The critics warn against giving ourselves over to artificial intelligence (AI) and trusting smart machines to make important decisions that already affect many aspects of our lives.
But Chwee Chua is more optimistic than that.
As Global Research Director for Big Data, Analytics & Cognitive/Artificial Intelligence at International Data Corporation, Chua says that inthe not-so-distant future, advances in AI and machine learning will propel us to the point where it's possible to build predictive, conversational, digital assistants like the ones you see in modern science fiction movies.
"Some smartphone manufacturers are already developing mobile phones with neural networks inside," says Chua. "We're on the verge of seeing powerful AI systems running locally on your mobile phone."
"That's the kind of digital transformation that's going to disrupt how we work and live in the not-so-distant future," says Chua.
In this Digital Trailblazer interview, Chua helps us understand why procrastinating on AI adoption could put your organization at the mercy of your rivals.
Hope you enjoy the conversation.
Appian: Welcome to Digital Trailblazers. The topic of artificial intelligence is grabbing lots of headlines these days. There are plenty of vendors out there pitching AI, and there's a ton of hype that goes with that. What advice would you give senior executives who want to cut through the hype about AI?
Chua: I think you're right. There's a lot of hype about AI. Almost every company out there is pre-fixing "AI" in front of their products and services to ride along the hype wave. The truth is, about 80% of it has nothing to do with AI at all.
"Over a year ago, we looked at some of the chatbots out there. And 80% were nothing but FAQ's that involved typing a keyword search and returning some kind of response. So for executives looking to get into the AI game, it's important to understand the capabilities of the technology that vendors are trying to sell you."
The question is, is it really AI, or just something that's been re-branded as AI?
Appian:So, it sounds like there's some confusion about AI and what it is. Based on your experience, what are the biggest misconceptions about AI?
Chua: One of the biggest is that AI is easy. Actually, it's very difficult. I would say that about 95% of AI is not just plug and play. It's not that easy. We hear a lot of hype about self-driving cars.
But they're not ready for prime time yet. It's really expensive. It requires lots of technology, lots of sensors, and lots of algorithms to get it to work really well.
So the misconception that you can just take a plug-and-play approach to AI is a problem. It can happen in some instances with APIs (Application Programming Interfaces). But AI is generally not that easy to implement. It takes time to get there.
Appian:Given the difficulty of conventional approaches to adopting AI, what are some of the biggest mistakes you see with AI implementations?
Chua: It gets back to jumping on the bandwagon before figuring out the business problem they're trying to solve in the first place. The trick is to define the right use cases first. Define the problem you're trying to solve and the data you need to solve it.
Appian: You mentioned data. Many experts say that data is the oxygen of AI. Why is data so essential to implementing AI?
"Problems with data represent one of the biggest barriers to successful AI implementations. Many companies only become aware of how problematic data can be once they actually start to try and implement AI solutions. The first step in the AI adoption process is to produce clean data sets to feed into your AI system."
We see many companies struggling with this issue. To be successful with AI, you need lots of good data. And you also have to resist throwing money at a solution that hasn't been carefully thought out.
Appian: In a recent appearance on CNBC, you said that we're in the Golden Age of AI. What did you mean by that. And how does that concept fit into the big picture of digital transformation?
Chua: We've come through about two or three AI winters to get to where we are today. In the past, we just couldn't seem to deliver on the hype cycle. But I think we're now in a perfect storm of convergence between the three different components that are needed to deliver on the promise of AI data, software, and hardware.
We're at a point now where all of these things can help us accelerate to the business value of AI. Information transformation is one of the dimensions of digital transformation. And that is where AI resides consuming, ingesting data to deliver the business value we need for digital transformation.
Appian: You mentioned that we're at an inflection point, that we have the capacity to accelerate to the business value of AI.
Chua: Yes, what I'm seeing right now in terms of accelerating AI adoption is the use of applications programming interfaces (APIs).
"At IDC, we expect that by 2021, more than half of large enterprise will see about 40% of their AI integrations come through API eco systems."
This will allow companies to expand their digital reach far beyond their own, typical, customer interactions.External APIs allow companies to accelerate innovation and expand the range of products and services they can offer. It's a catalyst for digital transformation.
Appian: Where (in which industries) do you see the biggest opportunities for AI-driven digital transformation? On the flip side, which industries have the greatest exposure to disruption from AI adoption?
Chua: Across the world, we see a lot of marketing companies stepping up as early adopters of AI. They were among the first companies to realize that they were sitting on a treasure trove of digital data. So, marketing companies were among the first adopters.
We're also seeing that banking and financial services are other areas where we're seeing fast adoption of AI right now, followed by cyber security.
As for which industries are facing the biggest risk of disruption? I think it's any industry with repetitive manual processes, because AI's most immediate benefit is automation.
Appian:As you know, many of the companies with these massive manual processes date back to the pre-digital era. What steps can these traditional companies take to prepare for the rise of AI?
Chua: The big challenge, right now, is not to be like an airplane circling the airport waiting for a chance to land. Don't be like Kodak. They invented the digital, single-lens reflex camera. But they decided to stick with traditional film technology.
Then, they went out of business, because everybody pivoted to digital.
So, the biggest challenge for traditional companies is to integrate AI into their business model to create value for their customers. That's the first step.
Appian:Another challenge business leaders face is understanding the different flavors of AI. Experts talk about general AI and specialized AI. What's the difference?
Chua: Specialized AI is dedicated to a singular task. Sometimes it's called narrow AI. This would be like Optical Character Recognition (capability for computers to recognize printed or written text), scanning and translating and doing sentiment analysis.
Self-driving cars is another example of narrow AI, because it represents just one dedicated skill set. On the other hand, general AI can literally replicate any human task.
We haven't reached that stage yet. The closest thing to it is probably in robotics and smart homes. But we're still in the very early stages of general AI.
Appian: So, when you think about the mainstreaming of digital assistants like like Amazon's Echo and Apple's Siri, that's just scratching the surface on general AI.
Chua: Yes. But soon we'll have more intelligent machines that can do much more than the consumer interactions processed by today's digital assistants.
Appian: While we're on the topic of human-AI interaction, what impact do you see AI having on the workforce of the future? Is this something we should be concerned about?
"We need to look at AI as the next industrial revolution. It's going to augment how we work. Low-skilled, manual, jobs will get displaced, just like in the first industrial revolution. But this will free up workers to focus on higher-value tasks."
Like any new technology, AI will displace some jobs. But AI won't be a massive job killer. It's going to augment how we think, how we process information, and how we make decisions. And it will create new categories of jobs as well.
Appian: Machine learning and deep learning are also getting lots of press coverage. What are your expectations for growth in these AI-related technologies? They're going to get faster and better.
"But what we've got to watch out for are adversarial attacks designed to cripple systems powered by machine learning. That's a scary notion, as more and more enterprises adopt machine learning and deep learning especially in public safety."
In fact, at certain airports, they're already scanning passports and using facial recognition, without the help of custom and immigration officials. In these cases, adversarial attacks on AI systems can have huge implications.
But, generally, I expect growth in machine learning and deep learning to explode. Just as we've become accustomed to the convenience of GPS routing, we'll also expect the convenience of machine learning to make tasks and jobs a lot easier to do.
Appian: For C-suite execs not familiar with AI lingo, how would you describe the difference between machine learning and deep learning?
Chua: We get that question a lot. In general, machine learning is a subset of AI. And it encompasses a range of algorithms that enable trend or pattern recognition over time. It comes in two flavors supervised and unsupervised with no inputs from humans
Deep learning is a subset of machine learning. It has many applications in the digital world. From speech recognition to image recognition, even biomedical informatics. Think of it as many layers of nonlinear processing, where each layer uses the output from the previous layer as input.
Appian: You've also written about how machine learning platforms will enable companies to build self-service machine learning capabilities, without deep technical skills. What are the business implications of that?
Chua: I think there's huge significance in it.
"C-suite execs aren't expected to be highly technical. But they're expected to make critical business decisions. The good news is that automated, machine learning, platforms allow non-technical execs to do things that were previously impossible."
A friend of mine likes to jokes that machine learning can make you feel like you're able to launch a space shuttle, because it enables you to do very complex calculations (laughter). Of course, that's not the case. But think about modern cars.
They're packed with so much AI and accident avoidance technology, they can make even the worst driver look good. This is how AI augments our natural intelligence with the ability to make decisions and do things that we previously couldn't do.
Appian:It seems like we're seeing a growing convergence of automation and AI. What do you make of this trend?
Chua: Automation is not just robotics alone. Automation is only part of the equation. We also need AI-enabled automation. Robotics without AI will not maximize productivity.
I go back to what I said earlier about Kodak, and how they lost out because they waited to long to adapt to digital. The digital camera companies themselves were disrupted by the picture-taking capabilities of the modern smartphone.
Today, companies that figure out how to squeeze AI into mobile devices will drastically change how you and I work, and how brands interact with consumers.
Some smartphone manufacturers are already developing mobile phones with neural networks, which would enable powerful AI systems to run on them.That kind of change is going to disrupt how we work and live in the future.
Appian: You mentioned data quality as one of the barriers to production-ready AI. What are some of the other roadblocks to successful AI implementation?
"The talent gap is a big problem the shortage of data scientists and AI specialists. In fact, 73% of organizations surveyed indicated they have none. This is a big deal, because organizations without AI expertise will be out-competed by companies that have it."
Appian: Finally, what are your expectations for AI in 2018 and beyond?
Chua: In the Asia Pacific region, one of the predictions that we came out with in 2017 is that over 30% of large enterprises will start generating Data as a Service revenue from the sale of their internal raw data because of AI.
"By 2019, over 40% of digital transformation initiatives will use AI services. In fact, by 2021, we expect 75% of commercial enterprise apps to have AI built in, and over 50% of consumers will interact with AI."
Two years down the road, personal digital assistants and bots will influence 10% of all sales. And artificial intelligence will be a key driver of growth for many organizations.
Appian:So, you see AI as a catalyst for business growth. But AI has been around for at least the last couple of decades. And many people are still skeptical of it. Have we finally gotten over the hump? Is AI now more reality than hype?
Chua: Yes. We've seen more acceleration in AI development than we've ever seen before. And organizations that don't take advantage this trend will be casualties of disruption.
Want to learn more about the state of AI? Listen to the podcast below featuring Cognilytica, an AI and Machine Learning focused analyst and advisory firm. It breaks down:
[podcast id="16341224" text="Practical Applications of AI for Today's Organization" on Spreaker."]
(This blog was originally published on April 30, 2018)
Appian helps organizations build apps and workflows rapidly, with a low-code platform. Combining people, technologies, and data in a single workflow, Appian can help companies maximize their resources and improve business results. Many of the world’s largest organizations use Appian applications to improve customer experience, achieve operational excellence, and simplify global risk management and compliance.