Can You Patent AI? The Truth About Artificial Intelligence & Patents

In the first episode of Season 2 of Figure 1: an IP conversation, Albright IP Managing Director Robert Games is joined by Freddie Noble, head of Albright IP's computer science patent team, to answer one of the biggest questions facing innovators today:

Can artificial intelligence be patented?

As AI technology, machine learning tools and software-based innovations continue to transform industries, understanding how intellectual property protection applies to these developments has never been more important.

This episode explores the legal challenges surrounding AI patents, software patents and computer-implemented inventions, including the recent UK Supreme Court decision in Emotional Perception AI and what it could mean for businesses, developers, founders and technology companies.

The discussion covers:

• Can AI inventions qualify for patent protection?
• Why software and computer programs face unique patentability challenges
• What does patent law mean by a "technical effect"
• Examples of patentable software innovations
• Why some AI and machine learning inventions succeed while others fail
• The Emotional Perception AI Supreme Court case explained 
• How the UK Intellectual Property Office assesses software patent applications
• The differences between UK and European approaches to software patents
• What founders, developers and businesses should consider before filing a patent application

Whether you're building AI products, developing software, investing in technology, or simply interested in intellectual property, this episode provides practical insight into one of the fastest-moving areas of patent law.

Figure 1: an IP conversation brings together Albright IP's patent and trade mark attorneys to discuss intellectual property, innovation, business growth and the legal issues shaping modern technology.

Figure 1: an IP conversation. Real insights. Real strategy. The protection your business needs.

Subscribe for more episodes covering:
• Artificial Intelligence (AI)
• Patents
• Trade Marks
• Software Protection
• Intellectual Property
• Innovation
• Technology Law
• Start-ups and SMEs

Want advice on protecting your AI or software innovation?

Visit Albright IP:
 www.albright-ip.co.uk

Transcript

Intro

Robert Games 0:00

There's been an important decision from the Supreme Court, hasn't there?

SPEAKER_01 0:03

Yeah, absolutely. So the Supreme Court gave their judgment in this case called emotional perception AI.

Robert Games 0:09

Uh is that going to make it easier for people to get patents for software?

SPEAKER_01 0:13

We get software patents in a sense all the time. Where the, you know, the invention, the new bit, is in software. And there are big classes of things where there's no particular problem with that.

Robert Games 0:23

So you're using the software to control it, basically.

SPEAKER_01 0:26

Yeah, that's right.

Robert Games 0:27

So what actually happened to emotional perceptions application? Did they get their patent?

SPEAKER_01 0:31

I do think they've got potentially a clever and significant idea. But my personal opinion, I I think they probably will still struggle. If you've got a technical solution to a technical problem, then you should be able to get a patent.

Robert Games 0:43

Hello and welcome. I'm Robert Games, Managing Director of Albright IP. Welcome to our podcast, Figure One and IP Conversation. Today I'm joined by Freddie Noble. Amongst other things, Freddie has a first class honours degree in computer science and engineering from Jesus College at Oxford University and leads our patent team in the area of computer science. AI is on the tip of everyone's lips at the moment, although the technology has been in development for a long time. IBM's Deep Blue defeated the chess champion Gary Kasparov in 1997, showing that a supercomputer could beat the chess world champion. Since then, we've come a long way, and everyone has access to an AI search engine. So, Freddie, can I patent a new idea in the field of artificial intelligence?

Can you patent an idea in field of artificial Intelligence?

SPEAKER_01 1:30

Well, maybe. It depends what the idea is. And I suppose that's an unhelpful answer, but uh well, why not? Why should there be a problem with protecting a new idea in the field of artificial intelligence? We can protect new mechanical inventions, um, we can protect new chemical inventions. So why should um AI or software more generally be any different? Well, the the catch is with software that there is an exclusion in the Patents Act and in the European Patent Convention that that says the following things are not inventions, and there's a list of four or five things actually, but the one that we care about today is a scheme rule or method for performing a mental act or doing business or a program for a computer. So that program for a computer exception is obviously highly relevant here. Sure. Um now the act does go on to say that the thing is only excluded, so the an invention is only excluded to the extent that it relates to that thing, so a computer program or a business method or whatever, as such. So that certainly seems to leave the door open, but I suppose the question is it wide open or is it ajar? Um it the door is somewhat open to inventions that use computers and use software. And we have had over the years quite a lot of cases, and what I think they've all boiled down to is um is there a technical effect? Um, so you know, does the computer software have a technical effect on some technical process?

Robert Games 3:15

So we know that patents are for technical inventions, right? But but what does that really mean? I mean, everything with computers is technical in a sense, isn't it?

What counts as a "technical" invention in patent law?

SPEAKER_01 3:25

Yeah, that's right. It's all very well me saying um, well, it's patentable if it's a technical invention with a technical effect, but it it all depends on what you mean by technical, doesn't it? Yeah. This can be a difficult problem. And it's not just me saying that. There was a case in 2012 called uh Really Virtual Co-Limited, where the judge said um that actually this whole business of whether something's technical is not readily amenable to systematic analysis, is based on an impression, um uh uh given the the cases and kind of drawing analogies. And your you know your previously decided cases are never going to be exactly the same as the invention you've got right there in front of you, are they? Because we're always talking about a new invention. Sure. So it it can be it can be really hard, but nonetheless, actually, there are a lot of I think quite clear cases, and the we get software patents in a sense all the time, where the you know, the invention, the new bit is in software. And there are big classes of things where there's no particular problem with that. So, one example is where you've got some control of a process, you know, it might be a production process, control of a machine, something like that. So you've got an invention in a technical field, which is not the technical field of computers. Um, you know, say you're controlling a furnace or something like that, you've found a you've changed the software in the controller to make it burn more efficiently. Well, really, the invention there is in the technical field of running a furnace. So, although the the the old furnace is exactly the same as the new furnace in terms of its hardware, and the only thing you may have changed is the software, actually, the technical invention there isn't a computer program, it's a way of controlling the furnace, which is implemented in software, as many things nowadays are. So there should be no problem with things like that.

Robert Games 5:22

Okay, so so that that makes sense. Um, if you've invented a better moulding machine or production line or furnace or whatever it is, it doesn't really matter that you've implemented the change in the software. The invention is really about the plastics moulding or the furnace monitoring, a clearly technical thing.

SPEAKER_01 5:41

Yeah, absolutely. Um the and there should be no problem with those kind of inventions. And in fact, I think a lot of the time we don't even realize we're writing software patents um when in a sense we are. Um so there should be no problem there. Um I I worked on a a stack of cases that were to do with um traffic light control, for example. And you know, this is a problem that everyone's experienced. Um, if you've driven in a busy city, um can get quite frustrating, can't it? You sat and there's a red light in front of you, and it it changes green, and two or three cars go and then it goes red again. And you know, everyone complains about whichever idiot it was who set the traffic light timings, but in fairness, they're probably not an idiot, it's just that it is quite a difficult problem um to get that just right. And um, you know, I worked uh with a client who was doing um sort of machine learning to try and build better control algorithms, and of course, all the hardware in those traffic lights are the same. I mean, they've still got a red light bulb and an ample light bulb and a green light bulb, they've still got the sensors um to see who's queuing up, um, but it's the control system, um, and they were using machine learning to try and improve that. And there is, you know, new software there, but a technical effect in the real world, hopefully. Um, you know, better traffic flow through the city.

Robert Games 7:01

So these are an invention, these are inventions, sorry, where that that there's a real world effect, something in the real world that you're actually controlling. So you're using the software to control it, basically.

SPEAKER_01 7:12

Absolutely. And and and that's an invention in a technical field of of running a furnace or or controlling traffic or or whatever. But then that that's not the only kind of software invention that you can get. There's another sort of big class of things um where there's a kind of internal technical effect and it in a technical effect inside the computer. Um, and and I've done uh again myself uh a number of applications around the themes of things like backup and recovery software. Um, you want to make your backups, obviously, in case uh a hard drive or uh component in a computer fails, um you want to have a backup that you can recover from. There's then a question of will the backup work? How do you set up recovery possibly on different hardware? Um, and this is you know a problem with the whole business of running computers. And where we've got inventions that solve problems that really span across anything you might do with the computer, it doesn't matter whether you're doing your accounts or playing games or drafting patent applications, whatever you're doing with the computer, it makes it a better computer if backup and recovery is managed better. And and so in those sorts of things, we have an example of a kind of internal technical effect, and those don't tend to present a problem either. I've you know I've done many of those applications, and I we've we've never had to argue that they're outside the scope of the exclusion because an examiner has simply never raised it. Sure.

Robert Games 8:44

So we have a few examples where you shouldn't hit a problem. Um, there are presumably some other areas um where

Inventions that can't be patented: no technical effect

Robert Games 8:52

there are clearly no technical effects, inventions which basically have no hope of getting a patent.

SPEAKER_01 8:57

Yeah, well, and in fact, according to the patents that they're they're not inventions under the definition. Okay. Um well, ooh, that might be contentious actually. But yeah, yes, there there are some things where um where there's basically no prospects of success. And we we obviously do get a lot of inquiries. People ring us up and say, Um, you know, I've got an idea for a new app. Um, and a new app may be participant, it depends what it does, because it may, you know, it may have an external or an internal technical effect. But quite a lot of the time, what people have really got is is an idea for a new business, uh a new model for for doing business. And of course, a lot of the sort of apps that have really flown are really kind of a new way of doing business, and of course, that's implemented using software, but the software is really doing its ordinary job as application software, it's doing things that are you know technically routine, um, and it may well be novel and non-obvious, at least in the sense that um, you know, the business model or what it's actually doing is is new and inventive. Um, but it it is going to cut across probably the computer program exclusion um and maybe the business method exclusion as well.

Robert Games 10:23

So, Freddie,

Could the Supreme Court open the door to more software patents?

Robert Games 10:24

there's there's been an important decision from the Supreme Court, hasn't there? Uh, is that going to make it easier for people to get patents for software?

SPEAKER_01 10:31

Yeah, absolutely. So the Supreme Court gave their judgment um in this case called emotional perception AI. That was just last month. And it it is a significant decision for sure. Um, but the effect, I think, in terms of which things are eligible for patent protection and which things aren't, I think could be quite subtle. So, what what the Supreme Court have said in emotional perception is this test that we've been using effectively for the past 20 years, it's called the Aratel test after a case that was in the Court of Appeal um well, 20 years ago now, um, is wrong. So that so that test that the courts and the patent office have been using, they were going through these four steps um to try and determine whether something fell within that computer program as such, exclusion or not, um, is is no longer applicable. And so that's that's big news, really, because it's a framework that the examiners at the patent office have used in pretty much every case um that that might cut across this exclusion to work out whether it does for a long time. So that's gone out of the window now. And they've given us an interpretation of the law, which actually isn't a new interpretation at all because the EPO have been um uh have been doing it for a long time, and the Supreme Court have have kind of said, Well, why are we different? Um the EPO isn't wrong, um, and and we should follow their the the very logical way that they have set things out. And to cut a reasonably long story short, um what this this I call it a new test, but it's the the EPO's test that they've been doing for a long time. What it says is well we look at the claim, and if there's any hardware at all, any technical thing whatsoever in the claim, then then the whole invention isn't excluded. So it's much much simpler test than this this previous four-step test to apply, and and that's where we are now. That sounds like quite a big change. Um, but presumably that's not the end of the story. Yeah, so it sounds like a massive change. Um, you know, I've just but I've just said it's gonna be quite subtle. So given that when you look at the claim, if there's any hardware feature at all, it's not excluded as such. Well, why isn't that a massive change? Why isn't that opening the floodgates to you know protecting everything and everything as long as you've you know if you've got a keyboard and mouse on your claim, it's all right, right? You say the software's new. Yeah, but I think the catch is this we still have to assess novelty and inventive step. And we have to make sure when we're assessing novelty and inventive step that we're making that an assessment on an invention. And remember, the law says that an invention is not a computer program as such. So, what we have to do under this new or EPO-inspired test is strip out any bits of the claim that don't contribute to a technical effect. So just because there's a thing in the claim, it doesn't mean that it's relevant to the assessment of novelty and inventive step. We actually have to look at what technical effect the claim has, and then look at only the bits of the claim that contribute to that technical effect in order to work out whether it's novel and inventive. So this difficult question of what is technical is still there. Um, software features can count as technical sometimes, hardware features can count as well, but only if they contribute to a technical effect. And in some extreme cases, you know, if you've got no technical effect at all, then you might be stripping out everything in your claim apart from the computer. Yeah. Now that that's okay. A computer's not excluded, but of course it's not new either. Um so I I think this is why, although we seem to have a very dramatic change to how these inventions are assessed, I think at least those clear cases that I've talked about will stay probably on the side of the line that they're on. And those cases, certainly, where there's you know a new app that basically implements a new business method, I think they're still going to be rejected. It might be that the heading on the rejection says novelty or inventive step and not patentable subject matter, um, but you're still not going to get a patent for them.

Robert Games 15:01

Sure. A lot of people commentators are saying that things will get a bit easier with this new decision. So so do you do you think they're right?

SPEAKER_01 15:09

Well, uh I mean time will tell, I suppose. I I tend more to well, the effects will be subtle. I don't say non-existent, I say more subtle. Um I I think those clear cases that I've talked about, those clear cases that are on the were on the right side of the line before, that will stay on the right side of the line. And I think those clear cases I told you that were on the wrong side of the line before, I think they will stay on the wrong side of the line. There are, of course, some difficult borderline cases where reasonable attorneys might disagree um as to uh whether they are patentable or not. You know, it might be that emotional perception's actual application might even be an example of one of those difficult borderline cases.

Robert Games 15:50

So, what

What happened to the Emotional Perception patent application?

Robert Games 15:51

actually happened to emotional perceptions application?

SPEAKER_01 15:53

Did they get their patent? Uh well, we don't know yet. Um so so what the Supreme Court have said is well, the patent office's rejection was was wrong because it was based on this Eritel test, which we now know is is wrong. Um But they they haven't the Supreme Court haven't reassessed that invention in the light of the tests they've now got to follow. They've effectively sent it back to the UK IPO to say, well, now you've got to make the decision again in the right way. Now now we've told you what the law means, you need to go and assess that invention against the requirements of the law properly interpreted. So it's back to the IPO. Um, and who knows, it could be appealed again. Um, I don't know, maybe even as far as the Supreme Court. Um, I mean, I I've I've read their claim, I've looked at their application. Funnily enough, I I do think it's potentially quite a clever system, and and I haven't heard anyone say it's not even novel, which sometimes happens with these cases. Yeah. Um I I do think they've got um potentially a clever and significant idea, and it's about um you know recommendations on on platforms like Spotify, for example. So it could be really commercially valuable. You know, there's a lot of there's a lot of money in getting people to watch and listen to things now, isn't there? So uh but but my personal opinion, I I think they probably will still struggle, and the reason I say that is I think fundamentally what they've developed is is probably a different way of you know organizing information, which the case is both at the EPO and the UK um tend to see as not technical. But I mean I don't know, I can see some arguments on the other side as well. Um, they are as part of the steps of their method, they're analysing audiophiles, and uh there may be some technical character there. Um, but but we'll we'll have to see.

Robert Games 17:56

Uh sounds like if you were a betting man though, you were you you probably put your money on the decision still not ending up in a patent for emotional perception.

SPEAKER_01 18:04

I I think I would, yes. Although I I wouldn't bet with money I can't afford to lose. That's that's always good advice, anyway, of course.

What impact will this decision really have?

Robert Games 18:11

So so you know what what difference do you think this new decision is really going to make then?

SPEAKER_01 18:16

Yeah, so I mean let's be clear about it. I I I do think it's a good decision. I agree with it completely. Okay. Um I do say I think it's the effect is going to be quite subtle potentially. But I I think there are some positive points. I think firstly, this this new understanding of the law, um not new understanding, the EPR have known about it for years, but I I think this new to UK courts understanding of the law is perhaps a little bit easier to apply and um may make things more predictable. Um it doesn't remove the difficult question of what is technical, but I think what the Aertel test was doing is is kind of putting a massive distraction in the way um of another argument over what was the contribution, and that was making the the difficult question of is it technical perhaps harder than it needed to be. Um, so we're still faced with that question. Um, but I I think we might just have a pathway that's a little bit easier to apply. I think there's still going to be blurriness at the borderline, but maybe that's got a little bit sharper as a result of um getting rid of this Aerotel um test. I think you know we're still going to get clear cases on one side or the other. I think we're still going to get clients come to us and we sort of scratch our heads and say, well, maybe that's patentable. Um and we could give it a go. Um, but perhaps it will be a bit easier to understand and explain the risks that are there. Um, so I I do think it's positive in those ways. Um there is already an indication that the UK IPO are perhaps going to stop refusing to search some of these uh applications so readily. So we used to put applications in, and if the UK IPO thought they were excluded, they quite often refused to carry out a search. Um that was that could be quite frustrating. Um and oddly enough, I I'm not really sure that the difference between the Aerotel test and the new sort of any hardware approach. I'm not sure there's any particular logical reason it should change that practice, but I think that practice is changing. We've already, even you know, a month after the decision seen some indications that that practice is changing, and a lot of people uh are sort of um assuming that that will change, so that's quite positive. And then I I suppose because the analysis is going to move from patentable subject matter as a sort of standalone point to an assessment of inventive step. I I suppose some people are thinking, well, you know, novelty and inventive step is our bread and butter. We have to deal with that if we're you know doing a mechanical invention, invention about a new kitchen gadget or a toy or whatever, and we have to do that if we're doing a chemical invention or any other type of invention. So it's kind of our bread and butter, there's nothing special anymore. And and normally with inventive step, you do get a certain dose of the benefit of the doubt at the application stage. You know, if you can put forward a reasonable sounding argument, the patent office will will tend to sort of say, Well, well, okay, I'll take your word for it. Um, is always open for someone to challenge after grant, of course.

Robert Games 21:52

Sure. So if you get over that hurdle of a technical advantage, then you stand a chance of getting home, really.

SPEAKER_01 21:57

Yeah, although I I I do think I think that you know, where there's doubt over the technical effect, um, the UK IPO, I think, are probably still going to be quite robust there. Um, you know, it is a a matter of law, I think, whether there's a technical effect. And if they don't think there is one, my prediction is that it it you know they're going to stick to their guns unless you've got some proper arguments that persuade them.

Robert Games 22:22

Sure. So it's all about just like for all patents, really, a technical solution to a technical problem.

SPEAKER_01 22:29

Yeah, absolutely. If you've got a technical solution to a technical problem, um, then you should be able to get a patent. Um, there are still going to be some difficult cases. Those difficult cases will be where the technical effect is is controversial, i.e., you know, you can disagree as to whether there's a technical effect or whether the effect counts as technical. Um, but I I think there is a a bit of a rule of thumb, and again, I've been saying this for years, I don't think emotional perception really changes this. When you've sort of wondering if if your effect counts as technical, is to say, well, is a technical person going to be impressed by it? Because, you know, when people talk about their inventions, they often very naturally point to, you know, what advantages is it going to bring, you know, to the end user? Um, what what features are we going to sell? What are we going to put in our you know, billboard adverts for our thing does this that you couldn't do before? And that's all very well because that you know that's what the customer wants at the end of the day. Um, but a feature being particularly useful or um doesn't mean it's technical. And I think what you have to look at is would a technical person be impressed by a sort of technical step um that you've that

Outro

SPEAKER_01 23:52

you've made? And if the answer to that question is yes, then then maybe we're on something. It is a rule of thumb and not a legal test, but I think it can be quite a valuable.

Robert Games 24:01

No, it sounds good. So if if you are a technical person, impressed by a piece of technical software development where there seems to be a technical advantage, it's well worth getting in touch with Freddie and the team to consider whether you may have a patentable invention. Uh thank you, Freddie, and thank you to our listeners for joining us on Figure One and IP conversation brought to you today by Albright IP. Thank you.