 Hello everyone, my name's Ed Parsons. I'm afraid I'm only going to be presenting to you virtually I would have loved to have been at the conference, but unfortunately Some travel issues I had on the morning of my presentation prevented me from doing so but I'll try to present to you now what I would have said and Hopefully you'll find it useful. I contact details are at the end. Do feel free to get in touch via Email or via social media and I'm happy to continue the conversation So what I wanted to talk about is a change that I think is happening in our industry where we are moving from A time perhaps where we regarded ourselves as as artisans as Developers of a code as a craft to a time where the world has Fundamentally changed and much of that has been driven by the emergence of artificial intelligence, but not not totally Let me start off by just describing who I am and making use of actually artificial intelligence to do so This is the response that barge gave me so our chat GPT equivalent of Answering the question who is Ed Parsons? And I think this is largely content that has appeared probably from from Wikipedia and other content on the web To tell me that I am a geospatial technologist, which actually is true I am the Google Maps guy in many ways. That's the easiest way of thinking about it I'm really interested in how we make sure that Google Maps is is current up to date It represents the the world as accurately as we possibly can and also things like making sure that that little blue dot that appears On the map to represent your location is done accurately as possible so I come at open source really from a data point of view data is is Almost 80% of what we do in terms of a building Google Maps So you may have seen this idea is this this statement that the data is the new code particularly in the context of of of AI and I think we can draw a comparison to that Perhaps less than a helpful statement made earlier that that data was new oil I don't think any of us really thought that the data was a Commodity in the same way that oil is but I think Looking at data in the context of how we develop software and how we develop new services Actually is really important and data is a new code I think encapsulates this idea that actually many of the new products and services are very much dependent on large volumes of data to make them work but our focus is not just on on AI and On I suppose machine learning to use a perhaps more correct term for what we're actually talking about most of the time There was actually a confluence of forces that are impacting on the open source community That have emerged around the same time, but I think we need to look at them Together because they they bounce off each other. They have impacts on the same idea at the same time So what are these these forces that are that are emerging at the same time? The first obviously is this one of around a machine learning a machine learning training How do we manage how do we control access to data sets? Which are going to be used to to create new foundation models for example How do we understand and accept the concerns that many content providers have that well They they didn't realize when they were publishing data online that their data was going to be used In this fashion in a way that you know Could be seen that they are being replaced by a machine doing their particular activity You can understand this from authors and and perhaps the creative types But actually you know your code that you may have uploaded to repository is used to train software development tools That will help individuals write code based on the code that you have written Now you can understand why people might be resilient to this not wanting that to happen But the flip side of that switching off large sections of the web to prevent people using this could have many negative Consequences we don't want to balkanize the web any more than it already is Certainly don't want to switch off sections of the web to prevent people from using it for one purpose as opposed to another The web is going to be you know The enormous value that it has been largely because of of open access to it And the other element here that is about the sort of diversity of the ecosystem around machine learning and foundation model development is that we have to recognize that the barrier to entry is actually quite high and We don't want the situation where it is only the large system provided as the large platform vendors That can build foundation with us that have the access to data and the computational resource to develop models We need to work hard to create an egg I ecosystem that is diverse that allows individuals or smaller communities to access to participate To take the technology in directions that are that are different that are perhaps counter to what some of the large Organizations are doing how we do that I think is still an open question and it's not one that is going to be necessarily straight forward to solve But particularly I think that there is a value in focusing on that data access Point and making sure that everyone gets access to the same data to train their models At least that's something that we probably can control where the parameters that goes into the Particular models and how they are developed Perhaps will always be a bit less transparent And we we need to be really careful about the language that we're using here when we talk about an open ML model What do we actually mean? Now the second Changing in focus. I think the second second force that we need to worry about is a focus on regulatory on regulatory insights and Regulatory access to software development is a reflection of the modern world that we live in the modern world is very much driven By software systems you think about you know the many activities you do During your day-to-day lives many of them are controlled by software systems that are by their very nature complex and complicated or multifaceted and have very long and complicated Mechanisms behind them Unsurprisingly a policy makers and regulators want to understand how this works and wants to be able to regulate that A bit more than they have done in the past because we've seen where you know Harm has come to individuals to organizations where software systems have failed In Europe, particularly, I think this has come to the fore with the emergence of the cyber security resilient act And in many ways this has introduced this idea of accountability for code creation You know the old software development License common Term of you know, no warranty is is no longer Accessible I guess it's no longer acceptable to policy makers But you no longer acceptable to individuals who may be impacted or by a software that doesn't work or ends up with providing Decisions that are difficult to understand or not transparent Software development working in technology is my day job. My real passion though is actually in aviation I am an aviation if he's an ab geek and I want to draw a parallel Between the software industry and the aviation industry in the early days of the aviation industry These are the Wright brothers everyone was an artisan. These guys ran a Bicycle shop they built bicycles and from their experience moved from building bicycles to building aircraft the right flyer in 1903 now it's In many ways, you know, they were perceived as as artisans. They were craftsman. They knew their trade They built what they did and to be honest, there wasn't a huge warranty with their air product initially They killed many of their customers because this was an emerging field and everyone knew the risk that it was associated with the early days of flight, but very rapidly That had to change people had to become as Manufacturers of aircraft much more responsible for what they did processes had to be implemented that controlled Managed the construction of aircraft. I am working with a team at this point in time who are building a a spitfire and and this is a Photo of the workshop very much like any other workshop you could see with individual components of a spitfire that are hand-built by rolling Aircraft grade aluminium But you can see there are little yellow labels inside the plastic bags that contain the various components And they document the construction of that particular component by the person that built it and by the supervisor the chapter So, you know, I guess drawing the analogy in the old days, but you used to be able to develop software Anonymously that came with no warranty That's no longer going to be relevant in the age of things like the CRA We will have to have that similar level of documentation With what we do there's a fundamental infrastructure by which we build Software is changing in the same way that the infrastructure that build aircraft is changing Likewise the modern industry as you'd expect is is is highly regulated Quality and responsibility will become and like they are in the aviation industry pillars of how we operate And this is where that overlap with the open-source community. I think operate works at the same time How do we make sure that? This increase the requirement for documentation It means that we can carry on without the Software industry being buried under requirements to develop lots more Technology lots more process to allow us to develop software We're going to need to create some sort of a compromise to allow us to do that But but I'm sure we will be able to to move forward part of that I guess Comes with the recognition that you know, there is a new landscape here We need to recognize the fact that we will need to work much more closely with legislators And in the past we can see examples where that hasn't necessarily work Well, it hasn't been a good communication between trained technologists software developers and legislators Key example of this is the equal regulation. You may Be aware that there is a button in your car in Europe Particularly because the legislation requires it that you compress and you will be put in touch automatically with the emergency services and the The car will share information with those emergency services about your location and so on The legislation was very prescriptive when it was written and it required the cars to Communicate using a 3g cellular network as you'll be aware perhaps those 3d networks are being currently switched off But yet you can buy a brand-new car today and it will have a 3g based equal system Even though that system is not likely to operate when you take the car out And that's an example of poor legislation where the communication broke down between the technologists and the legislators The idea was was absolutely right But it was it was poorly crafted what we want to make sure is that doesn't happen in the future And to do that we need to be in the room where it happens We need to be active Participants in these communications and I think to be honest around CRA perhaps we allowed the Discussion to Billy become a bit too confrontational And I think to keep the open source ecosystem healthy. We need to be active participants often that means meeting in rooms like this one with no windows and Going through perhaps some quite dry discussion, but I think we need to be active participants What is Google's role here? Well, we remain committed to open source we remain Wanting to be stewards of the open source ethos. You know, we have had some challenges in terms of How we're able to do this over the past year But we'll we'll keep on moving forward or keep committing to things like summer of code But we need your help as well We need you to be active participants in these discussions There is going to be a requirement to talk much more probably with legislators as that focus looks on AI and and and software from a perspective of Control in some ways, but we need to be part of that debate. So we need you to be participants Participate in whatever communities you feel comfortable with and I think the the interesting part of this debate over the last Year or so is that there's been quite a diversity of discussion. There are many different groups talking about this There is a lot of political visibility of this and it's important for us to To be active participants with the rationale with the idea behind it that we want to keep a diverse ecosystem where as much of Access to this is is is open as is possible So look up and keep up with the software and foundations as they emerge in this space Be active participants in the mailing list reach out to the academics who are looking at this And yes, Google does have some interest in this space look at things like ML Commons where we're trying to develop And a common approach for training data sets for machine learning But increasingly I think you know the the message I want to communicate is let's move forward together You know, we are a big player here. We have some Abilities that perhaps individuals smaller contributors don't have in terms of talking to legislators But but we want to understand the broad view of the ecosystem in this space Well, we don't want the the the process to become too bureaucratic to Driven only by by the large vendors. We want an ecosystem that is healthy moving forward So do keep in touch do be vocal in your views But do sorry no constructive fashion as I said I think a key change in our industry is that we are going to end up talking to Allegiance way to two policy makers much more than we have done in the past Whether that's done as individuals that's going to be really difficult So perhaps work with the foundations with the organizations that you feel best represent you Moving forward and and we will be as as active in participating in that process as we possibly can be So once again, thank you for your attention. I'm sorry. I couldn't make the conference in Bilbao Although I did get to the airport. So the aviation enthusiast of me tick that box, but it wasn't really the same So, thank you very much for listening. I look forward to hearing any Comments any questions you might have get back to me via my email or via social media And once again, thanks for your attention