or ML algorithm does not mean a project will qualify as R&D for tax relief. We must consider if what has been achieved is an appreciable improvement on the available knowledge and technological capabilities within the field. The data from ML comes with big costs and storage requirements; however, HMRC has made a change to help R&D tax relief claimants with these costs. For accounting periods starting on or after 1 April 2023, data license and cloud computing services costs can be qualifying expenditure when employed in activities that directly contribute to the resolution of scientific or technological uncertainty. What Information Is Crucial to a Software R&D Claim? An appropriate competent professional must be involved in the project and claim preparation. HRMC has provided additional guidance about what would be expected and what evidence would confirm that this criteria has been met. When considering a software claim, we must consider the technological input regardless of the commercial output of the system. This means we must look at how the system is built and if the information as to how to do this is readily available or readily deducible to a suitably qualified competent professional in the industry. As stated by HMRC, the work will not qualify if any of the following are true: The issue is readily resolvable. The needed knowledge is readily available to a competent professional. The needed knowledge is publicly available outside your company. R&D project boundaries must be correctly identified. The R&D project will be shorter than the commercial project, and we must ensure we review each activity and stage of the commercial project against the activities which qualify for relief. AI in R&D There is still so much potential in relation to what we can possibly achieve using AI. As highlighted in a recent Forbes article, “2024 will mark a watershed moment for generative artificial intelligence, triggering a metamorphosis across the global economic landscape as businesses wake up to its broad potential.” The Research and Development Tax Credits Statistics: September 2023 saw the Information and Communication sector having the second highest number of claims and total expenditure behind the Manufacturing sector. There is no doubt that this is a trend that is set to continue moving forward.