Identifying eligible clients in AI and AR

Something we hear time and again from our clients is that they struggle to identify clients in their client base who would be eligible to claim R&D tax relief. So, in this blog series, we’ll be digging into some interesting sectors and discussing what to look for when assessing eligibility!

In the latest of this series, we’re looking at Artificial Intelligence and Augmented Reality. As with our previous subjects, the level of eligible R&D to be found in this area is misunderstood, and very often overestimated. However, if you know what you’re looking for this can be a very fruitful area to look for eligible clients.

A few definitions to start…

Before we get into what to look for in this area, a few definitions are in order. The terms artificial intelligence (AI), machine learning (ML), virtual reality (VR) and augmented reality (AR) are thrown around a lot in the context of cutting-edge technology, but it’s often unclear what they mean!

• Artificial Intelligence (AI): giving a computer or a robot controlled by a computer the ability to carry out tasks that would normally require human intelligence.
• Machine Learning (ML): A type of artificial intelligence where a computer is taught to predict outcomes without being explicitly programmed to do this. It usually involves ‘teaching’ the program through the analysis of vast amounts of data.
• Augmented Reality (AR): an experience of a real-world environment that has been augmented by computer generated perceptual information.
• Virtual Reality (VR): a computer-generated reality that can be interacted with in a seemingly real way.

Now that we’ve got that out of the way, let’s dive into the potential eligible areas within these kinds of work!

What to avoid

As with previous subjects, it’s worth taking the time to first think about the types of AR or AI work that isn’t eligible. Although these are highly technical and cutting-edge areas, there are several areas to steer clear of when preparing your client’s claim.

Firstly, all work done in the areas of AI, and especially AR, will involve elements of aesthetic design. While the projects will depend on this design work to be successful, HMRC are very clear that this kind of purely aesthetic work is not eligible and must be excluded from the claim. 

Over the past few years, big players in software development, such as Google, have made AI libraries publicly available. These libraries can be used by anyone with even a basic understanding of software development to (relatively!) easily and quickly add AI elements to other software. These libraries are well-tested and documented, so any project that involves taking them and using them as intended would not be eligible for R&D tax relief. 

Lastly, remember that the use of true AI is less widespread than the claims of software companies would have you believe! Often, companies that claim to use AI are simply creating algorithms and datasets that can give the ‘feel’ of AI, but can be achieved using established techniques. While this work may have some eligible elements, it’s important to remember that not all AI work is eligible, and not all work labelled as AI is even AI!

Ok, so what should I look for?

So, now that we know what ineligible AI and AR R&D work looks like, what areas should you be focussing on?

Artificial intelligence

Much of the actual AI development work currently being carried out will have some elements of eligibility, so assuming that you’ve discarded any aesthetic work—and anything achieved through the use of standard components and techniques—you should be fairly safe to make a claim. That said, there are a couple of areas within this sector that it’s worth focussing on in particular—security and computing power. 

Computing power

The amount of computing power required to help a computer ‘learn’ through machine learning and AI is immense. As super computers are replaced by cloud-based services, companies carrying out this work have to devote more and more time to either working out how to increase the performance of these machines, ensure that cloud connections are robust enough to handle the computing needs, and improve their algorithms to reduce the amount of power needed for each operation. All of this work will require advances to be made in software science, and should be eligible for R&D tax relief. 

Cyber-security

AI is often used to improve or enable automation of systems, from production lines all the way to automating the control of ocean-going vessels. This can reduce costs and increase efficiency but does leave the systems vulnerable to cyber-attacks and potential takeover by hostile groups. Companies working in these areas are therefore very likely to devote huge amounts of time and money to security related R&D. Given the need to remain one step ahead of the hackers and their techniques, work done in this area is almost always eligible. 

Augmented reality

In common with AI work, work done in the areas of VR and AR is often highly eligible, assuming that all aesthetic design elements have been excluded. As long as you’re happy that the work goes above and beyond what was possible at the outset of the project, and advances software science in some way, the claim should be eligible. There are a couple of hot topics in this area that it’s worth digging into—integration with disparate hardware, and data stabilisation.

Hardware integrations

AR and VR must be served to users through some kind of hardware, be that mobile phones, VR headsets or multiple wall screens. All this hardware will have been designed to work in a certain way and give a certain level of performance, so the challenge for AR/VR developers is to integrate newly developed software with this existing hardware in a way that is stable and performant. As long as this work focuses on using hardware in a way it was not designed to behave, or extending the performance of this hardware, you should be able to make a claim. 

Data stabilisation

In both AR and VR, the smoothness of the imagery is of paramount importance to the success of the created environment. The stabilisation of the data required to create the environment is therefore a huge focus of the R&D efforts in this area, with developers working to ensure that massive amounts of data are served at the right time and in the right way to give a completely seamless experience to the user. Again, as long as companies are working to advance software science and push the boundaries of what was previously possible in this area, this kind of work should qualify for R&D tax relief. 

A word about data costs

In AI, and especially ML, vast amounts of data are required to ‘teach’ the software, which usually has to be purchased by the clamant company for use in their R&D projects. Unfortunately, up until now, data costs have not been allowable for R&D tax relief, to the detriment of companies carrying out legitimate R&D in this area. Happily, in April 2023, legislative changes will come into effect that will enable companies to claim for all data costs related to R&D. Any claims being submitted before this date will have to exclude these costs, so make sure you know what data costs have been incurred by your clients and when they were, or will be, incurred in the run up to this date.

We’ll be covering this topic in detail in our upcoming webinar ‘Identifying R&D Eligibility in Artificial Intelligence and Augmented Reality’ on Tuesday, 24th May at 12:30–13:15.

Tickets are free—book at the button below!

Claim your spot here