Artificial Intelligence

How Adobe is building practical AI for creators and SMBs

Conversations about AI, machine learning, and automation are often sensational, said Adobe's David Parmenter, but for today's business artificial intelligence is now mission-critical.

Dan Patterson spoke with Adobe Document Cloud's director of data and engineering, David Parmenter, about how Adobe is building practical AI for creators and small to medium-sized businesses.

Patterson: Conversations around AI, machine learning, and natural language processing tend to skew into the science fiction realm, but science fact is just as fascinating. David, thanks a lot for your time today. I wonder if we could start with this idea that I'm very fond of, which is everyday AI. We know that businesses are integrating AI systems, but often that turns into jargon soup. I wonder if we could kind of start with the types of artificial intelligence and machine learning that are being used by enterprise companies, and some of the problems they're helping solve.

Parmenter: Right. First off, Kevin Kelly, who's a leader, a visionary, in our industry, he made a really smart comment. He said that basically AI stops being AI when it works. Then everybody just accepts it as a technology. Really concrete examples that we see now are like your spam filter, was at one point AI. It was something really annoying. You'd get a lot of spam, and now you get much less spam. The solution to that is by taking data, by analyzing the data, and then producing ... basically giving you back more of your time, nobody thinks of that as AI, but at one point, that was a mystifying process. That's something that I feel like everybody can relate to with their own experience. Even now, of course, we do get a fair amount of spam, and you get stuff you want in your junk folder. That's the very typical example of the challenge that we have in the AI space. That's an everyday example.

Parmenter: I feel like we're also seeing chatbots that actually work now that are making things more productive. I had an opportunity to speak at the CCW, Customer Contact Week, and everybody there is basically a call center professional. They're deploying these systems now. The really interesting thing is that it basically ... The people who work in a call center are actually happy with that, because it allows them to work on higher-value contacts. They don't have to take somebody's serial number, or tell them to reboot their system and so on and so forth. As far as we can see, these solutions are being accepted by the people especially, and it's crucial that we draw the line between machine and person, especially if it's really easy to opt out. We're seeing acceptance both at the enterprise level, because no small business is going to deploy a chatbot, but we're seeing it accepted it at that level and in the everyday case, in technology that's in your operating system, that's on your phone, et cetera.

SEE: The latest Adobe Photoshop update continues to flex its AI muscles (TechRepublic)

Patterson: Yeah, I love the idea of AI offloading a lot of the cognitive load for kind of drudgery tasks and allowing people to focus on, allowing humans to focus on, bigger challenges and higher quality tasks. I wonder if you could help us understand for SMBs some of the opportunities in the AI space.

Parmenter: Well, I think the first message I would have to everybody who's an SMB is you have data that is particular to your business, and you have expertise. It kind of starts with a question like, how would ... I have this problem. How would I use data to solve this problem? Okay, and it might be that you want to generate more leads. It might be that you want to retain your customers better or increase your satisfaction. In every one of those cases, your solutions run the gamut from, "I have an intuition about this," to, "Okay, I have a data scientist. I built a model. I made a prediction. I'm going to test my prediction," to, "I don't know what the answer is, but I have a lot of data. I'm going to let these algorithms loose on the data and see what kind of insights they have." All of those, I think, are really possible. My No. 1 message is: your data's very valuable. Don't give it away. Don't give it to Google, don't give it to Amazon, and don't give it to Adobe for free. You really, really, want to figure out how to take advantage of that data for your business.

Patterson: I know that, especially in the consumer space but even in business space, smart offices are becoming pretty routine, and voice is a big part of the smart home and the smart office. How are some innovations in voice technology changing the applications of artificial intelligence?

Parmenter: Well, I think the first thing that's really interesting, I actually worked in one of the first commercially successful speech recognition systems in the '90s, Dragon NaturallySpeaking. The first thing that you see now is it is socially acceptable to talk to Alexa or to Siri in a public setting. That used to be an awkward thing. Now, people expect that that's a normal thing that will happen. I have an Alexa, and I have two Echos in my kitchen, one for my music and one for everyday life. I have to remember which one I'm talking to, so I'm already having social interactions with my two voice bots.

SEE: How Adobe proved me wrong about AI in photography (TechRepublic)

On a day-to-day basis, the kind of thing that you would expect is that there will be some kind of a portal, it could be Google Home, it could be Alexa, and that's going to be your primary way into the ecosystem. Amazon in particular, a company I really admire, has done a really nice job of building out the ecosystem in terms of templates and in terms of skills, so that I can operate my security system, I can operate my lights, by voice. This kind of thing, I think, is really normal. Again, we don't really think of it as AI. We just sort of think, "Oh, that worked." I think for the consumer, the future is very bright.

Obviously, there's a big privacy concern. What's going on with my data? Is it listening all the time? We need to be very mindful of privacy, but if you can get past privacy and the trade-off works for you, there's a lot of interesting things that are being done. I think for people who are evaluating technology, the thing I would look at is make sure that it's not a walled garden. Make sure that it's something that participates in a larger ecosystem. It doesn't have to merely be the Amazon ecosystem or the Google ecosystem, but fundamentally these technologies should be generally applicable and available to people, and not something where I'm locked into a specific vendor.

Patterson: David, I know that you guys at Adobe have done a ton of research trying to learn not just about AI, but learning how companies adopt and use machine learning in AI. I wonder if you could leave us with a forecast. If we look at, say, the next 18 to 36 months, I know there's been a lot of hype, but as we know, there's a lot of practical application. I wonder if you could kind of point us in the direction that practical AI is heading in the next couple years.

Parmenter: I think that one of the big areas that really needs to be solved is sort of self-service. If I'm a small business and I want to deploy a solution, can I do that myself? Can I take my data, upload it into a Google or a Salesforce solution, and have it just, quote-unquote, work? Right now, that's hard to do. Getting to the point where individual people could take advantage of this technology without a lot of preparation and a lot of building up the infrastructure, it is still early days. For example, to deploy a chatbot would be a big deal. I'm really looking for self-service.

Also see


About Dan Patterson

Dan is a Senior Writer for TechRepublic. He covers cybersecurity and the intersection of technology, politics and government.

Editor's Picks

Free Newsletters, In your Inbox