Newco Shift Forum 2018
As CTO of Adobe, Abhay Parasnis oversees the company’s technology strategy, and he’s betting big on narrow but deep AI.
Now that every business worth its stock price has moved to “the cloud,” creating massive technology winners like Salesforce, Amazon, and Google along the way, the technology industry finds itself searching once more for a metaphor that can drive its seemingly endless cycle of identifying and building the “next big thing.” And while it seems almost too obvious to identify artificial intelligence, or AI, as that next thing, Abhay Parasnis, CTO of Adobe, makes a strong case for why the received wisdom may yet prove true.
Famous for betting the company on the cloud five years ago (and winning big), Adobe is making an even bigger bet on a certain kind of AI — what Parasnis calls “narrow AI.” Adobe’s goal is to leverage narrow AI across its core suite of products in creativity, marketing services, and business services, in the process simplifying them and making them accessible to a magnitude of order more potential customers. Forget Terminator references, where a generalized AI takes over the world, Parasnis told me at the recent NewCo Shift Forum. Think instead of the magical world of Harry Potter (minus Voldemort, of course). Below is the video and full transcript of our conversation, edited for clarity.
John Battelle: It’s time to talk about the future of technology with the CTO of one of the largest technology companies in the world. Adobe has made an extraordinary transition, now a classic Harvard Business Review case, of how you transition from a packaged goods software company that shipped shrink-wrap software to a cloud company.
But we’ll discuss what’s beyond that, not just for that company, but for the entire technology industry. Please join me in welcoming Abhay Parasnis to the Shift Forum stage, CTO of Adobe. [applause]
Abhay Parasnis: Thank you. Great to be here.
I mentioned that transition, it was a remarkable pivot for a very large company. But it’s part of what feels like an ongoing cycle: I started covering technology when mainframes were out, and client-server architecture was the cool thing. Then client server was out because networked PCs were the cool thing. Then networked PCs became the Internet, and that was the cool thing. Then mobile, then the cloud.
It seems like we’ve been in the cloud long enough to say, “OK, what’s next?” We must have another cycle or we are going to run out of ideas here! Let me put that to you as a senior technologist in one of the largest companies in the Valley. What might be past the cloud?
First, on the transformation — just to tee up, thanks for all the kind words — that transition will probably bridge a little bit to what we see from both Adobe’s perspective, but more broadly from the industry standpoint as we look at what’s beyond the cloud.
But it’s worth stepping back because there are a lot of parallels between what we just went through in our own business and as I look at the future. As you say, there’s a lot of discussion has happened around Adobe’s transformation. I would say the two or three core things that may not be as well understood and that leads us to what’s next, is that you have to recognize when maintaining status quo is just not enough.
For us, in the last transition, there were two factors. First, back in 2008 with the economic slowdown, where we realized that the recurring revenue that we had as a company was such a small percentage of the business. And when a big macroeconomic event happens, we basically said we are never going to let that happen again as a business, both for ourselves but also from a market standpoint.
The second was with the advent of iPhone and the smartphone wave starting to happen, we realized we were fundamentally misaligned with the pace of innovation. We were still delivering 12- 18-month box software cycles and the market was moving much quicker.
And so those two factors led us to a point where we basically said no matter what the cost, we have to transform ourselves. I think that conviction and that notion that status quo is not enough is actually pretty critical for companies that look at transitions in the marketplace.
Frankly in hindsight, it looks obvious, but at the time customers were not super happy with us. Not all of the customers were happy. Some were, but some customers were quite pissed with us, and actually felt we were not listening to them and not doing the right thing by them. We had to have a balancing act of listening and adapting to our current customer needs, but also maintaining the conviction of where we had to go relative to this space of innovation and continuous innovation.
This is also an industry where yesterday’s success is really not that relevant. As far as we are concerned the whole cloud transformation of our business is done. It’s behind us and we are going to think about what’s around the corner.
When you ask what’s beyond cloud, first I will say, it’s not so much what’s beyond, but rather what’s building on all the changes that have happened. Each of the changes you mentioned, maybe except the network computer, but mobility and the cloud have really laid a foundation of computing model that we think unlocks pretty drastic set of innovations for the next decade.
From my vantage point, there are two or three areas that are super exciting. One, I think software over the next decade is going to go through a very transformational shift of not just being an adjunct in our lives. Software has already had a pretty profound impact if you look at last two, three decades from automation to productivity to touching lots of our professional lives and personal lives.
Over the next 10 to 20 years, we think software is going to go from just being this adjacency in our lives, to fundamentally permeating every part of our life in ways that we haven’t even thought possible today. The shift driven by artificial intelligence, I’m sure every speaker here probably has talked about…
We’ve done a really good job here so far, it hasn’t yet become an eye roll.
I am sure there are people in the room who will do the eye roll.
At a fundamental level, we believe in the notion of AI and intelligence — and not so much the technology aspects of it, but the profound potential it has to transform what software can do in industry after industry and in every aspect of our daily lives.
We are making a even bigger bet than what the cloud shift was for us — very much an existential bet for us — because the power of AI combined with these devices and new immersive media and new types of experiences can completely change the role of software in our lives.
There are all these phrases around software “will eat the world” and stuff. But we believe AI absolutely has the potential to be the defining aspect of what the software business looks like 10 years from now.
We’ve spoken before so I have the benefit of your insight on this, but I would say you are say profoundly optimistic about this. We have mobile, we have cloud, we have the infrastructure in place, we have the commitment from very large companies. You believe the opportunity is there. Yet, when the world first heard about this idea of software being everywhere, it was eating things. That’s not necessarily good. And when they saw AI and machine learning in the wild and when the world woke up to it, it was because it was being used to corrupt our elections.
How do you counter the narrative which seems to be, “I’m not really happy about the clouds waking up, and some algorithms starting to work out there, because it doesn’t feel like it’s in control”?
Yeah. I think that’s a really good point, and in some ways one of the nice things about an event like this is that you get a lot of people with different perspectives from different parts of the industry, not just the tech echo chamber.
First, I’ll acknowledge whenever you have a massive shift like what we think AI will be, it is very healthy to have a dialog on pros and cons, what’s the upside, what are the downsides, and on the social policy and other implications of this, beyond just technology.
I would say we spend a lot of time thinking about that. That said, you and I have talked about it, we happen to be more on the optimist end of the spectrum. Clearly, a lot of the examples that you mention, there are a lot of topics right now around misuse of this kind of technology and approaches. Internally, we joke there are a lot of these current views of the world which are more Terminator-like, if we use the movie example, which is machines taking over and people getting displaced in jobs and stuff which we have to be mindful of. At Adobe given our DNA of creativity, design, and amplifying human emotion and human storytelling, we have much more of a Harry Porter view of the world of what AI can do.
You have a Harry Potter view of the world. It’s magic?
It’s magical. It’s more accurate.
There’s Voldemort, let’s not forget.
There always is. Movie analogies aside, we believe if done right, these technologies really have a potential to profoundly democratize how everyone can tell their story. The other side of misuse of these elections, is (technology) giving everyone a voice, giving everyone better tools to tell their stories. That said, we have to be mindful about the ethics of data, the ethics of privacy, of AI, you have to really watch out for those.
Given that, and I love an optimistic view, I think it’s probably time for me to ask you, we’ve heard how IBM is doing AI, and we know Google and others are doing AI as a white-label service.
How is Adobe doing AI? Because you are well-known for both your marketing software and also your creative software? Are you going to help me draw better than stick figures with AI? How literally is AI going to change your products?
First of all, as you said, there are a lot of different players in the software and tech market that are all suddenly applying their own perspective and their own unique vantage point on where AI will lead. I’m happy with the level of focus and attention and everyone is approaching with their innovation mindset.
One thing I will clarify is we are not trying to build a general-purpose AI or a AI machine that can take any problem, cure cancer, predict weather. There are some in the list you mentioned that are going after very broad aspirational general-purpose AI kind of type things.
We are very unique and focused in our idea around this, where our business and our mission, more than even our business, has always been to change the world through digital experiences.
By that what they really mean is building tools like what Photoshop has been, what PDF has been, Illustrator, InDesign. These are tools that creatives around the world have used for the last couple of decades to really bring to life their stories, their emotions, their vision.
So our view on AI is very specific, very narrow. We take a lot of pride in our approach being extremely narrow and deep. We are not trying to solve general-purpose problems.
In terms of how you are likely to experience it, one of the things I see right now…I have a 12-year-old son, and when I look at his use of computing and his friends, they are growing up with a very different set of devices and experiences, and communication being at the core of everything they do. So democratizing these tools … One of the challenges Adobe has had is our tools are very high-end but they cannot be necessarily used by people without investing months or years in learning them.
One of the advantage of AI is take something like Photoshop and imagine if we can bring it in simplicity and usage, not just to tens of millions of people but billions of people on the planet, and being able to use that to tell stories in creative expression for everyone not just for a select few.
I remember the first time I ever made in iMovie or I used Photoshop or Illustrator, you have to get over this hump of learning how to use it, and then I was just terrible at it from that point forward. I just never mastered it but I got to the point where I could kind of do it. What you are saying is that you can add a couple of zeros to the number of people who could do amazing things with these tools?
Yes, and the simplicity with which they can do it. Even today if you look at something like Lightroom on a phone, one of our products, it’s already dramatically more simple than it was two years ago where you had to learn a lot more. That is absolutely the Holy Grail for us with AI is to democratize who can use these tools to tell better stories.
Now let me turn that back to the Voldemort side.
Harry Potter is going to stick around.
When anyone can use these tools, and you’ve already proven in your labs that you can take snippets of speech and make anyone say anything. We can make anyone look like they’re saying anything in a video or in a photo. One of the things that you told me in response to this question when I talked to you last was that you were thinking very deeply about how to approach that problem through audit trails, watermarks.
There are kinds of things that will ensure that as a society we have a capability to understand what is real and what is not. Even as we wish to suspend disbelief and tell stories, we also need to know what’s true. Can you say a bit about that?
Yeah. I think that’s a good point. First, it’s not new. The technologies and AI may be new now, but even Photoshop. For the longest time people have always asked the question if I can create anything with Photoshop, how do I know what’s real versus what’s Photoshopped? The reality is that we look at it two ways. Our role in this picture is to continue to push the limits of technology, innovation, what’s possible, but we also have to be mindful to your point about the implications, the policy, society, and so on.
One of the examples I’ll give you: With AI, even as we push the limits of algorithms and what we can do with intelligence and dramatically simplify experiences, you have to be very mindful of things like ethics of data or biases not being built into your algorithms because of these systems.
One of the things as CTO I did at Adobe early on is, even as we formed a research lab and a deep focus on AI, we also formed a group of ethicists thinking about data and how actually we are using our own algorithms, are we implicitly or explicitly building in biases in them, or the notion of piracy. When you are creating something, how do you maintain a trail of what was created by who?
We are very focused on it, though I will say it’s more than one technology company’s job. It’s a dialog that the entire society is going to have to evolve as these new tools and these new mediums come into existence. Everyone is going to have to figure out what are the norms, what are acceptable things versus what is not.
NewCo’s Shift Forum is underwritten by “Pillar” partners who help us identify and program the high level content delivered on the Shift Forum stage. We’re extremely proud that Adobe has been NewCo’s partner for two years, working on the theme of “Business transformation” with us this year. For more, see the NewCo Shift Forum site.