In this video, I am going to talk about how AI can help you improve, process, and generate process-related innovation. It's probably no surprise to you that the explosion of big data is ripe for advanced AI, and data analytics, and various techniques. From social media, mobile media, every single second we generate thousands or tens of thousands, millions of impressions, text messages, comments on social posts, reaction on social posts. All that involves explosion of text, pictures, videos, and sounds, and although they may come from user-generated data, but much of that data can also come from business processes. Once firms can carefully curate their business process data and really analyze with advanced machine learning analytics tools, this can lead to dramatic improvement in their process innovation. Let me give you an example on Google's data center. Google runs some of the world's largest data centers, and these data centers are extremely energy-intensive. That means they can supply more than 100,000 servers, and it's important to keep these servers cool. The cooling challenge is further compounded by the fact that the computing load is varying quite a bit because it's difficult to predict how much power is being used over time, depends on how much people are using the servers. It also gets affected by outside temperature, humidity, et cetera, et cetera. In the past, humans are typically controlling these pumps, these coolers, these cooling towers, and other equipment. It's really important to keep the data center at the right temperature. Otherwise, you have a server malfunction. These people will monitor their thermometers, the pressure gauges, and many other sensors to make decision over time. Deep learning was a machine learning platform. Try to see we can automate some of that. What they've done, they took years of historical data on data centers' computing loads, various sensor readings, environmental factors such as humidity and temperature readings, and put all that information to train a big neural networks. That's deep learning model. That neural network is used to control all available cooling equipment. This is also a reinforcement learning algorithm to train the deep learning neural network. Can see on this graph on top, in beginning, we have pretty stable power usage effectiveness, pretty stable, and then we turn on this DeepMind platform to reduce the energy consumption or making it more effective. You can see a dramatic drop, after machine learning, is used to control these temperature pumps, coolers, and all the systems related to keeping the servers cool. Over time, you can see it's lowered dramatically, and once the system is turned off and back to human control, you can see the power usage effectiveness has actually gone back to the prior level. After the system is deployed, energy use fell by 40 percent and the overhead associated with this energy use also improved by 15 percent. This is probably the one of the biggest improvement ever seen in this area. There are many other examples you probably use or have heard of them. Amazon, other firms have dramatically used machine learning to improve product recommendation and the process of managing inventory. Infinite Analytics has used machine learning to predict what ads a user will click, and all that data can then be used to predict conversion rates and likelihood a person returning to the website or buying a product. Cybersecurity firms such as Deep Instinct have used this processed data to better detect malware. In many different cases, they can protect the malware before the actual security breach has happened. The insurance company is also a big user of machine learning to analyze their customers data and also to improve their customer support. All that is enabled because these firms can capture the data associated with their business processes and machine learning uses data and apply the appropriate algorithm to improve their existing processes. I gave you a bunch of examples of how AI and data analytics can improve process innovation. But to really see this is one-off thing or this is actually applicable for most firms, so to do this, we actually conducted a large survey of more than 300 firms in a partnership with McKinsey. In this survey, we actually asked many questions about firms. One such section is on process space innovation. There eight questions we've asked on a scale of one through five, how likely a firm is oriented to improve their process innovation. We asked the question such as to what extend the following statement describe the work practices and environment of your entire company? One of the such statement is we have a strong ability to make incremental changes or improvements to our business processes. We also asked questions such as, please list important core activities of your primary business. If you listed process development, process quality, process management, or improvement. Then we also classify you as a firm that orients to our process improvement. We also asked questions such as, how is your organization in proactively engaging with business leaders to refine existing processes and systems? We aggregate all these scales and the firm that score very high on this scale, we label them as firm who have interests oriented toward process innovation. What do we find? This is distribution of the firm that are process oriented. You can see not all firms are process oriented. Some are less so and some are more so. But as a follow-up pretty interesting bell-shaped curve and that's what we expect. It's not a case where everybody say, oh yeah, of course we are interested in processes and [inaudible] or everything or highest score and everything. Similarly, we also looked at data analytics and AI, how do we measure how much the firm invest in data analytics and how much the firm invest in AI. What we did is we actually used the resume data of these firms. We look at data on the employees skills, whether they have skill sets related to data analytics and skill sets related to artificial intelligence. We can look at general AI measurements to see the words such as do words like business intelligence, data center, data-driven, data integration are showing on your resume. Similarly for algorithms, we look at words like A/B testing, machine learning, natural language processing, neural network, etc. We also look at systems and tools to see if they actually use AI system or system related to AI such as cloud computing, Hadoop, MapReduce, etc. This is growth of data analytics and growth of AI skills over time. You can see we see a big jump in the same timeline. We see huge improvement or huge increase in skills in data analytics, the big number of people with data analytic skills, and we see even sharper increase in the AI skills in the recent years. AI skill mirrors data analytic skill, follow in a similar pattern, except they're even little bit more dramatic in the most recent years. What we find is that data analytics and AI can greatly facilitate process-based innovation. Not only that, we find that firm that use AI tools and are oriented toward process improvements experience a even greater productivity gains as measured by their revenues. We measured that one standard deviation increase in investment AI tools are associated with seven per cent more productivity. Similarly, if a firm invest in AI but we're not interested in improving their processes and they will not experience that seven percent productivity gains. Only when you have both, that you have both the investment in AI tool and you're oriented toward process improvements, then we see the productivity gain. To give you one factor that could explain AI innovation paradox in the sense that if you are using AI to improve your processes, you're more likely to see greater gains.