Chapitre 1
The Dawn of Machine Intelligence: Thriving in the Fourth Industrial Revolution
Ever felt that strange sensation when your smartphone seems to know what you're thinking before you do? That's just the beginning. "What To Do When Machines Do Everything" isn't just another tech prophecy-it's a practical field guide for navigating what authors Malcolm Frank, Paul Roehrig, and Ben Pring call the Fourth Industrial Revolution. This 2017 bestseller quickly became required reading in boardrooms across industries, with business leaders from GE's Jeff Immelt to Microsoft's Satya Nadella citing its influence on their digital transformation strategies. The book's prescient insights have only grown more relevant as AI has exploded into the mainstream-what seemed futuristic when published now feels like today's business reality. Unlike philosophical treatises on AI's distant future, this book delivers pragmatic advice for thriving alongside intelligent machines in the next five years, positioning the technological shift as comparable to historical transformations led by figures like James Watt and Thomas Edison.
Chapitre 2
From Technological Stall to Digital Boom
Many people feel economically stalled amid stagnant wages and anemic productivity growth. Despite massive technological advances, headlines often foretell job losses and robot overlords. But what if this economic "stall zone" actually signals an upcoming period of unprecedented growth?
According to the authors, we're transitioning from the Third Industrial Revolution to the Fourth, powered by what they call "the new machine"-the combination of artificial intelligence, algorithms, bots, and big data. This coming "digital build-out" will spread technology's benefits from Silicon Valley to the entire economy at unprecedented speed.
Concerns about machines replacing human labor have persisted throughout history. During the First Industrial Revolution, Luddites smashed power looms that threatened their textile jobs. Similarly, agricultural employment in the US dropped from 80% of the workforce in the early 19th century to less than 2% today. Each technological revolution brings similar fears, but history shows these fears typically appear just before technology-led economic booms.
Major technological revolutions follow an S-curve pattern: slow initial adoption (bottom of the curve), explosive growth 25-35 years in (middle), and eventual flattening as the technology matures (top). We're currently transitioning from the stall zone to rapid expansion in the Fourth Industrial Revolution. This explains the contradiction between optimistic technophiles and pessimistic economists-both are right from their limited timeframe perspectives.
Three parallel trends are driving this transition: "Ubiquitech" (technology embedded in everything), the recognition that current systems will seem primitive by 2030 standards, and enterprises mastering the "Three M's" (raw Materials in data, new Machines in AI systems, and business Models that monetize these systems).
Take New Classrooms Innovation Partners as an example. Founded in 2011, they're using data to personalize teaching and break the industrial model of education. Their approach places students in smaller groups working through different learning "modalities" at various "stations," with multiple teachers guiding 60-80 students in larger spaces. Lessons are online, grading happens automatically, and each student receives an individualized learning plan by 6 A.M. the next day based on their progress. This "no back rows" philosophy ensures no student falls through the cracks, bringing personalized education typically reserved for private academies to public schools.
The key to prosperity in the coming digital build-out lies in understanding the new machine and positioning it within the right business model.
Chapitre 3
The Great Job Transformation
While the new machines will help move economies from stall to boom, this transition will involve significant job disruption. However, the authors challenge extreme predictions like Oxford University's forecast that 47% of U.S. jobs could be automated by 2025. Based on their analysis, a more realistic figure is around 12% of jobs being automated away, balanced by approximately 13% new jobs being created directly from new machine growth.
Rather than mass unemployment, they predict massive job transition, with the labor force changing in three ways: job automation (12%), job enhancement (75%), and job creation (13%).
The authors distinguish between automation of manual labor and knowledge work. While industrial automation has eliminated factory jobs that won't return, knowledge automation follows different principles. Knowledge work isn't zero-sum-when automated, knowledge assets can be reused, recombined and enriched, expanding output rather than simply substituting labor. Unlike physical tasks that happen once (like attaching a lug nut), knowledge can be personalized and repurposed countless ways, as demonstrated by today's customized news feeds that would be mathematically impossible for traditional newsrooms to produce.
Doomsday job analyses fail to distinguish between jobs and tasks. Rather than viewing jobs in binary terms (automated or not), knowledge jobs comprise multiple tasks-some automatable, others not. Using an accountant as an example, tasks like documenting transactions might be automated while strategic counsel won't be. Forrester Research's analysis shows that often only 20% of routine job portions will be automated, primarily the tasks humans find boring.
Automation has historically provided net benefits to society despite initial disruption, following three consistent patterns: new abundance is created as automated products become more affordable; overall employment rises despite less labor per unit; and society gains higher living standards. The authors warn against "Luddite thinking" that could make companies noncompetitive, arguing that automation is inevitable and will be positive if managed wisely.
AI will impact knowledge jobs in a "slowly, slowly, suddenly" pattern, gradually automating tasks until reaching tipping points that fundamentally transform job roles. Understanding both the task-based nature of work and the value of remaining human elements helps predict how quickly specific professions will be affected.
Chapitre 4
Understanding the New Machine
At the heart of the Fourth Industrial Revolution are "systems of intelligence"-the technological foundation powering digital leaders like Uber, YouTube, Facebook and Google. These systems combine software (algorithms, machine learning), hardware (servers, sensors), data, and human input to create seemingly magical digital experiences.
Systems of intelligence have three distinguishing features: self-learning software, massive processing power through cloud computing, and unprecedented data collection. The authors clarify that artificial intelligence should be viewed simply as "machines that learn" rather than human imitations. They distinguish between three AI types: Narrow AI (ANI) which is purpose-built for specific tasks and represents all current AI; General AI (AGI) which would have human-like general intelligence; and Super AI which would far exceed human capabilities. Only Narrow AI is relevant for current business applications.
These systems share a common anatomy while differing from traditional systems of record. Though they share elements like user interfaces and databases with older systems, they're distinguished by their learning capabilities, processing power, and data utilization.
The app interface serves as the front door, framing the user experience while keeping the underlying technology invisible. These interfaces must meet the high design standards set by tech giants for elegance and ease of use, explaining the rise of "design thinking" in corporate IT.
The core consists of three elements: digital process logic that transforms manual processes into automated ones; machine intelligence comprising algorithms and learning systems; and a software ecosystem of interconnected tools linked through APIs.
Systems of intelligence rely on diverse data inputs from both traditional systems and real-time instrumentation-the "Code Halos" surrounding products, people, and places. This data comes from sensors in mobile devices, apparel, equipment, cars, and countless other physical entities.
Netflix exemplifies a system of intelligence in action. Its anatomy reveals how each component works together: 75 million subscribers interact through device-agnostic apps; sophisticated process logic delivers content on-demand; machine learning powers recommendation engines processing billions of events daily; an open software ecosystem leverages tools like Java and Hadoop; sensors collect data from various devices; a 10-petabyte data warehouse tracks viewing habits, searches, and ratings; and infrastructure services fully provided by Amazon Web Services.
Truly effective systems of intelligence get smarter over time as they absorb more data; maintain open rather than closed architectures; integrate human expertise rather than attempting to replace it entirely; focus on narrow, specific use cases; provide one-to-one personalized experiences; and are typically bespoke rather than off-the-shelf.
Building your own new machine is becoming increasingly accessible. With just a credit card, you can lease access to machine-learning code, infrastructure, and databases through platforms like Google Cloud Platform or Amazon Machine Learning-technology that would have cost millions just years ago.
Chapitre 5
Data: The New Oil
Data is the primary raw material of this industrial revolution, just as previous revolutions were catalyzed by coal, steel, oil, and electricity. Today's winning organizations have real-time insights that were impossible before: they know exactly how an aircraft engine performs during a specific flight, how interest rate changes will impact bonds before lunchtime, or how a student is performing in today's calculus lesson.
Like oil, data must be mined, refined, and distributed-but unlike oil, it's potentially infinite, multifaceted, and can rapidly grow in value. However, without proper management, it can become worthless or even burdensome. The scale is staggering: a single Airbus A350 generates 2.5 terabytes of sensor data daily-equivalent to 524,000 copies of Shakespeare's complete works.
In the 1850s, oil was viewed primarily as a problematic byproduct of coal mining-brown sticky "muck" with limited applications. It wasn't until Scottish chemist James Young developed efficient distillation processes that oil was reconceptualized as something tremendously valuable. Today's business leaders face a similar reconceptualization challenge with their data. Most struggle with what they perceive as costly, complex data systems with unrealized value.
Like the oil industry's upstream (extraction), midstream (refining), and downstream (distribution) structure, data requires a similar three-pronged supply chain approach. Data proves superior to oil in several critical ways: it's cheap to mine; it's infinite and shareable rather than zero-sum; it's proprietary rather than commoditized; it's inexpensive to distribute; and most importantly, its value grows exponentially rather than linearly with volume.
By 2020, data harvesting and distribution will become standardized "table stakes"-Ford won't gain advantage over Chevrolet merely by instrumenting carburetors. The competitive battleground will be in refining data into meaning through business analytics-the tools, techniques and processes that transform data into actionable insights.
Just as we now view rotary phones as "dumb," by 2025 we'll consider today's uninstrumented objects equally primitive. The miniaturization and improved price/performance of sensors makes it economically viable to instrument virtually everything. The strategic question shifts from "what should we instrument?" to "what shouldn't we instrument?"
Traditional companies with physical assets have a unique advantage in the digital economy. Their "jujitsu move" involves leveraging existing infrastructure as data generators. Century-old firms like American Express, airlines, banks, and insurance companies possess valuable data stockpiles ready for extraction and refinement. However, this advantage is highly perishable, unlike oil which remains unchanged over time. Uber demonstrates this threat by pooling massive data from assets owned by others and applying its platform model beyond taxis to flu shots, food delivery, yachts, and helicopters.
Chapitre 6
Transforming Business Models for the Digital Age
Silicon Valley's digital disruptors are targeting every industry with new machine-powered, data-fueled business models. To compete, companies must transform beyond just adopting technology-they need fundamentally new business models.
The winning business model isn't purely digital but hybrid-blending physical and digital elements appropriately. Airlines will still fly physical planes but digitize passenger experiences and flight operations. Hospitals will maintain emergency rooms while heavily instrumenting their processes.
Based on extensive work with companies undergoing digital transitions, the authors identify four critical traps that derail even well-intentioned digital transformation efforts:
1. The "doing digital" vs. "being digital" trap - Simply adding digital interfaces to industrial processes without fundamentally rethinking business models. This superficial approach never reaches true digital transformation.
2. The FANG trap - Blindly imitating Facebook, Amazon, Netflix and Google is counterproductive because these companies play a different game with different starting points and destinations.
3. The "boil-the-ocean" trap - Going "all in" on massive digital initiatives often fails. Starting small with targeted "digital process acupuncture" on specific pain points builds momentum through quick wins.
4. The digital denial trap - Believing your industry is immune to digital disruption ignores that all industries have different "melting points" but none are immune to information transformation.
For companies working to get ahead in the digital economy, the authors identify five distinct approaches to business model transformation, which they call the AHEAD model: Automate, Halo, Enhance, Abundance, and Discovery.
Industrial-model leaders are investing heavily to become "hybrid winners" in the digital economy. Kevin Plank of Under Armour envisions fundamentally affecting global health through Connected Fitness. Frans van Houten of Philips calls for radically rethinking healthcare delivery through technology. Jeff Immelt of GE declares that industrial companies are now in the information business "whether they want to be or not." The authors frame this as "the management challenge of a generation."
Chapitre 7
Automate: The Foundation of Digital Transformation
Many tasks we once did manually-writing checks at banks, visiting travel agencies, creating fonts for presentations-have already been automated away. This new wave of white-collar automation represents a once-in-a-generation opportunity for managers to fundamentally change their firm's cost structure while simultaneously increasing operational velocity and quality.
The economic case is compelling: rather than driving just 3-6% cost reduction, new machine applications can remove 30-60% or more from operational costs. Companies like TriZetto are using software robots to decrease healthcare payer costs by up to 90% for some middle-office processes.
Automation isn't new-we've been consuming it for decades without noticing. Think about a typical airport journey: passing through E-ZPass tollbooths, using automated parking ticket dispensers, airline check-in kiosks, and ATMs. Just twenty years ago, all these touchpoints required human workers.
Automation's greatest potential lies in the back and middle offices of organizations-areas tucked away from customers where core operations happen. The back office includes supporting functions like IT, finance, HR, and administration, while the middle office encompasses industry-specific processes like claims processing in insurance or trade settlement in banking.
The authors provide four practical "rules of the road" to kick-start automation initiatives:
1. Set your 25%-25% automation imperative - Challenge teams to achieve 25% cost reduction with 25% productivity increase, forcing them to think beyond traditional approaches.
2. Find your process-automation targets - Focus on highly repetitive tasks performed at scale, tasks with low demand for human judgment, tasks requiring low levels of empathy, and tasks generating high volumes of data.
3. Break through the "brass wall" - Address resistance from middle managers who fear automation threatens their positions by explaining automation's organizational necessity.
4. Build a repeatable process to obliterate work - Follow seven steps: set bold goals, start small with specific pressure points, apply existing automation tools, develop prototypes, pilot and scale gradually, analyze results honestly, and repeat continuously.
Robotic process automation is our modern equivalent of the loom or steam engine-a transformative technology that generates both cost savings and valuable data. These savings provide the capital needed to invest in new markets and ideas, while the data enables better products and customer relationships.
Chapitre 8
Halo: Creating Digital Value from Physical Assets
"Code Halos" represent the idea that any person, place, or thing has both a physical self and a virtual self. When instrumented and tracked, objects develop an invisible halo of code around them, creating a "digital twin" that often provides more insight and value than the physical item itself.
This is how companies like Amazon and Netflix can understand consumer preferences better than friends and family do, despite never meeting customers in person. The cost of connecting objects to the internet is now so low and the potential value so high that it's practically corporate malpractice not to instrument everything.
South African insurance company Discovery Limited transformed their business by connecting insurance with wellness tracking. By collecting data on customer health behaviors through apps and devices, they offer benefits like discounts on healthy food and Apple Watches to members who maintain healthy lifestyles.
The authors identify three key principles for successfully competing with code: 1) Instrumentation is no longer elective but core curriculum; 2) Code is more valuable than things, transforming traditional products into relationship-building platforms; and 3) Connections must be always-on rather than episodic to generate exponential value.
"Know-It-All" businesses use sensors and instrumentation to collect and analyze information about everything, contrasting with traditional approaches where businesses had to rely on guesses, hunches, or delayed information. GE exemplifies a 125-year-old company reinventing itself by instrumenting its industrial equipment. GE's Tier 4 locomotive, equipped with over 200 sensors, creates a "rolling data center" that enables zero unscheduled downtime through predictive maintenance.
The authors provide practical guidance for organizations to start capitalizing on code:
1. Buy an extra-large box of chips - Liberally apply sensors to physical objects as the first step in building effective systems of intelligence.
2. Find a squad of data scientists - Once instrumentation generates massive data, organizations need data scientists to extract business meaning.
3. Build a halo business model - Establish clear links between insight and value, following the money to demonstrate ROI.
4. Redesign your customer experiences - Design is no longer an afterthought but central to success, as demonstrated by Apple's design philosophy.
5. Keep away from the dark side - Maintain transparency and trust in data relationships with customers by clearly explaining the value exchange.
6. Monetize your data - Leverage data assets to create entirely new revenue streams with enormous margins.
Chapitre 9
Enhance: Human-Machine Collaboration
Consider navigating to a meeting in 2003 versus 2017. In 2003, a driver gets lost using printed directions, misses highway exits, and fears being late. In 2017, the same journey is guided by Waze's turn-by-turn navigation, ensuring on-time arrival despite traffic. This example illustrates how technology now enhances our capabilities, making previously difficult tasks effortless.
Human evolution has always been tied to our tools, from primitive stone implements to today's advanced computing systems. These tools have consistently enhanced our capabilities, allowing us to exceed our natural physical and mental limitations. Now, AI-infused digital assistants are expanding this enhancement into office environments, helping workers manage their time and responsibilities more efficiently.
The core premise of enhancement is that nearly every job can and must be improved through technology. Even knowledge workers already using technology extensively can be further enhanced-automating administrative tasks like expense reports, using digital assistants for scheduling, or employing voice commands for presentations.
While some see only the downsides of automation as a "race against the machine," the authors emphasize we're actually in a "race with the machine." Technology has always both given and taken away jobs, but this dynamic ultimately drives efficiency, productivity, and higher-value work.
McGraw-Hill Education's ALEKS system exemplifies enhancement technology in education. Based on Knowledge Space Theory, ALEKS creates personalized learning pathways through dynamic knowledge maps that adapt to each student's abilities. The system handles routine grading and analytics, freeing teachers to focus on meaningful instruction, small-group interaction, and creating engaging experiences.
The symbiotic relationship between humans and intelligent machines is illustrated by the famous Go match between world champion Lee Sedol and Google's AlphaGo. Though AlphaGo won the series, Sedol's surprising "Move 78" in game four-which even AlphaGo didn't anticipate-demonstrated how humans can learn from AI. Sedol himself acknowledged that playing against AlphaGo "opened his eyes" to new strategies.
To begin enhancing your work, the authors recommend two major steps:
1. Double-down on being more human - As technology handles routine tasks, human qualities become more valuable. Companies like Zappos, Pret A Manger, and Apple demonstrate how automation can free employees to deliver more humanity, not less.
2. Build your white-collar exoskeleton - While physical exoskeletons enhance our bodies, "exoskeletons for the brain" are enhancing intellectual work. Palantir Technologies creates products that "make people better at their most important work." In fields like radiology, humans partner with AI systems like IBM's Avicenna that analyze massive amounts of patient data to improve diagnostic accuracy.
We're becoming smarter because our tools are becoming smarter. While humans today aren't inherently more intelligent than Aristotle or Shakespeare, our tools certainly are. These intelligent tools drive our progress.
Chapitre 10
Abundance: Creating New Markets Through Affordability
The concept of abundance is simple economics: as prices drop, demand rises. While AI-generated abundance is new, the underlying principle has driven every industrial revolution-from looms creating abundant clothing to assembly lines making refrigerators ubiquitous. Before these innovations, such products were rare luxuries; afterward, they became democratized.
Narayana Health (NH), founded in 2000 by Dr. Devi Shetty in India, exemplifies abundance through digital technology. By applying systems of intelligence to cardiac surgery processes, NH has reduced costs approximately a hundred-fold, providing bypass surgery for around $1,200 compared to $100,000 in the United States-with comparable mortality and infection rates.
As core processes become instrumented and digitized, entirely new thresholds of price, quality, and customization emerge. Companies like Spotify and WeWork create abundance by automating the matching of supply and demand.
Setting new price points isn't a one-time event but a continuous process. As products and services transition from human-based to machine-based delivery, they become tech-centric and subject to Moore's Law-the foundation of innovation and affordability in technology for decades. While insurance policies, doctor's visits, and educational services have maintained relatively stable prices for 20 years, that's about to change as these offerings become digital to the core.
To kick-start momentum toward abundance, seven approaches have proven effective:
1. Obsess about the start-up community - Create teams focused exclusively on tech startups targeting your business to reveal emerging threats and opportunities.
2. Kill your company - Ask your sharpest employees to develop ideas that could put your company out of business, leveraging their perspective on the gap between their seamless digital personal lives and outdated work processes.
3. Play the "tomorrow it's free" game - Imagine your premium products at 10% of current prices to prepare for abundance markets.
4. Manage your innovator's dilemma - Use frameworks like the Three Horizons Model to separate business units by timeframe.
5. Make like a maker - Harness the talents of the maker movement by embracing employees who tinker with new technologies.
6. Think like a corner shop - Focus on personalization at scale, creating one-to-one connections with customers using systems of intelligence.
7. Apply digital Taylorism - Use AI and analytics to optimize knowledge work the way scientific management revolutionized manufacturing a century ago.
Chapitre 11
Discovery: Innovation for the Digital Economy
Discovery represents both a catalyst for and outcome of the AHEAD model. It's not merely a side project but a central philosophy and rigorous practice essential for remaining relevant in the great digital build-out. While machines will increasingly handle current work, innovation will allow us to discover entirely new opportunities that are currently impossible to imagine but will form the core of future human work.
The new machine is becoming the essential platform for innovation. When businesses instrument, automate, track, and analyze their core operations using machine learning, they unearth innovation opportunities based on empirical data rather than informed opinions.
Netflix exemplifies how AI drives innovation at unprecedented speed and scale. Rather than relying on assumptions about regional viewing preferences, Netflix deploys algorithms to understand what works globally. Their team discovered that age, gender, and geography matter less than actual viewing behavior-a 19-year-old might enjoy documentaries about wedding dresses, and Japanese anime finds audiences worldwide.
Even the world's largest investors recognize the importance of discovery over short-term results. BlackRock CEO Larry Fink has urged S&P 500 companies to focus on long-term value creation rather than quarterly earnings, emphasizing the need to embrace technological innovation.
The key strategy for discovery is to embrace human imagination. Despite techno-dystopian fears, our fundamental curiosity-the DNA attribute that defines human intelligence-will continue driving innovation. To kick-start discovery:
1. Apply Digital Kaizen - Focus on small, continuous improvements enabled by the new machine that accumulate into significant impact. The University of Kentucky demonstrates this approach by using real-time analytics to improve student retention by 1.3%-a seemingly small improvement with substantial downstream impact.
2. Let Hits Pay for Misses - Establish a portfolio of discovery initiatives with clear lifecycle methodology, balancing incremental digital Kaizen with blue-sky innovation. Even with data-driven insights, the future remains unpredictable-75% of venture-backed firms don't return capital, and 70% of movies lose money.
3. Leave the Past Behind - Legacy technology severely hampers discovery initiatives in Fortune 500 companies. Without clearing away systems no longer fit for purpose, organizations undermine their ability to find budget, time, and energy to invest in the future.
4. Play the Wayback Game - The Internet Archive Wayback Machine offers powerful perspective on technological change. Websites from just 20 years ago feel as dated as wool bathing suits. Humans struggle to extrapolate future developments, believing we're at the zenith of development when we're merely at a camp on an infinitely high mountain.
The next 20 years will produce a new generation of iconic innovators who transform industries by connecting the Three M's and applying the AHEAD model. Though we cannot predict who they'll be or what they'll create, they will share a fundamental belief that something better can be created.
Chapitre 12
A Pragmatic Call to Action
As AI debates intensify between utopians who see technological marvels and dystopians who fear malevolent machines, pragmatists must take control. Neither extreme view will prevail-the future will contain both opportunity and disruption. Intelligent machines will inevitably permeate all software and physical products, while systems of intelligence will expose non-intelligent systems.
The digital innovations of the past 70 years are merely precursors to the revolutionary impacts coming in the next 15 years. Just as 1870s England was utterly transformed from the 1840s, the compounding nature of digital progress will fundamentally metamorphose our world. Many institutions that have barely changed in decades will become unrecognizable by 2030.
As the old S-curve declines, companies mastering the Three M's will lead the new wave: aligning the new raw materials (data), new machines (systems of intelligence), and new models (data-based personalization). Even when machines can do everything, people remain the ultimate X factor. It's still humans who must decide to instrument everything, harvest data, teach algorithms, invest in technology, and make difficult decisions about product lifecycles.
The AHEAD model provides clear direction: Automate everything possible to strip costs and improve quality; Instrument everything to generate previously invisible data; Enhance every person to improve performance across your organization; drive prices down to increase market size toward abundance; and Discover possible futures through innovation.
This transformative era demands courage, not timidity. Recent headlines illustrate the pace of change: e-sports broadcasts outperforming NBA Finals viewership, traditional manufacturers like GM placing bets on future car-sharing models, business speed becoming the new currency, blockchain revolutionizing financial markets, AI composing Beatles-style music, and quantum teleportation advancing.
Rather than fearing AI, we must embrace it as "the next level of productivity tool." Innovation has always propelled humanity forward, and AI's development cannot be inhibited. The winners in this digital build-out will be those who stop debating and start building-partnering with machines to invent the future rather than merely predicting it.
