Digital Transformation, Innovation
and The AI Dialectic
By Peter B. Ritz and Dr. Rajarshi Das
The dawn of general purpose computing foretold an unrequited dream of giving machines intelligence. Partly moved by the notions of intelligence explored through the Turing test Turing, A.M., Computing Machinery and Intelligence. Mind, 59, 433-460. (1950), popularized today on many web sites by the combination of Captcha and chatbot user experience extensively, see also the formal instantiation of the Turing test as part of the Loebner Prize , IBM’s S/360 mainframe – a pioneering general-purpose computing platform designed to drive a full range of commercial and scientific apps – demoed examples of nascent artificial intelligence (AI) through handwriting recognition and language translation at the 1964 New York World’s Fair Luke Dormehl, Thinking Machines: The Quest for Artificial Intelligence — and Where It’s Taking Us Next (2017) . S/360 went on to inspire Stanley Kubrick’s hyper-intelligent if psychotic machine lead HAL 9000 Aisha Harris, Is HAL Really IBM? Slate’s Culture Blog, Slate (January 7, 2013) in his 1968 masterpiece, 2001: A Space Odyssey. IBM S/360 also became the initial member of the world’s most transformative business growth enabler. In the following decades, while AI played only bit roles outside Hollywood, Intel and Microsoft joined IBM in democratizing PCs with hardcoded software for business masses. This marginal role for AI ensued while the disruptive advances in client server, Internet, mobile, social and cloud drove a virtuous innovation cycle. Enterprise resource planning (ERP), customer relationship management (CRM), financial and collaboration apps also grew without significant AI impact to optimize the digital economy. Yet, the world’s business infrastructure was revolutionized and global productivity exploded.
The last half-century records a profitable trajectory for how digital innovations drive business growth. The next phase in this cycle promises to be turbo-charged by AI to learn business outcomes at hyperscale See, FatBrain’s blog post “Google's GCP Next Signals Hyperscale AI for the Enterprise: Modern Enterprise Architecture means Software that Learns Business Outcomes,” highlighting Google's long-game corporate commitment to enterprise, poignantly and pointedly reaffirmed by its executive chair Eric Schmidt, in the keynote address (2017) orchestrated via cloud, where the geometric growth in data powers transformations in apps, and vice versa. We call this cycle the “AI Dialectic” The approach was inspired, in part, by Karl R. Popper’s “Conjectures and Refutations,” Routledge & Kegan Paul, London (1963), p 312-355, first published in Mind N.S., vol. 49 (1940), as well as a presentation by Kathryn Hume at NYU Future Labs AI Summit (April 5, 2017) and a thoughtful paper by Peter Wagner and Martin Giles, “Microservices, Containers and The Digital Dialectic”,(2016) .
The latest escalating turn in this Dialectic revolves around software that learns rather than counts and fundamentally aims to automate the scientific process At least one ancient meaning of Dialectic goes back to Plato, reflecting a ‘scientific method’ used to construct explanatory theories and the critical discussion of these theories to account for the empirical observations. Id, Popper, fn. 4 . These principled innovations are driving a radically different way to use software, dubbed machine learning (ML) and business AI We distinguish business AI from general AI, compare for example Netfilx’ movie recommendations (business AI) vs. HAL in 2001: Space Odyssey, Ibid., and in ex-Machina (general AI) . At their core, business AI services package advanced algorithms with modern architectures. This enables enterprises to process images, speech and natural language, to make predictions and recommendations, and to infer optimal strategies and to learn outcomes without legions of PhD data scientists and cloud engineers, obviating the corresponding costs and hiring hassles. Just as the last century’s IT systems revolutionized productivity, so the combination of AI, big data and cloud will transform Matt Turck reaches a similar conclusion in Firing on All Cylinders: The 2017 Big Data Landscape, “With the killer combination of Big Data and AI, we’re heading towards the “harvesting” part of the cycle. Beyond all the hype, the possibilities are enormous.” (April 7, 2013) not only how enterprises develop apps, but also how they achieve business outcomes and overtake competitors. The data geometry and scale connecting the millennial world are breaching the hardcoded limits of enterprise IT systems unable to cope with real-time, “celebrity treatment” demands of consumerized experience. Multi-trillion dollar markets comprising digital transformation enabled by the business AI services are up for grabs CB Insights, State of AI Report, e.g., quoting Andrew Ng, former Chief Scientist at Baidu and co-founder at Coursera, “Just as electricity transformed everything almost 100 years ago, I actually have a hard time thinking of an industry that AI will not transform in the next few years.” (March 2017), see, also, Id. Turck .
Delving into the AI Dialectic
Data, compute and software have been evolving in a geared push-pull relationship since the 1960s. Driven by markets and science, innovations in these systems have made computing cheaper and faster. These shifts successively spark modernizations in app architectures and customer experience, which in turn fuel optimizations in infrastructure and devops fabric to support the new app paradigms at hyperscale.
The figure above highlights how the app architectures evolved from the monolithic mainframes to open systems in the 1990s, triggering software innovations in client-server as well as N-tier design comprising web front-ends, business logic modules, app server middleware and database back-ends. In 2000s, virtualization further drove the IT makeover, enabling businesses to share compute between apps and laying the foundation for public and private clouds. Recently, the microservices and container tech underpinning composable and serverless modern architectures started to enable the cloud-first/native efforts of enterprise devops transformation. Significantly, minimal change to the business data and enterprise workloads was a key success factor in recent growth of virtualization and early cloud deployments.
Hardcoded rules driving enterprise workloads will not work with millennial data geometry and scale. App development cycles take too long and cost too much to keep up with the firehose of continuously changing, multidimensional data. Outside the enterprise comfort zone, hyperscale giants such as Google, Amazon, Facebook and Microsoft are leading the way to commercialize AI Google et al are using business AI to dynamically adjust strategic decision making to the millennial data geometry, consider e.g., Google’s Search switched using hardcoded rules for AI, such that now searching for “Giants” automatically infers football if done in NYC, vs. baseball if searched in SFO. . Forced to accelerate their own needs for scale and speed via internal R&D efforts, many are open sourcing their results, such as Google’s Tensor Flow See, Google Research Blog (Nov 9, 2015) , see also for deep learning.
Practical AI initiatives for the enterprise
What will the transition to business AI thinking The term ‘business AI thinking” or BAIT is inspired in part by Twilio CEO Jeff Lawson’s 2016 AWS (re) Invent Keynote presentation, alluding to the next generation of Design Thinking (see, e.g., IDEO Tim Brown’s Jun 2008 HBR eponymous article), which Lawson coined as Software Thinking, explaining it means “using software to build a competitive advantage in business” mean in terms of practical steps towards digital transformation and innovation for the enterprise? The strategic corporate push by Google into the “AI-first” cloud gear See Ibid fn. 4. See also, Ibid, CB Insights, State of AI Report and Id. Turck is a strong signal that an arms race is under way to capture the cognitive enterprise markets, where other contenders include AWS and IBM Watson at the AI-friendly cloud level, as well as Palantir leading the usual cadre of consultants aiming for the professional services and systems integration business. Significantly, none of these players control a critical tech toll point This, as expected, generally parallels the dynamics of the open source and cloud-native movements in modern architecture app development cycle – rather, value resides in orchestrating and packaging business AI thinking to marshal the millennial data geometry to continuously effect business outcomes. This sets the stage for a new wave of software services purpose-built for the emerging reality. We use the term "Structured Theme Substrates" or STS Themes bind words and behavior into signals and insights based on billions of web pages and decades of search intent, reflecting digital pathways of transactions and engagement, while also enabling a principled monetization of hyper-personal narratives by brand, goal, category, or demographics. to describe the family of Cognitive Theme technologies underpinning these innovations. Theme tech uses casual analytics to automatically explore and exploit actions based on observations Smart companies have moved to an agile 'sense and respond' approach, enabling quick adaption to a rapidly changing environment. By continuously learning from real-time data about customers, competitors, suppliers and operations, companies can principally evolve their strategies, product offerings and profitability. Indeed, the smartest firms run real-time experiments to test different policies and products. For example, Internet companies routinely run A/B tests: presenting customers with different interfaces, measuring which is most effective, then adopting the most successful. Tom Mitchell & Erik Brynjolfsson, Track how technology is transforming work, Nature 544, 290–292 (20 April 2017) . Here are some early examples of using Theme tech to get started and to operationalize business AI thinking within the enterprise.
The figure above illustrates a classic enterprise goal of attaining lifetime loyalty during their customer journey and beyond. FatBrain’s STS services enable any enterprise to quickly and cost-effectively deploy a principled way to understand and rank one or more areas of the journey e.g., churn risk or spending trends, or customer acquisition costs. Moreover, over time and as driven by business priorities, other sources of business data independent of its geometry or scale may be transitioned onto the same common STS denominator. This means principled, automated and strategic decision-making which spans both “what is” and “what if” to drive the desired business outcomes, without changing or refactoring existing enterprise software or using an army of T&M consultants.
Eyes on the AI prize
Prior phases of the AI Dialectic have given rise to new sets of companies with significant market clout due to productivity and technology differentiation. We predict this cycle will yield similar results. The opportunity is now for many leading enterprises and technology innovators to engage with business AI thinking to purposefully and strategically take advantage of the millennial data geometry and its corresponding customer demands. This is when disruptive innovation value is created. We’ve all seen such trends sustain strong growth before: the open systems movement; the commercialization of the Internet; the virtualization and cloudification of enterprise IT. Now the multi-trillion dollar fabric of business transformation itself is in play—and the opportunities couldn’t be more exciting.
Home - About - Contact