Presidential Lecture Series | Shantanu Narayen

Thank you so much. Thank you, Preston.

No, thanks for having me and it's great to be here. Engineering has always been one of the hallmarks of what I have studied and what you have done here is amazing, so thank you for having me.

Well, again, maybe for the audience here, a little bit about Adobe. The company was founded with the first fundamental invention being PostScript, and we helped also shepherd in the desktop publishing revolution. And when I talk about PostScript or I talk about WYSIWYG, a lot of people don't even know what WYSIWYG computing is about, it just shows how much computing has advanced, which I think is a great thing. But the company's mission has always been how do we empower people to create and how do we, you know, we used to talk about changing the world through digital experiences. And for many, many decades, the company was incredibly successful as you mentioned, building products like Photoshop or Illustrator or Premiere or FrameMaker. And the idea was that it didn't matter what media type you were using, it didn't matter what device you were using. If you had an idea or a story to tell, we wanted Adobe to help you tell that story. And what happened, I've been as you said at the company for almost three decades, but it was clear that as there were these further tectonic changes in technology, what was happening with mobile, what was happening with the move to cloud, the pace of innovation was dramatically changing. And our business model at that point was really one of these, as you pointed out, we shipped software in boxes, so we used to do product releases every 12 or 18 months. And we just didn't feel like the pace of innovation at Adobe in terms of what was happening internally, the business model allowed us to bring that out externally. And I'll give you a story actually. I was meeting with the Photoshop team one day and the Photoshop team, this was when DSLR cameras were just coming out, and the Photoshop team wanted to support these new DSLR cameras. And you know, being an engineer, I said great, let me sit down and if the DSLR camera is coming later, why don't you build a plug-in that'll update itself? And they said Shantanu, we don't need you for your engineering skills, we got plenty of that, but we don't know how to deal with these financial accounting standards bodies restrictions that did not allow us to innovate because if we added to the software during those 12 or 18 months, we had to carve out revenue. And so the real impetus frankly for what we did in terms of moving to the cloud was if you're a product company in your DNA and you have to innovate, anything that prevents the product people from innovating at the pace at which they needed to innovate is going to come in the way of your delighting customers. And so the hard part was clearly communicating to the street why we were doing it. The hard part was recognizing that if you're delivering software in the cloud, you have to be an always-on, always-available, and so every part of the company was transformed. But the fundamental idea was actually quite simple. It was like let's embrace the cloud, let's allow people access to this technology wherever they want. Inspiration doesn't strike only when you're sitting behind a desktop. Inspiration strikes for your daughter. If she's an artist and she wants at the time of inspiration to use a product, why aren't we allowing them to use it? And that was the fundamental motivation for it. Everything else came as a result of that basic assertion and hypothesis that we had to innovate faster.

Well, the core business model when we transformed from a perpetual software, as it was called at that point, licensed software, to a SaaS-based subscription software, the good news about it is you can attract customers at any time. And once you pay us for the monthly subscription, I think the onus is very much on a company like Adobe to constantly innovate at the pace at which you need it to innovate. Most recently, for example, you now have the ability within a PDF or Acrobat to have what we call a conversational interface. So you can actually, if you get a PDF file, a lot of researchers I'm sure at Purdue University, you can say hey summarize the research that's in this paper and allow me to ask questions so I can actually interact with the document rather than just be a passive reader. And so the ability for us to actually innovate at whatever time the product people can innovate is one of the beauties of the subscription model. And the other thing that we can actually do, if you think about the old model, the old model because the incentive was really about adding a lot of features between these 12 or 18 months, you actually ended up putting in a lot of features that perhaps weren't discoverable by customers. Whereas when we deliver it on an ongoing basis, what we have found is actually the ability for people to consume that new innovation is actually also better because they're getting this innovation in bite-sized chunks. And so for the engineering team it's a real benefit because they can also engage with their customers. I mean one of the things that we love to do is to say if you have something that's an alpha software or beta software, put it out there, engage with customers, get that feedback back, make sure it's ready. And so the additional benefits of actually migrating to this business model were evident a lot later. You can make it more affordable. It used to be when the perpetual software was a one-time fee, it was more expensive and less affordable. Actually, at many times, Adobe software and other software used to be the most pirated in the world. Now, we're really happy that it's become more affordable. So there have been a ton of benefits that have come associated with that move to the subscription model. But to your point, the beauty of PDF was always that it actually was an offshoot of PostScript. The beauty of PostScript was that you could take any file that you had created and by creating a print stream you could print it on a printer, right? And the ingenious part about PDF was like if you could tap into that print stream and then make it a readable format, then it would actually have this lasting value that people have now created. So the amazing thing about technology is that most fundamental technology, the insight is actually pretty obvious in retrospect. And it was just about hey if you want to share files why not allow people to share files across any medium across any data type.

That'd be great. Okay. Doing demos real live is always, you know, a little bit tricky. But you know with the Wi-Fi gods hopefully... we do need reliable high-speed connectivity, Wi-Fi or 4G, 5G, and then large cloud data centers for cloud to work, otherwise the business model innovation would not have the foundation. But I believe I'm seeing a Wi-Fi connectivity here.

That's right.

Give it a try.

I can dance for sure. So let's see what Firefly is going to do. All right. Now, it's going to send it to the cloud. Let's see if we can make it work. Generation... due to high demand.

This is why we don't do real-time demos. But I guess it's good news. It's in such popular demand.

Well, Meta is building one and Amazon's building one. So I don't think it's a Wi-Fi problem. This is a cloud capacity. So, let's wait a little bit.

Well, there are two separate issues and let me touch a little bit on that. I mean the first thing as it relates to intellectual property, when you think about AI, if I could zoom out for just 30 seconds, I think what people are talking about, what are computers really good at? Computers are really good I think at pattern matching. And if computers can interpret what a human wants to do and accelerate that, then we believe that computers are doing their job. And so at Adobe we really think about it and saying the art of creativity, if computers can help you take whatever's in your mind and create it, then we're adding value. So, specifically, as it relates to AI, when we think about AI, we think about data. We think about models, AI models, and we think about workflows or interfaces that people are using to create whatever they wanted to create.

Okay.

Well, if you don't like smart products... No, no, it's better than I do it in person. So I think there are some benefits to that. But, you know, going back to the question that you asked and maybe we can take it off so you don't have to be subject to that.

Exactly.

I think you asked two really important questions. So again, as we think about AI and as we think about how can these computers actually be used to allow you to accomplish what you needed to accomplish. You know, I think most people would talk about AI as what is the data that you're using to train the models? What are you doing with the models? And then how does all this innovation become accessible to a user? What's the interface that they're using? What's the workflow that they're using in order to be able to do that? I think as it relates to creatives and creative professionals, Adobe has been very clear from day one that we believe that the work, the creative work of artists, it's their intellectual property. So Firefly is actually the only model that has been created that is completely and 100% trained on data that we have licensed for. So we have not like the other companies actually gone and scraped the internet and used that data in order to be training our models because we believe that this is really the intellectual property of the creative professionals. So that's the thing that we did on Firefly and because we want creative professionals to be able to monetize the work that they've created. Now in terms of then what we do as an application, we monetize either the Firefly application which allows you, if you want to do text to image, text to video, text to any kind of content, and why are we interested in that and to your point about the apprehension that might exist. I really believe two things. The first is human ingenuity will always triumph what a computer can do in terms of creative expression. But anything that we can do to help accelerate that creative ingenuity is what our customers would like to do. I mean think about it. When you're doing a project whether it's a creative project or whether it's a research paper that you're writing, I would say most people are really terrified of that blank page. How do I start the process? And so I think where Firefly and where all these other models are being used is in the ideation part which is like I want to ideate. I want to be able to think about what I want to do. And if it can serve as this on-ramp, if Firefly or the other products that are being created with AI can serve as an on-ramp then it performs a really important task because it can do what's required in your mind much quicker than you can do it. The second area where I think Firefly as well as all of this AI can really help is in automating tedious tasks. If you're creating high-end video or you're creating high-end animation, the task of then frame by frame getting that perfect... I haven't met a creative professional in my life who would like to take a day to accomplish something that they can do in 10 minutes. And so our job again is the more we can allow them to create things faster, it allows them to actually in their profession do more and if they can do more then they're making a better living. And so that's the second part and the assertion that we are doing which is our job is to make them productive. Our job is to help them ideate. Our job is to help them automate inefficient tasks. And interestingly in Photoshop we have these features called generative expand or generative remove. I mean how many times have you taken a picture and you're taking a picture maybe with the Eiffel Tower in the background and there's some other tourist and you want to remove the tourist. So this is all about trying to find those use cases that add a lot of value. So that's our belief as it relates to AI which is if AI can help people become more productive and automate that, the emotional connection to content, that emotional connection to creativity, I think that's still going to be human. Part of what I say is I don't think Steven Spielberg is sitting around saying I'm going to make a complete movie by just doing a prompt. Right? And so the prompt helps accelerate it but it's never going to replace human ingenuity. I think your second question was all about content provenance and authenticity and how do we deal at Adobe as well as an industry with things that are called deep fakes. We were very proud we started what's called a content authenticity initiative. We now really fortunate we have thousands and thousands of other companies that have joined that and the whole idea is how do you, whether you're a camera manufacturer who was taking a camera or a person on a mobile device, you should be able to say this was taken at this time. If somebody then makes an edit to it, we have the ability to have the metadata that actually says who edited it, when did they edit it, how did they edit it. And we worked very hard with as I said people at the front of that creative process, the camera manufacturers, to say can you please put this content or think of it as a nutrition label. When you are eating something and you want a nutrition label, if for every piece of creative content that's created if you can have this nutrition label, when was it created? How was it altered? And that technology actually exists. And we work with the distributors of this content. We work with everybody to say, can you please make sure that you're carrying that nutrition label with you? The issue that we haven't completely solved is does the consumer who's actually consuming that creative content, do they actually care? And that is an educational process, right? Because when you have all of this content that's out there and some of it was clearly created by AI and some of it was clearly plagiarized, unless the consumer also says I really care about trying to understand the content provenance and the authenticity, you will never completely close the loop. I mean, think of it. When the internet first came out and people were really trained to say when I'm going to put my credit card in a web browser, I wanted to make sure that I saw that HTTPS to make sure it was secure. So, I think there's an education part. This is not a technology problem. With technology, we can say whether technology was used in that piece of content and I think it'll evolve. But I think I would net it out by saying we care very, very deeply about the intellectual property for creative professionals as we've said with Firefly. We care very deeply about their ability to monetize their content because it is their secret sauce. And we care a lot about making sure that the entire industry rallies around how do we recognize content provenance.

Yep.