<p>Artificial Intelligence has been a topic dominating the discourse across the board, for almost all organizations, for the last few years. The key challenge statement is - where to start? How to leverage the technology marvel to bring value to the table?</p><h2><span style="font-size: 14pt;">The Google Story</span></h2><p>In 1998, when Larry Page and Sergey Brin were trying to bring Google to market- what was the key commodity they were dealing with? What was their core product? </p><p>It was a search- we all know it. Over the years, across the journey - they constantly improved, simplified, and made accessible ' Search. </p><p>Before Google, it was possible to do what Google was doing, but it was costlier, more complex, and cumbersome to Search for information, and at times, it needed special access, privileges, and skills to carry out the same. It was less of a search- more of 'Research.'</p><p>With its foundation, Google handed over the power to perform that 'research' in the hands of anybody having access to its search portal. Then there were advancements like Contextual Search, Maps integration, Integration with different apps, etc., to the core search product.</p><p>Of course, Google diversified with time into many other fields, but primarily- Google Search, as we know it, was the first attempt at Artificial Intelligence by machines. There was this search portal that was performing four activities - which any AI will have to perform:</p><p><strong>Sense</strong>: The needs and requirements of the user. This was done through the users' queries for the information they were looking for.</p><p><strong>Comprehend</strong>: Understand what is being requested and the context behind it. This is the part where Natural Language processing is used- to accurately decipher the jumble of words being typed or spoken in any order.</p><p><strong>Respond</strong>: Find the most appropriate results in reference to the user context and present them</p><p><strong>Learn</strong>: The search engine learns from its behavior and how the user clicks successful outcomes to provide feedback on its contextual awareness to produce more accurate and relevant results in the future.</p><p>If we try to look around the use cases of AI operating around us - each one of them is doing the same set of 4 steps, whether it's Alexa, Siri, Maps, or the advanced Fraud detection tools used by your Credit card company or Bank. Try to relate- same four steps. What Google did eventually was the commodification of 'Search' as a capability. The exact process is possible for any AI.</p><h2><span style="font-size: 14pt;">Favorite Avenger and AI</span></h2><p>I think all of us have a favorite Avenger- for me, it's Iron Man. Yes, he is sharp, smart, and Brainiac and knows how to execute stuff in his mind, but for me, his biggest strength is his ability to know near perfect -' division of Labour.' </p><p>When we see Iron Man on screen, we see a perfectly orchestrated act of 3 distinct entities - Tony Stark, The Suit, and Jarvis (the AI). Here it's quite interesting to understand what each is doing. </p><p>The Suit is the power engine, providing different combat and defense capabilities at the command of Tony Stark, as delivered by Jarvis. Jarvis being the AI providing - Prediction (Yes- Prediction, the Suit power is 17% and will last only 30 seconds, I have got two additional missile options left, but the target is out of range to waste the same, etc.,) and Tony Stark - is providing the judgment, around situations- whether to use or not use a particular option. </p><p>Keep in mind the four steps shared in the Google example- Jarvis is doing all of that. What plays out in the form of Iron Man is a perfect example of the "Human + Machine" intelligence working in perfect synchronization to achieve the desired outcomes. </p><p>I used this example- to highlight AI's key role- prediction. Immensely powerful and accurate prediction. The add-on levers like Machine learning, Neural network, etc., only help to strengthen this core capability of prediction.</p><h2><span style="font-size: 14pt;">Why Prediction?</span></h2><p>In very simple terms- Prediction is the process of filling in the missing information. It takes the available information and uses it to generate the information that is not available. When you type in your destination in the Maps application on your Phone, it senses what you need, trying to comprehend the available data on traffic situations, roadblocks, etc., and respond with the best possible route with timeframes. It is also learning, at the same time, what challenges came through.</p><p>In very simple terms- what Google did with 'Search,' AI is doing with 'Prediction,' and the entire focus of corporations engaged in harnessing its power or sharing it with the world is to 'make it cheaper, more accessible and scalable.' </p><p>Just a decade ago, the conceptual prototype of the kind of digital assistants (Google/Siri/Alexa/Cortana, etc.) we use every day would have seemed impossible to acquire for most of us, but now most of them are free and are indeed enabler for a discrete suite of services being sold. This is precisely what happens when a capability is commoditized and made more easily accessible and available. People start using it more and more, and in the case of AI- its ability to predict accurately will increase progressively as every single use instance adds up to its repertoire of data- that it leverages to Respond and Learn.</p><h2><span style="font-size: 14pt;">Importance of Data for AI</span></h2><p>We all know AI relies on data, which is the fuel for AI. But what type of data, so very simply, there are three types:</p><p><strong>Training Data</strong>- this is the data set used to train the AI around basic principles. For example, if you are building an AI engine for Problem Invoice detection, this is the dummy data you will use to train the AI to give examples around the kind of problematic invoices that can come.</p><p><strong>Input Data</strong>- this is the real deal. The actual instances which run through the AI. In reference to the example above, these would be the real Invoices that are sensed, comprehended, and responded to.</p><p><strong>Feedback Data</strong>- this is the data around exceptions, which strengthens the prediction capability of the AI engine. These are the exception invoices, which AI cannot comprehend, but when it flags it for human action, for the next subsequent instance, this data is fed back into the core engine to handle it autonomously in the future.</p><p>Prediction is one part of the equation- but an effective use case also needs judgment. Judgment around what is of value and what is irrelevant. That component is where Human operators come into the picture. </p><p>The machine can predict accurately, but the human providing the judgment will determine where to use that prediction. So here Outcome= Prediction + Judgement. After all, Jarvis can't operate independently- it needs a Tony Stark to seek command from.</p><p>Currently, the entire discourse amongst CXOs is around how to acquire/ develop/ leverage AI? How to bake in AI as a strategic lever etc. But in the coming few years, it is perfectly possible to listen to phrases like "AI of your choice" (just like Cloud of your choice) and "Prediction as a Service." </p><p>Are we ready for it yet?</p><h2><span style="font-size: 14pt;"><strong>Readiness Assessment and the Challenges Ahead</strong></span></h2><p>As mentioned earlier, AI needs data- specifically Input Data. Even in use cases where being used currently- the biggest failure that any AI encounters is the easy, accurate and timely data availability.<br /> <br />It's a real challenge- since most organizations do not capture data in a structured manner or the quantum which will feed AI to deliver real outcomes. Proper structuring of the existing data is another layer of challenge. </p><p>Manual processes are mostly reactive and are not built with a future potential of data leverage in mind, hence will fail to support any investment in emerging technologies like AI. </p><p>So, the question is - where to start? What can be done?</p><p>Based on my own experience working with large corporations across the world, many of which are already farming the benefits of AI and Machine learning, I can say with a certain level of confidence that organizations using structured workflows (with the ability to capture, define data objects in a custom way) are better positioned to derive gains from investments in Artificial intelligence. </p><h2><span style="font-size: 14pt;">Business Process Modeling as a Service</span></h2><p>This is where Business Process Modeling as a Service (BPMaaS) solutions like Pega, IBM BPM on Cloud, etc., is one of the most strategic investments an enterprise can make to be future-ready.<br /> <br />These tools provide the unlimited potential to digitalize their process and formalize the capture of data, offering Integration with existing/ new tools and ERP solutions, connecting data in a way- which is fit to provide an appropriate baseline for Input Data for AI. </p><p>As it will not be possible to go back in the past, and make up for lost data instances, to effectively 'make merry' in an 'AI-enabled competitive landscape,' investments in formalizing the input data capture through BPMaaS and Workflows is a first foundational step. A step that builds the foundation for the entire staircase.</p><p> </p><p><span style="font-size: 10pt;"><em>This article was contributed by our expert <a href="https://www.linkedin.com/in/mohit-sharma-cgma-25764411/">Mohit Sharma</a></em></span></p><p> </p><h3><span style="font-size: 18pt;">Frequently Asked Questions Answered by Mohit Sharma</span></h3><h2><span style="font-size: 12pt;">1. How artificial intelligence AI is changing the marketing landscape? </span></h2><p> </p><p>Marketing is more about reacting quickly to data inputs and using the right moment to qualify a lead. AI, whose primary capability premise is prediction, can help a great deal here. </p><p>Marketing can use AI technologies to make automated decisions about data collection, data analysis, and additional observations of audience or economic trends that may impact marketing efforts. The tools use data and customer profiles to learn how best to communicate with customers, then serve them tailored messages at the right time without intervention from marketing team members, ensuring maximum efficiency. Some common use cases include-Data analysis, Natural language processing, Media buying, Automated decision-making, content generation, Real-time personalization, etc. For example, when you search for an item in Google, through your Phone- Facebook ads, if there are no privacy restrictions, start appearing related to the best offers on those items. There is context sensitiveness around the time of the day you are free to make those decisions.</p><h2><span style="font-size: 12pt;">2. What is the future of AI in business? </span></h2><p>AI is more about prediction and less about automation. AI provides prediction, so a business transaction or incident could be responded to in a more automated way. It is estimated that around 70% cost reduction and 10X productivity gains will be possible in the future through the proper use of AI. Some aspects need to be managed, like Ethical AI- how to eliminate bias from AI algorithms. There has been a good deal of research on this recently, which assures this can be done.</p><h2><span style="font-size: 12pt;">3. How does big data influence AI? </span></h2><p>To make a prediction, you need baseline data- big data is the storehouse that provides the ‘Volume, Velocity, Variety and Veracity’ of data needed to run successful AI solutions delivering committed business outcomes.</p><h2><span style="font-size: 12pt;">4. How does AI analyze data? </span></h2><p>AI uses machine learning and deep learning techniques to identify patterns and trends in the analyzed data and to develop a recommendation or prediction score. That prediction score is used to trigger automated action to provide efficiencies and cost savings.</p><p> </p>
KR Expert - Mohit Sharma
Human insights are irreplaceable in business decision making. Businesses rely on Knowledge Ridge to access valuable insights from custom-vetted experts across diverse specialties and industries globally.
Our flagship service, phone consultations, enables you to get access to first-hand, grass-root level information from our global expert network to form or validate your hypothesis.