Protecting your invention

0:00Welcome to Tech Life on the BBC World Service, the programme about technology making an impact on all our lives. I'm Zoe Kleinman and this week we're learning all about patents, what they are and who wants one and how by analysing the global demand for patents we can spot future trends in tech. That's in a moment. What are doctors and patients around the world saying about the growth of AI use in hospitals? And we'll be finding out about Google's latest injection of artificial intelligence into online search.

0:51We start this week by looking at something we don't talk about much on TechLife. And by this, I mean patents. They're an important part of the innovation process because once something's been invented, the inventor often wants to protect it from being copied. there could be patents for all sorts of things.

1:19Did you guess what they were? I think I heard an old dot matrix printer, a vibrating smartphone and a microwave. Well, with many thousands of patent applications made for all sorts of technological developments, it's possible to identify trends and see how the world of tech is evolving. I had a really interesting chat with Steve Rowan from the European Patent Office, which covers 39 member states and received hundreds of thousands of applications from all over the world. Steve is Vice President of Patent Granting Processes. I started by asking him to explain the patent and why anyone would need one.

1:58A patent is basically a time-limited monopoly of 20 years. it's granted by a state so it's it's only valid in that particular state which is granted the patent and it's granted to a company or an individual who's developed something which is new and inventive and a solution to a technical problem and that technical problem can be very broad so it could be in agriculture it could be in health it could be in energy it could be in mobility It's basically about making our lives better. And really the purpose of them, they were invented a very long time ago in the 1300s. And the purpose is really to encourage and reward the time and money that's invested in coming up with these solutions.

2:45And society gets something back. So society gets the new idea, the new solution brought to market. But also the inventor has to fully disclose the invention in the patent application. And that means that when the patent expires after 20 years or if they don't renew before that, anyone can then reproduce that invention and also use it to make new ideas and to come up with new solutions. But surely everybody thinks they have a new idea. And the reality is that lots of people don't. Yeah. And that's the purpose, really, of the patent offices of the world. What we do at the patent office is we take the idea that someone has come up with, the written format of their solution to the problem, and then we do a very in-depth search.

3:30So we search through over 2.3 billion documents to see if anyone anywhere in the world has ever come up with that idea. So is it new? And then we also look at whether it's inventive. And what does that mean? Inventive means it's not obvious to somebody who is skilled in that particular area. Someone wouldn't go, oh, yes, of course, we would do it that way. They would go, wow, OK, that's interesting. I wouldn't have thought of that. And that's the role then of the patent offices. We check, we search and we then examine it against the criteria, as I said, for novelty and inventive step. And if all of those criteria are met, we then grant the patent to the applicant.

4:08What's the success rate? How many people fall at that hurdle? It's a really good question. what we see, of course, is that as a part of the process, the first thing we do is we do a search and what we call a written opinion. So we do a search of that database I mentioned, and then we look at whether or not the idea could be patentable. On a first application, we see around 21 % of people dropping out of that phase. So they see that their invention isn't that new, or it doesn't involve the inventive step that's necessary. Overall, we see around 60 % of the applications that we get proceeding to a granted patent.

4:44But nearly all of those are in an amended form. If we look at the statistics, so our search reports and written opinions, 93 % of them are negative. So in 93 % of cases, we find some element that is not patentable under the law. Of course, people can then amend. And so overall, then we see around 60 % of those applications proceeding to a granted patent. You track trends in patent applications. I'm going to take a guess that there's quite a lot of patent applications on your desk right now, which are AI themed, perhaps? Yep, definitely. I think if you look over the last sort of 20 years and growth in the fields of technology, the computer technologies have really, really taken off.

5:30And of course, more recently then, AI-related technologies have also taken off. And we see that growth year on year. If we look at last year, for example, there was a 3.3 % growth in the area of computer technologies. But the other element that we see, of course, again, over the last sort of 20 years, is that before, if you were looking at, say, an agricultural device for planting seeds, Years ago, that might be a device which would bore the right depth of hole and then would plant the seeds at the right distance. Now, of course, even something like that includes AI technology. Using GPS, etc., the farmer can actually completely identify where the seed has been planted and also optimize that planting so that we get the best food production from the crop.

6:19So AI is basically now everywhere. Computer implemented inventions are also in all fields of technology. And who is filing these patents? Is it individuals? Is it the kind of inventor in their shed? Or is it research organisations, big companies, big business? That's an easy question to answer because it's all of the above. Of course, we see a prevalence at a higher number of applications from the larger companies. They're the ones, if you look at what's needed to get a patent, you need a lot of research and development. A lot of ideas will fail. But we see a high number also from universities.

6:58Increasingly, universities are creating that patent technology and then spinning out companies and also from SMEs as well. So we see the whole range of people filing patents. But as I said, mostly it's coming from the larger corporations with their sort of R &D investment. And are you seeing the numbers of patents that you receive going up or down? In general, there's a global trend towards increasing numbers of patent applications year on year. What we see is that patents are actually quite closely tracked with regard to what's going on in the world. So, for example, if there's a bit of an economic crisis, companies pull back a little bit on their research and development.

7:40And what we see is that if there are other global crises such as wars, etc., we also see a slight pullback. but overall the global trend over the last 20-30 years has been an ever-increasing number of patent applications and an increasing number of countries filing patents. So for example, huge growth from China. But we also see that last year, for example, we see a slight stabilisation in our own filings. So we've had around 200 ,000 applications last year. That's a lot. And that's a very similar number. It's a lot. It's a huge number. If you look at China, they're at about 1.6 million. So that's an even bigger number.

8:14That puts it in perspective, doesn't it? But 200 ,000 is a record for the European patent office. Last year, we stabilised. If you look at it in detail, we're about 0.1 % down. So it's a high number. It's where we've been in the last couple of years. We saw growth, for example, from 23 to 24. And then in 24, we saw a slight stabilisation. We obviously track this very carefully with regard to our workload. So we try and have an idea of what's going to come in through the door. Do you think the world is starting to run out of ideas then? No, no. There was a very famous head of the USPTO who, I think at the start of the 1900s, said everything that's been invented has been invented.

8:55That's clearly not the case. What we see, of course, is the differences, some technologies pulling back. So, for example, the internal combustion engine, we see a fall off in applications there. But we see huge growth in electric motors and particularly in batteries. If I just look at the battery area, four-fold increase in the last five years in the number of applications coming in in battery and fuel cell technology. Oh, good. I'm pleased because we need better batteries. Exactly. And it's not just better batteries, it's also recycling. So if we just look at batteries, we see increased numbers, obviously more effective batteries, more energy efficient batteries, but also ones that can be recycled, different ways of making them part of the vehicle, different ways of attaching them.

9:39All of those things are really important to solve the problem of everyone wants to be able to move around, everyone wants to have access to clean technology and clean energy. So we need better batteries. That's Steve Rowan from the European Patent Office speaking to me from the Netherlands.

10:06you're listening to tech life on the bbc world service with me zoe kleinman there's nothing we like more than receiving your messages if you've got any thoughts on the show so far please do get in touch particularly if you've ever invented something that you're proud of we'd love to hear about it even if it's top secret we've been sent a whatsapp message from Carlson Amanui from Cameroon who says greetings I love your program that's great to hear thank you Carlson and we're still getting a great response from you about the one piece of tech you simply can't live without. Culwinder Sin emails us from Edmonton in Canada hi Culwinder he's a truck driver and says the Garmin truck GPS saved me countless hours by providing real-time traffic updates and truck-friendly routes it keeps me confidently on the road avoiding low bridges and restricted areas.

10:53Thank you, Colwinder. I can certainly see how useful that is in your job. Joe sent us a WhatsApp from England. He uses his laptop to design and animate 3D models, which he can then send to his 3D printer to create a physical version. I love a 3D printer, although I did once eat a 3D printed biscuit and it was very disappointing, I'm afraid. And hello to John Morris from Malawi, our regular listener to Tech Live. Nice to hear from you, John. He says a piece of tech he can't do without is the humble radio. Popular choice for many of our listeners and quite right too. John says I listen mostly to the BBC on the radio because I can do other things while listening.

11:31TV takes up too much time plus it's a battle with my daughters over the remote. Lovely to hear from you John and I promise we didn't pay him to say that. You can send us a message on WhatsApp on plus 44 330 1230 320 or you can email us techlife at bbc.co.uk is our email address. We'll repeat those details at the end of the show.

12:03Let's talk about another area of life where artificial intelligence is growing. For the last 10 years, the electronics and healthcare company Philips has published a future health index examining the role of technology in health systems around the world. This year, like everything else, it's focused on AI. Two online surveys were carried out by an outside company involving 1 ,900 healthcare professionals and more than 16 ,000 patients in 16 countries around the world, including China, India, the USA and South Africa. I've been speaking to Dr. Shez Partovi, Chief Innovation Officer at Philips. He told me more about the research.

12:42Medicine today finds itself in a really interesting crossroads. You have an ageing population with increasing demand in care, and then you have a shortage of clinicians and physicians, nurses, and so you have this incredible imbalance where patients are waiting months and months for care delivery. And everyone views AI as artificial intelligence as a way out of this quagmire. And what this survey was really interesting about was it revealed that while both clinicians and patients have a sense of trust in AI helping this imbalance and solving some of these problems, the interesting thing was the level of trust between clinicians, physicians and nurses, and patients was different.

13:21So when you actually ask the clinicians, do you think AI will help advance sort of the care delivery and increase access to care? About 79%, almost 80 % were very optimistic. And then you ask the same question from patients and only like 59%. So 20 % trust gap between what the health care professionals think is the, I don't want to use the word savior, but the tool that will help bridge this increase in demand and shortage of supply and the patients that thought that it would do the same were 20 % less. Why do you think that is? From a clinician's point of view, what they saw was that they were spending an inordinate amount of time with administrative tasks that they shouldn't be doing.

14:12And generally, they saw artificial intelligence. I mean, 85 % of them said, AI, we anticipate it will automate the things that I shouldn't be doing anyhow. And that means I can spend more time with patients, more face-to-face time and so on. So they saw this automation as a way of giving time back to them so they could spend with patients. So that, I think, was where they were optimistic. And patients, it was really interesting. About 75 % said, look, I'm eager. Bring technology forward. But I worry that it will actually decrease my face-to-face time with physicians. And so their replies, when you analyze it, and we did this analysis, they said, look, if you're going to use AI for scheduling and for things like check-in, I'm good.

15:03I'm like very optimistic. Use it for those things that are administrative and I kind of check it. But if you want to use it for care delivery and so on, I'm much less optimistic that it's going to be useful. That's patient speaking. So that I think what you saw was physicians that are in the throes of overwhelmed then overworked with administrative tasks saying, AI can help me. And what you saw from patients was, I don't mind if it's administrative, but if it involves you diagnosing me and caring for me, I want the physician in the loop. And actually, Zoe, there was one interesting data point there.

15:43When you ask patients what would increase your trust, they said, if you put the physician in the loop and have the AI informed by a physician, I'm 86 % optimistic. So actually, they came up to the same optimism of doctors if the doctors were in the loop, which is a clear signal that says, okay, I see the value of AI, but I want my physician in the loop. Yeah. Which honestly is part of the reason in the actual survey, we had a call to the industry to say, look, AI should be people first, improve in care, improve face-to-face time. It should create human AI collaboration because what we're seeing and what we're hearing is, this is how I will trust it.

16:28And so those are the recommendations we made from this survey. I met a lady last year, actually, who had very early stage breast cancer. And it was picked up by an AI tool looking at a routine mammogram that she happened to have because of her age. And the the human radiologist had missed it because it was so tiny. So without the AI tool, she wouldn't have known about it for another three years. She had no symptoms. And she was very pleased. But the trial also threw up hundreds of false positives because of the lack of data access that it was allowed. So it didn't know that, you know, other cysts and lumps and bumps that women had had for years that had been discounted by human doctors, it didn't know about those.

17:12So it was throwing all of those up as well. We have to have a big conversation at some point, don't we, about health data and how we use it. But it's very emotive for people. At the end of the day, what you want, when you're building AI models for predicting and diagnosing, the data source which you use is really critical because you want the data that the model is trained on to be very representative of the population in which you use the algorithm in. And so this idea that you want to avoid what's called bias in the data, in the model building. And even so you want the data to be as rich as possible, as deep as possible.

17:50So these algorithms are as accurate as possible. In fact, in the survey, clinicians, 65 % or so said that they actually are concerns about potential bias in the data set. And which means, of course, you need to have data that's rich. and then after you've released a model to actually do post-market surveillance to avoid, to understand it. But you're absolutely right. A, the more access to data, de-identified, of course, that one has to build models, the more accurate the prediction and diagnosing opportunity from the AI itself. And then that means they can support physicians and nurses in them delivering better care for more people.

18:29And of course, in the case of your friend, potentially even identify things that may have been overlooked. You're obviously very pro more AI tools being used in healthcare, which is also, it has to be said, good business for Philips, isn't it? Well, I mean, Philips' mission is to deliver better care for more people. This is why we come to work, 70 ,000 people come to work every day under this idea that we want to improve the health and well-being of 2.5 billion people every year by 2030, of which 400 million should be underserved communities. And so when we look at the current gap in between supply and demand, And we see that AI is the one mechanism to do that.

19:05And so for us, we do want to take advantage of AI to help clinicians, A, get more time back so they see more patients, and B, improve the diagnostic accuracy so they deliver better care. My thanks to Dr. Shez Partovi, Chief Innovation Officer at Philips, speaking to me from Amsterdam in the Netherlands.

19:28If you use Google for online searches, you're in plenty of company. I typed into Google, what's Google's global market share for online search, and up popped an AI overview saying the figure is 89.73 % this year. But scrolling down a bit more, a few search results suggest that market share has been dipping slightly with more competition available. It is still a huge slice of the pie though, and it's being closely watched by regulators. Google's just announced some changes to search, among other things, at its annual IO developer conference. I've been speaking to Pandu Nayak, chief scientist at Google Search.

20:05He told me about some of the updates. We have this feature called AI Overviews that provides you a little response to your more complex questions. AI Overviews is now going to be available in over 200 countries in more than 40 languages. More importantly, our most cutting-edge Gemini models, that's the 2.5 version, is now coming to both AI overviews and AI mode, which essentially lets us help you with your most complex questions. Another big thing is AI mode, which is effectively like AI overviews, but one where our sort of power users wanted to get AI responses for even more of their questions.

20:59So it's a special mode that we launched as a lab in the US earlier this year. And that's led to a lot of positive reviews from the users who entered the lab. And so we're going to make it available to everyone in the US. And will that look different? It won't look different than what it looked like in the lab, but it is a separate mode in search. You know, there'll be a little tab that you would click on to get get into that mode. And it's a place where where we will bring sort of the latest, latest models, the latest features. And then some of those things that sort of work well in AI mode will then flow on into AI overviews onto the main search page.

21:46So it's it's all AI, isn't it? The future of search is AI. I mean, we keep hearing that there are there are other AI companies that I talk to who are quite frankly gunning for Google search. Do you feel the heat of that competition? Our focus really is on serving our users, helping our users. That's that's been the key. And when we say that the future is AI, it's really very grounded in what we are seeing with our users. So AI Overviews, which is one of the big things that we launched last year, has found excellent product market fit in that users are loving it. It's leading to actual growth, which means people are using it even more, sending us more complex questions.

22:36And I think this is wonderful. I think while I understand the convenience of AI search, I mean, certainly it's easier to have one comprehensive answer than to wade through hundreds, thousands, millions of search results. Is there a trade off here with accuracy? I mean, there's a Google AI overview result that did the rounds the other day where somebody asked if the actor Al Pacino was in heat. And the answer was all about female reproductive cycles, which is, you know, a funny example. But is there a danger that, given that we know that they're not always accurate answers, that this could end up becoming less reliable?

23:17One of the things that we have worked hard over the years in search is to surface high quality, reliable information for users. And in fact, users have recognized this and we have built a lot of trust in Google from our users. That's one of the reasons people come to us for these questions. With things like AI overviews, we continue to hold a very high bar on reliability in this regard, right? There's a lot of testing. More importantly, the techniques we use make sure that there is corroborative evidence from the search results. And so to the extent that the search results are great, you get great AI overviews.

24:04There are environmental concerns around AI, of course. And someone said to me not so long ago, you know, if every Google search was an AI led search, then not only could Google not afford it financially because it's more expensive, but the planet couldn't afford it environmentally because of the resources involved in powering it. What's your response to that? So I think you make a great point, but in your question is also the answer to this, right? Like you said, we couldn't even afford to do that. And so we are incredibly incentivized to make our use of AI as efficient and cheap as possible. And in fact, even over the last year, we have seen our efficiency improve dramatically.

24:53So we've got work going on pretty much at every level of the stack, starting with the hardware. You know, we build our own AI chips. These are these so-called TPUs, tensor processing units that are very, very efficient. at the software level. People, we have lots of engineers who are trying to make it as efficient as possible. At the model level, there's all sorts of work going on. But the point is that at every level of the stack, we are trying to optimize it so that it is fast. And it's really central to us to be able to do that because we need to do this at scale. It doesn't serve us if we can't do it.

25:40My thanks to Pandu Nayak, the Chief Scientist at Google Search, speaking to me from London.

25:48It's time to wrap up this AI-themed edition of Tech Life. If you'd like to tell us your thoughts about the show or tell us about the tech you simply can't do without, please contact us. WhatsApp us on plus44 330 1230 320 or email us. Our address is techlife at bbc.co.uk. Today's Tech Life was produced by Tom Quinn, edited by Monica Soriano and presented by me, Zoe Kleinman.