martes, 23 de junio de 2026

LIVING BEYOND EARTH: THE GREAT SIMULATION TRAINING HUMANITY FOR MARS

The next space race is not about reaching another planet. It is about learning how to survive there.

For more than half a century, humanity’s space ambitions were defined by a simple question:

Can we get there?

We built enormous rockets, launched robotic explorers, created permanent orbital laboratories, and placed machines on Mars capable of revealing the secrets of another world.

But the next chapter of space exploration is asking a far more difficult question:

Can we live there?

Because reaching another planet is an engineering challenge.

Living there is a civilization challenge.

The first explorers of Mars will not only need advanced spacecraft, artificial intelligence, robotics, and life-support systems. They will need something much harder to engineer: human resilience.

They will have to solve problems without immediate help from Earth, cooperate under extreme pressure, manage isolation, repair failing systems, and make decisions when there is no instruction manual.

The future of space exploration will depend not only on rockets and computers, but on the ability of humans to function as a team in environments where failure can become catastrophic.

This is why scientists, space agencies, and private organizations are creating a new generation of “space rehearsals” on Earth.

One of the most ambitious experiments is the World’s Biggest Analog, a global project that connected 16 simulated space missions across different locations on Earth. Over two weeks, 76 participants lived inside habitats designed to recreate the conditions of future missions to the Moon and Mars.

The experiment delivered a powerful message:

The greatest challenge of becoming a multiplanetary species may not be technology. It may be ourselves.

Mars Is Not Only a Planetary Destination — It Is a Human Stress Test

Located in Utah’s desert landscape, the Mars Desert Research Station (MDRS) provides one of Earth’s closest analog environments to Mars.

The landscape is dry, isolated, and visually similar to the Red Planet. Participants wear simulated space suits, operate rover vehicles, collect geological samples, perform emergency drills, and follow routines inspired by real planetary exploration missions.

Among the participants was photographer and National Geographic Explorer Mackenzie Calle, who experienced the unusual transition from observer to astronaut simulation participant.

But the most important discoveries were not about equipment.

They were about behavior.

A future Mars crew will face a reality very different from Apollo-era missions.

During the Apollo program, astronauts were in constant communication with mission control. Future deep-space explorers will experience communication delays, limited external support, and greater independence.

The astronaut of tomorrow must not simply follow instructions.

The astronaut of tomorrow must become a decision-maker.

The New Astronaut: Engineer, Scientist, Leader, and Human Being

A Mars mission will require people with multiple abilities.

A crew member may need to:

Repair critical equipment.
Conduct scientific research.
Provide medical assistance.
Maintain psychological stability.
Resolve conflicts.
Adapt to unexpected failures.

Space agencies increasingly recognize that technical excellence alone is insufficient.

A brilliant engineer who cannot collaborate may become a mission risk.

A person with average technical skills but exceptional adaptability may become essential.

The selection process for future crews focuses increasingly on qualities such as:

Emotional intelligence.
Communication.
Learning ability.
Adaptability.
Team cooperation.

Emily Apollonio, involved in selecting participants for the analog missions, emphasized the importance of finding people who are “coachable” — individuals willing to learn and work effectively with others.

The future astronaut will not be the lone hero.

The future astronaut will be the ultimate team player.

The Moon: Humanity’s First Off-Earth Laboratory

While MDRS simulated Mars, another experiment, LunAres, focused on the challenges of living on the Moon.

The Moon represents humanity’s first realistic opportunity to establish a permanent presence beyond Earth.

Its proximity makes it an ideal testing ground.

A lunar base could teach humanity how to:

Build extraterrestrial habitats.
Produce resources locally.
Maintain closed ecosystems.
Grow food away from Earth.
Operate autonomous systems.

At LunAres, researchers experimented with growing microgreens under artificial conditions, exploring ways to extend food supplies beyond traditional astronaut rations.

This reflects a major shift in space exploration.

Early explorers focused on survival.

Future explorers must learn sustainability.

The Real Enemy: Complexity

The challenge of Mars is not one single problem.

It is thousands of interconnected problems.

A human settlement on another planet requires:

Energy systems.
Food production.
Communication networks.
Medical capability.
Engineering maintenance.
Psychological support.
Scientific operations.

Everything must work millions of kilometers away from Earth.

The World’s Biggest Analog also tested how future space missions might coordinate internationally.

A mission coordination center in Vienna connected different simulation sites, collected mission data, shared operational information, and helped maintain communication between teams across continents.

The lesson:

Space exploration will not belong to one country.

It will become a global ecosystem.

As Gernot Groemer from the Austrian Space Forum explained, space exploration is a team effort involving different cultures, disciplines, and generations.

When a Simulation Becomes Real

One of the most revealing moments occurred during a rover expedition at MDRS.

Two crew members exploring the terrain unexpectedly lost communication with their base.

Although the mission was simulated, the problem was real.

They were isolated.

Without reliable navigation.

Without immediate assistance.

They had to stop, analyze the situation, communicate clearly, and decide together what to do.

Eventually, they identified terrain markers and safely returned.

The lesson was simple:

Technology can fail.

Human judgment cannot be replaced.

The Human Factor: We Will Take Earth With Us

There is a romantic idea that Mars will represent a fresh beginning.

But simulations reveal something different.

Humanity will carry its nature into space.

Future astronauts will not only conduct experiments.

They will create friendships, traditions, humor, and social rituals.

During simulations, crews created shared experiences such as group activities, jokes, and communal meals to maintain morale.

These moments may appear insignificant.

They are not.

In isolated environments, culture becomes a survival system.

A Mars settlement will not be built only with technology.

It will be built with human connection.

The Next Evolution of Humanity

The first permanent human presence beyond Earth will represent one of the greatest transformations in history.

For thousands of years, humans have been a planetary species.

Now we are developing the skills to become a multiplanetary civilization.

But before building cities on Mars, humanity must answer deeper questions:

How will isolated communities govern themselves?

How will conflicts be resolved?

How will humans maintain identity away from Earth?

How will technology and humanity coexist?

The analog missions happening today are valuable because they allow humanity to fail safely before attempting the impossible.

The future of space exploration will not be defined only by landing on another world.

It will be defined by learning how to belong there.

Glossary: Key Concepts

Analog Mission

A simulation on Earth designed to recreate conditions of space exploration, allowing researchers to study human behavior, technology, and operations.

Mars Desert Research Station (MDRS)

A Mars simulation facility located in Utah, USA, used to study planetary exploration procedures.

LunAres

A Polish analog habitat designed to simulate lunar living conditions and investigate human performance in isolation.

Deep Space Exploration

Space missions beyond low Earth orbit, including travel to the Moon, Mars, and other planetary destinations.

Crew Autonomy

The ability of astronauts to make decisions independently without constant support from Earth.

Closed-Loop Life Support System

A technology approach where resources such as water, air, and food are recycled within a spacecraft or habitat.

Space Habitat

A structure designed to support human life outside Earth.

Mission Control

A ground-based team responsible for supporting and monitoring space missions.

Human Factors Engineering

The discipline focused on designing systems that consider human psychology, behavior, and limitations.

Multiplanetary Civilization

A civilization capable of maintaining human settlements on more than one planet.

References and Further Reading

National Geographic Magazine – July 2026
“This Isn’t Outer Space. Yet.”
Coverage of the World’s Biggest Analog and future human missions to Moon and Mars.
NASA Human Research Program
Research on human health, psychology, and performance during long-duration space missions.
NASA Artemis Program
Lunar exploration initiative designed as preparation for future Mars missions.
The Austrian Space Forum (OeWF)
International organization conducting Mars analog missions and planetary exploration research.
Chris Hadfield – An Astronaut’s Guide to Life on Earth
Insights into astronaut training, leadership, and decision-making.
Scott Kelly – Endurance: A Year in Space, A Lifetime of Discovery
Lessons from long-duration human spaceflight.
Andy Weir – The Martian
Fictional exploration of survival, engineering, and human problem-solving on Mars.
Robert Zubrin – The Case for Mars
A foundational argument for human exploration and settlement of Mars.
Elon Musk / SpaceX Mars Architecture Concepts
Private-sector approaches toward creating sustainable human presence on Mars.

Final Idea:
The first humans on Mars will not only prove that we can leave Earth. They will prove that we can take the best parts of humanity with us.

sábado, 20 de junio de 2026

The New AI Empire: How Frontier Models Became the Next Strategic Weapon

Introduction: The New Oil Is Not Underground — It Is Inside Artificial Intelligence Models

For decades, global power was defined by control over physical resources: oil, gas, minerals, trade routes, and military capabilities. In the 21st century, a new category of strategic power has emerged:

the ability to create, train, and control advanced artificial intelligence systems capable of reasoning, programming, designing, researching, and making complex decisions.

Frontier AI is no longer merely a technological product. It is becoming a strategic infrastructure comparable to nuclear technology, satellites, and the internet.

The central question of the AI era is no longer:

Who has the best technology?

It is:

Who decides who can access the most powerful intelligence systems?

This is the core argument behind The Economist’s analysis of America’s AI power grab: the United States has discovered that controlling access to the world’s most advanced AI models can become a new source of geopolitical leverage.

1. Anthropic: The Company That Turned AI Safety Into a Mission

Anthropic was founded in 2021 by a group of researchers who left OpenAI because they believed the race toward advanced artificial intelligence was moving faster than society’s ability to control its consequences.

Its leadership, especially Dario Amodei, represents a particular philosophy:

AI will become one of humanity’s most transformative technologies, but its power must be developed with safeguards similar to those used in aviation, nuclear energy, and medicine.

Anthropic’s identity is built around a simple idea:

AI can:

accelerate scientific discovery,
improve productivity,
solve complex problems,

but it can also:

enable cyberattacks,
amplify misinformation,
increase biological risks,
transform warfare.

Anthropic is not simply selling software.

It is selling a vision:

Powerful intelligence requires responsible governance.

2. The Silicon Valley vs Washington Conflict: Who Controls AI?

The central conflict began when the U.S. government moved to restrict foreign access to Anthropic’s most advanced AI models, including Fable and Mythos, citing national security concerns.

This creates a historical question:

When a technology changes the balance of global power, should control belong to governments or to the companies that create it?

History provides several examples:

Nuclear technology became state-controlled.
Cryptography exports were restricted.
Advanced military systems remain tightly regulated.

But AI is different.

A nuclear weapon is a physical object.

AI is software.

It can be copied.
Modified.
Distributed.

The challenge is that artificial intelligence is a digital technology capable of escaping traditional forms of control.

3. From Software Product to Geopolitical Infrastructure

The most important idea emerging from this debate is that AI is moving from being a technology product into becoming a global infrastructure of power.

The future question may not be:

“What computer do you own?”

but:

“What intelligence system are you allowed to use?”

A country with access to the most advanced AI models could:

accelerate drug discovery,
automate industries,
improve military systems,
increase productivity,
strengthen scientific research.

A country without access could become technologically dependent.

This creates a new type of dependency:

cognitive dependency.

In the past, nations depended on foreign oil.

In the future, they may depend on foreign intelligence.

4. The AI Economy: The Winner Controls the Ecosystem

Anthropic followed a different strategy from many competitors.

While other companies focused on consumer AI assistants, Anthropic concentrated on enterprise customers, especially software development and business automation.

This strategy proved powerful.

Developers adopted tools such as Claude Code because they can perform tasks that previously required hours or days of human effort.

The next AI battle will not simply be about creating the smartest chatbot.

It will be about becoming the invisible intelligence layer behind global businesses.

The future belongs to:

AI agents,
autonomous systems,
digital workers,
intelligent enterprise platforms.

The next generation of software will not just be something humans use.

It will become something that works alongside humans.

5. Compute: The New Oil of Artificial Intelligence

One of the most important insights is that America’s advantage does not come only from better AI models.

It comes from computational power.

Advanced AI requires:

specialized chips,
enormous data centers,
massive energy supplies,
global networks.

The geography of computing power is becoming the new geography of influence.

The United States currently dominates AI infrastructure, while other regions remain far behind in computing capacity.

The future competition will not only be about building models.

It will be about building:

factories of intelligence.

6. Europe’s AI Sovereignty Challenge

Europe faces a difficult strategic choice.

Regulating AI is not enough.

A region that regulates without building technological capability risks becoming dependent.

Europe needs three major transformations:

1. Build domestic AI capacity

Through:

investment,
talent attraction,
research,
startups,
computing infrastructure.

2. Create technology alliances

Complete independence is unrealistic.

The future model is:

sovereignty through interdependence.

3. Reform institutions

The speed of AI requires:

faster decision-making,
stronger investment incentives,
less bureaucracy.

The question is not:

“How can Europe avoid AI?”

The question is:

“How can Europe shape AI?”

7. The Security Paradox: Excessive Control Can Destroy Leadership

There is a fundamental contradiction.

If America restricts AI too aggressively, it could weaken its own advantage.

Why?

Because technological ecosystems grow through adoption, experimentation, and global interaction.

History shows:

The internet expanded because it was open.
Cryptography spread because it could not be contained.

AI may follow the same pattern.

Too much control may create competitors instead of preventing them.

8. The New Cold War: Intelligence Against Intelligence

The emerging technological competition involves three major players.

But the key difference is:

AI is not just an industry.

It is becoming a national capability.

9. Emerging Trends Shaping the AI Future

Agentic AI

The next revolution will not be chatting with AI.

It will be delegating work.

AI agents will:

write software,
manage processes,
conduct research,
operate businesses.

AI will evolve from assistant to autonomous collaborator.

AI + Robotics

Artificial intelligence will move from the digital world into the physical world.

The combination of:

AI + sensors + robotics

will transform:

manufacturing,
logistics,
healthcare,
defense.

AI as a Scientific Discovery Engine

Future AI systems may become tools for scientific breakthroughs:

designing molecules,
discovering materials,
analyzing complex systems.

AI could become as important to science as the microscope or telescope.

Conclusion: The Future Belongs to Those Who Control Intelligence

History remembers those who controlled:

oil,
oceans,
nuclear power,
the internet.

The next strategic question will be:

Who controls artificial intelligence?

The conflict between Anthropic and Washington is not merely a corporate dispute.

It is the first chapter of a new era where governments, companies, and societies compete to define who controls humanity’s most powerful intellectual technology.

Artificial intelligence will be a tool.

But it will also become a territory.

And, as history has repeatedly shown:

those who control the frontier shape the future.

Glossary

Frontier AI
The most advanced artificial intelligence systems available at a given moment.

Compute
The computational power required to train and operate AI models.

AI Agent
An AI system capable of performing complex tasks with limited human intervention.

AGI (Artificial General Intelligence)
A theoretical AI system with broad capabilities comparable to or exceeding human intelligence.

AI Governance
The policies, rules, and institutions designed to manage AI development and deployment.

AI Sovereignty
A nation’s ability to develop, control, and strategically use artificial intelligence capabilities.

Complementary References

The Economist USA June 20, 2026
Anthropic — research on Claude models and AI safety.
OpenAI — development of advanced generative AI systems.
NVIDIA — AI computing infrastructure and accelerated computing.
Stanford AI Index Report — global trends in AI investment, capability, and adoption.
OECD AI Policy Observatory — international AI governance research.

miércoles, 17 de junio de 2026

Competing in the Age of AI Revisited: When Algorithms Become the Organization

A Review and Critical Reflection on Marco Iansiti and Karim R. Lakhani's Vision of the AI-Native Enterprise

Few business books attempt to redefine the corporation itself. Most offer new management techniques, improved leadership frameworks, or fashionable technologies destined to become obsolete within a few years. Yet Competing in the Age of AI by Marco Iansiti and Karim R. Lakhani is more ambitious. It argues that artificial intelligence is not merely another technological wave; it is reshaping the very architecture of the firm.

Reading the book today feels remarkably similar to revisiting an early map of a continent that the rest of the world has only recently begun to explore. Published in 2020, before ChatGPT, before the generative AI boom, and before the emergence of autonomous AI agents, the book anticipated a reality that is now unfolding across industries: organizations are increasingly becoming software-defined systems in which algorithms, data, and digital networks drive value creation.

The central claim of the book is deceptively simple:

The most successful companies of the twenty-first century will not merely use AI; they will reorganize themselves around it.

This insight remains one of the most important strategic ideas of the decade.

GET YOUR COPY HERE: https://amzn.to/4eiGn2M

The End of the Traditional Firm

For over a century, companies were constrained by three factors:

Scale
Scope
Learning

The larger an organization became, the more expensive coordination became. The broader its product portfolio, the more difficult management became. The faster markets changed, the harder it was for firms to learn and adapt.

Iansiti and Lakhani argue that AI removes these historical constraints. Organizations built around data, algorithms, and digital networks can expand rapidly without proportional increases in labor, infrastructure, or managerial complexity. This is why companies such as Amazon, Alibaba, Google, Microsoft, and Ant Financial achieved unprecedented growth trajectories.

The authors describe these enterprises as AI-centric organizations.

Unlike traditional firms, they do not rely primarily on human-operated processes. Instead, value is created and delivered through software systems capable of continuous learning and adaptation.

This distinction is crucial.

The book is not about implementing AI tools.

It is about redesigning the operating model of the enterprise.

The AI Factory

One of the book's most enduring contributions is the concept of the AI Factory.

Traditional firms operate through departments:

Marketing
Operations
Finance
Sales

AI-native firms operate through continuous feedback loops:

Data → Analytics → Prediction → Action → New Data

This cycle continuously improves itself. Every customer interaction becomes an opportunity to learn.

The more customers the system serves, the smarter it becomes.

The smarter it becomes, the more customers it attracts.

This creates a powerful self-reinforcing mechanism that economists often describe as increasing returns to scale.

In the industrial era, economies of scale eventually plateaued.

In the AI era, learning itself becomes scalable.

This idea helps explain why digital leaders often dominate entire industries with astonishing speed.

Why Strategy Changes in the Age of AI

The book's most provocative argument is that AI changes the foundations of strategy.

For decades, business schools taught firms to compete through:

Cost leadership
Product differentiation
Market positioning

These principles still matter, but they are no longer sufficient.

According to Iansiti and Lakhani, competitive advantage increasingly depends upon three assets:

Data
Algorithms
Networks

As AI systems improve through use, organizations with superior data accumulate advantages that become difficult to replicate.

The result is a new form of competition.

Traditional barriers such as factories, supply chains, and physical assets become less important.

Digital ecosystems become more important.

The winners are often those who occupy the strongest network positions rather than those who possess the largest physical resources.

Strategic Collisions

One of the most fascinating sections of the book explores what the authors call strategic collisions.

These occur when AI-native firms confront traditional firms.

Consider transportation.

Traditional taxi companies operated fleets.

Ride-sharing platforms operated algorithms.

The collision was not between two transportation businesses.

It was between two fundamentally different organizational architectures.

The same pattern appeared in:

Retail
Banking
Hospitality
Media
Advertising

The lesson is profound.

Organizations do not merely compete with products.

They compete with operating models.

And AI-centric operating models often possess structural advantages that analog organizations struggle to match.

Leadership in a World Run by Algorithms

Perhaps the most underappreciated aspect of the book is its discussion of leadership.

Many executives assume AI adoption is primarily a technological challenge.

The authors disagree.

They argue that AI transformation is fundamentally a leadership challenge.

Leaders must rethink:

Organizational design
Governance structures
Decision rights
Talent development
Ethical responsibilities

The transition requires more than purchasing software.

It requires changing how the enterprise thinks.

The CEO becomes less a commander of hierarchical processes and more an architect of adaptive systems.

This insight has become even more relevant in the era of generative AI.

What the Authors Predicted Correctly

Looking back from 2026, the accuracy of many predictions is striking.

The book anticipated:

Software-Centric Enterprises

Today nearly every major company describes itself as a technology company, regardless of industry.

Data as Strategic Infrastructure

Data has become one of the most valuable corporate assets.

Ecosystem Competition

Competition increasingly occurs between interconnected ecosystems rather than isolated firms.

Continuous Learning Systems

Organizations now deploy AI systems capable of improving through ongoing interactions.

Platform Dominance

Network effects continue to reinforce leadership positions among major technology platforms.

In many respects, the book predicted the operating logic behind the generative AI revolution before the revolution arrived.

Where the Book Shows Its Age

No serious review would be complete without acknowledging limitations.

The most obvious limitation is timing.

The book was published before:

Large Language Models
ChatGPT
Claude
Gemini
Agentic AI systems
Autonomous digital workers

As a result, the authors viewed AI primarily as a prediction engine.

Today AI increasingly functions as a reasoning engine, creative partner, and autonomous actor.

This distinction matters.

The next generation of firms may not simply automate decisions.

They may automate entire workflows.

Recent research on agentic AI suggests that organizations are beginning to move from AI-assisted processes toward partially autonomous business systems.

If the original book described AI as the nervous system of the enterprise, the next generation may describe AI as both nervous system and workforce.

The Missing Concept: Human-AI Collaboration

A second limitation concerns people.

The book focuses heavily on organizational architecture and less on the emerging relationship between humans and intelligent systems.

Modern enterprises increasingly rely on hybrid teams composed of:

Human experts
AI copilots
AI agents
Automated workflows

This collaborative model has become one of the defining management challenges of the 2020s.

Future strategy may depend less on replacing humans and more on designing productive partnerships between humans and machines.

Why the Book Matters for Financial Institutions

For banks, insurers, and microfinance institutions, the book remains exceptionally relevant.

Traditional financial institutions often digitized customer interactions while leaving core operating models unchanged.

Iansiti and Lakhani argue that true transformation requires deeper redesign.

For organizations such as MiBanco, the implications are significant:

AI-assisted credit evaluation.
Real-time risk monitoring.
Personalized financial recommendations.
Predictive collections management.
Continuous customer learning systems.

The opportunity is not merely efficiency.

It is the creation of entirely new operating capabilities.

Institutions that successfully build AI-centric architectures can potentially serve more customers, make faster decisions, and learn from market behavior at a scale previously impossible.

The Book's Enduring Legacy

The lasting importance of Competing in the Age of AI lies in its reframing of the strategic question.

Most executives ask:

How can AI improve my business?

Iansiti and Lakhani ask a far more powerful question:

What would my business look like if it were designed around AI from the beginning?

That shift in perspective changes everything.

It transforms AI from a technology initiative into a strategic imperative.

It forces leaders to rethink not just products and services, but the very structure of the enterprise itself.

Conclusion

Competing in the Age of AI deserves recognition as one of the foundational strategy books of the AI era. Its authors correctly identified that artificial intelligence would not simply improve existing firms—it would redefine what a firm is.

The book's greatest achievement is not its discussion of algorithms, data, or platforms. It is its recognition that AI represents an organizational revolution.

Today, as enterprises experiment with generative AI, autonomous agents, and AI-native operating models, the core message remains remarkably relevant: companies will not win because they possess the best technology. They will win because they redesign themselves around new forms of intelligence.

The firms that thrive over the next decade will not merely adopt AI.

They will become AI-native organizations whose strategies, processes, decisions, and cultures are inseparable from the intelligent systems they create.

In that sense, Iansiti and Lakhani were not simply describing a technological transition.

They were describing the emergence of a new species of enterprise.

Glossary

AI-Centric Organization
A company whose core operations are driven by data, algorithms, and digital networks.

AI Factory
A continuous learning system that converts data into predictions, actions, and further learning.

Algorithmic Competition
Competition based on superior analytics, data, and machine intelligence rather than physical assets.

Digital Ecosystem
An interconnected network of organizations, platforms, partners, and users creating value together.

Network Effects
A phenomenon where a product becomes more valuable as more people use it.

Scale
The ability to increase output efficiently.

Scope
The ability to expand across products, services, or industries.

Learning Loop
A cycle in which data continuously improves predictions and decisions.

Agentic AI
AI systems capable of initiating and coordinating actions with limited human supervision.

AI-Native Enterprise
An organization designed from inception around AI-enabled processes and decision-making.

Recommended and Verifiable References

Harvard Business Review – Competing in the Age of AI
Harvard Business School – Competing in the Age of AI Book Overview
Competing in the Age of AI Official Site
Co-Intelligence: Living and Working with AI by Ethan Mollick
Rewired: The McKinsey Playbook on How Leading Companies Win with Technology and AI by Eric Lamarre and colleagues
The Coming Wave by Mustafa Suleyman
Research on AI-enabled firms and dynamic capabilities by David Teece and collaborators.
Emerging research on autonomous business models and agentic AI.

Co-Intelligence: Why the Most Important Employee of the Next Decade May Not Be Human

A Review of Ethan Mollick's Co-Intelligence: Living and Working with AI

Introduction: The Arrival of a New Colleague

Every technological revolution arrives with a promise and a threat.

The steam engine threatened muscle. The computer threatened routine cognition. The internet threatened information scarcity.

Artificial intelligence threatens something deeper: our monopoly on thought.

In Co-Intelligence: Living and Working with AI, Wharton professor Ethan Mollick argues that we are witnessing a transition unlike any previous technological shift. For the first time in human history, ordinary people have access to a system capable of generating ideas, writing reports, coding software, designing products, tutoring students, and participating in creative work previously reserved for humans. Rather than viewing AI as a replacement for human intelligence, Mollick proposes a new framework: co-intelligence—a partnership between human and machine.

We can observe that Mollick is less interested in predicting the future than in teaching readers how to inhabit it. His book is not a manifesto, nor is it a dystopian warning. It is a field guide for navigating an unfamiliar intellectual landscape where machines have become collaborators.

The central question is deceptively simple:

What happens when intelligence becomes abundant?

GET YOUR COPY HERE: https://amzn.to/4xr6kEC

The End of Intellectual Scarcity

For centuries, organizations were built around a fundamental assumption:

Human expertise is scarce.

Companies hired specialists because expertise was expensive to acquire and difficult to scale.

Generative AI changes that assumption.

Mollick argues that systems such as ChatGPT represent the first broadly accessible "general-purpose" technology capable of participating in knowledge work. Unlike earlier software, these systems are not limited to predefined rules. They generate text, synthesize information, brainstorm ideas, and solve novel problems.

This changes the economics of cognition.

Tasks that once required teams of analysts can now begin with a conversation.

Reports that once took days can be drafted in minutes.

Ideas that once emerged from lengthy brainstorming sessions can be generated instantly.

The implications extend far beyond productivity.

They challenge how organizations define expertise, talent, and competitive advantage.

AI as Co-Worker

One of Mollick's most influential concepts is that AI should be viewed as a collaborator rather than merely a tool.

Historically, tools waited for instructions.

Generative AI behaves differently.

It contributes.

It suggests.

It critiques.

It improvises.

This creates a paradox.

AI often behaves enough like a human that users naturally anthropomorphize it, yet it remains fundamentally different from human intelligence. Mollick advises readers to exploit this tension: interact with AI conversationally while maintaining awareness that it is not a person.

In practical terms, AI becomes:

A brainstorming partner.
A research assistant.
A writing collaborator.
A coding companion.
A strategic sounding board.

The most successful professionals will not be those who compete against AI.

They will be those who learn to work alongside it.

The New Productivity Frontier

Perhaps the most important managerial insight in Co-Intelligence is that AI amplifies capability unevenly.

The greatest gains often occur not among experts but among average performers.

Studies discussed by Mollick and subsequent research have shown substantial productivity improvements when knowledge workers integrate generative AI into their workflows. In consulting environments, AI-assisted professionals completed tasks faster and often produced higher-quality outputs than those working alone.

This finding carries profound implications.

Traditional organizations are built around differences in expertise.

AI compresses those differences.

The mediocre writer becomes competent.

The competent analyst becomes excellent.

The excellent analyst becomes dramatically more productive.

This does not eliminate the value of expertise.

Instead, it raises the baseline for everyone.

The competitive battlefield shifts from knowledge acquisition toward judgment, creativity, and execution.

The Four Rules of Co-Intelligence

Throughout the book, Mollick proposes several principles for interacting effectively with AI.

These can be summarized as four managerial rules.

1. Always Invite AI to the Table

Many professionals still treat AI as optional.

Mollick argues the opposite.

Every important task should begin with the question:

"What would AI contribute here?"

This does not mean accepting its output.

It means leveraging its perspective.

Ignoring AI increasingly resembles refusing to use spreadsheets in the 1990s.

2. Keep Humans in the Loop

AI can generate convincing errors.

It can hallucinate facts.

It can produce flawed reasoning wrapped in persuasive language.

Therefore, human oversight remains essential. Mollick repeatedly emphasizes that responsibility cannot be delegated to algorithms.

Organizations that remove humans from critical decisions expose themselves to significant risk.

The future belongs not to automation but to supervision.

3. Treat AI Like a Person—But Remember It Isn't One

This may be the book's most subtle insight.

Conversational interaction often produces better results.

Asking questions.

Providing context.

Requesting revisions.

Offering feedback.

All improve performance.

Yet users must never confuse simulation with consciousness.

The machine is imitating understanding, not experiencing it.

4. Assume AI Will Improve Rapidly

Mollick frequently reminds readers that today's AI is likely the weakest version they will ever use. Technological progress continues at extraordinary speed.

Organizations that wait for perfect systems risk perpetual delay.

Experimentation becomes a strategic necessity.

Education: The First Industry to Be Rewritten

Few areas receive more attention in the book than education.

Mollick, together with collaborator Lilach Mollick, has extensively explored how AI can serve as tutor, coach, mentor, simulator, and learning companion.

This challenges traditional assumptions about teaching.

Historically, education relied on scarcity:

Scarcity of instructors.
Scarcity of feedback.
Scarcity of personalized guidance.

AI reduces all three.

A student can now access individualized tutoring at any hour.

Feedback becomes instantaneous.

Learning pathways become adaptive.

The challenge shifts from information delivery toward cultivating critical thinking and judgment.

In a world where answers are abundant, asking better questions becomes the primary educational skill.

Creativity in the Age of Machines

One of the most controversial themes in Co-Intelligence concerns creativity.

Can machines be creative?

Mollick's answer is nuanced.

AI can generate novel combinations of ideas.

It can produce surprisingly useful creative outputs.

Yet creativity remains a collaborative process between human intention and machine generation.

Research associated with Mollick suggests that prompting strategies can significantly influence the diversity and originality of AI-generated ideas. Human guidance remains essential in shaping outcomes.

This transforms creativity from solitary genius into orchestration.

The future creative professional resembles a conductor directing an increasingly capable ensemble.

Leadership in an AI-Native Organization

Perhaps the most significant implication of the book lies in leadership.

Traditional management focuses on coordinating people.

Future management will coordinate people and intelligent systems.

This requires new competencies.

Leaders must learn:

Prompt design.
AI evaluation.
Human-machine workflow design.
Algorithmic governance.
Digital ethics.

The CEO of the future may spend as much time designing cognitive systems as designing organizational structures.

The question becomes:

How do you manage intelligence that is not human?

Few management books address this challenge as directly as Co-Intelligence.

The Book's Greatest Strength

The greatest strength of Mollick's work is its practicality.

Unlike many AI books that oscillate between utopian enthusiasm and apocalyptic fear, Co-Intelligence occupies a productive middle ground.

It neither worships AI nor dismisses it.

Instead, it encourages experimentation.

This pragmatic optimism explains why the book has become one of the most influential management texts of the generative AI era.

The Book's Main Limitation

Its optimism can occasionally appear understated regarding structural disruption.

Since publication, organizations have begun experimenting with AI agents, autonomous workflows, and large-scale workforce redesign.

Some critics argue that the book devotes insufficient attention to potential labor displacement and broader societal consequences. Academic reviewers have similarly noted that ethical concerns deserve deeper treatment.

Yet this limitation may also be its virtue.

Mollick focuses on what leaders can do today rather than speculative futures.

Why the Book Matters More in 2026 Than It Did in 2024

Ironically, Co-Intelligence has become more relevant with time.

The emergence of agentic AI, autonomous digital workers, and enterprise-wide AI adoption reinforces its central premise.

The most successful organizations are discovering that value emerges not from replacing humans but from redesigning workflows around human-machine collaboration.

Recent discussions by Mollick increasingly emphasize that uniquely human judgment, taste, and decision-making may become more valuable as technical skills become automated.

The future is not human versus machine.

It is human plus machine.

That distinction may define the next decade of management.

Conclusion

Co-Intelligence is not merely a book about artificial intelligence.

It is a book about adaptation.

Ethan Mollick's central insight is that AI should not be viewed as a competitor waiting to replace us. Instead, it should be understood as a new participant in the human enterprise of thinking, learning, creating, and solving problems.

The organizations that thrive in the coming decade will not necessarily possess the most advanced algorithms. They will be the ones that learn how to combine human judgment with machine capability most effectively.

In that sense, Co-Intelligence may ultimately be remembered as one of the first serious management books of the AI-native era—a guide to a future in which intelligence itself becomes a shared resource.

The question is no longer whether AI will transform work.

The question is whether leaders can transform themselves quickly enough to work alongside it.

Glossary

Agentic AI
AI systems capable of performing multi-step tasks with limited human intervention.

Co-Intelligence
A collaborative model in which humans and AI work together to achieve outcomes superior to either acting alone.

Generative AI
Artificial intelligence capable of creating text, images, code, audio, and other content.

Hallucination
An incorrect or fabricated output generated by an AI system.

Human-in-the-Loop
A governance approach requiring human oversight of AI decisions.

Large Language Model (LLM)
A machine-learning model trained on massive text datasets to understand and generate language.

Prompt Engineering
The practice of designing instructions that improve AI outputs.

Knowledge Work
Professional work centered on information processing, analysis, decision-making, and creativity.

AI-Native Organization
An enterprise designed from the ground up around AI-enabled processes and workflows.

Digital Worker
An AI system that performs tasks traditionally executed by human employees.

Complementary References

Books

Competing in the Age of AI — Marco Iansiti & Karim Lakhani.
The Coming Wave — Mustafa Suleyman.
Power and Prediction — Ajay Agrawal, Joshua Gans & Avi Goldfarb.
The Worlds I See — Fei-Fei Li.

Academic and Institutional Sources

Wharton Knowledge: Co-Intelligence: How to Live and Work with AI.
Ethan Mollick & Lilach Mollick, Assigning AI: Seven Approaches for Students.
Ethan Mollick & Lilach Mollick, Instructors as Innovators.
Review in AI and Ethics: The Process Is the Product.
Google Books and publisher descriptions of Co-Intelligence.

Overall Rating (HBR Perspective): 9.5/10

A foundational management book for understanding how organizations, leaders, educators, and professionals can create value in the age of generative AI.

domingo, 14 de junio de 2026

The Moon Problem Nobody Is Solving Honestly

Starship's Orbital Refueling Bet Is Riskier Than SpaceX Admits — And the Alternatives Aren't as Simple as Critics Think

The race back to the Moon has a dirty secret: the single most critical technology required to get there has never been demonstrated at operational scale, faces years of delays, and sits at the center of an architecture that compounds risk with every flight.

That technology is orbital propellant transfer. And if it fails, the entire Artemis program fails with it.

This isn't a fringe concern. NASA's own Office of Inspector General has flagged it repeatedly. Internal SpaceX documents obtained by Politico in late 2025 confirmed what independent analysts had suspected: the Artemis III timeline had slipped again, now targeting 2028 at the earliest. The reason, more than any other, is that transferring over 1,200 metric tons of supercooled liquid methane and liquid oxygen between vehicles in low Earth orbit — across 10 to 14 sequential tanker flights — remains an unsolved problem at mission scale.

The conversation about what to do about this, however, has been frustratingly shallow on both sides. SpaceX defenders wave away the concerns as routine engineering challenges. Critics propose elegant-sounding alternatives that quietly relocate the same problems rather than eliminate them.

It's time for a more honest accounting.

What the Current Architecture Actually Requires

The Starship Human Landing System is a 50-meter vehicle that cannot reach the Moon under its own power after launching from Earth. The physics simply don't allow it at the required mass. The solution SpaceX and NASA have committed to is orbital propellant transfer: launch the HLS empty, then send up a fleet of tanker Starships to fill it in orbit before it departs for the Moon.

The numbers are staggering. Current estimates call for approximately 10 to 14 tanker flights, each carrying hundreds of tons of cryogenic propellant. Each flight requires a successful launch, a precise orbital rendezvous, a hard dock between two vehicles exceeding 1,000 metric tons when loaded — an operation with no precedent in spaceflight history — and a cryogenic fluid transfer that must succeed completely before the propellant begins boiling off.

Boiloff is not a minor issue. Without active zero-boiloff cooling systems, liquid oxygen and liquid methane evaporate at approximately 1% per day. That means the entire campaign of tanker flights must execute in rapid succession, with no margin for weather delays, vehicle anomalies, or the kind of iterative troubleshooting that defines SpaceX's development philosophy on the ground.

As of mid-2026, SpaceX has demonstrated small-scale cryogenic transfer between two Starships at roughly 5 metric tons — a proof of concept, not a mission capability. A full-scale propellant transfer demonstration remains scheduled for later this year. The fully integrated depot-and-tanker architecture needed for a lunar mission is now realistically targeted for 2027 or 2028 by independent analysts, with 2028 being the more defensible estimate given current progress.

This is not an argument that SpaceX will fail. SpaceX has an extraordinary track record of solving problems that looked intractable. But it is an argument that the architecture carries concentrated, sequential risk that is frequently underrepresented in public discussions.

The Alternative That Moves the Problem Around

A growing number of voices in the space policy community have proposed what sounds like a cleaner solution: abandon orbital refueling entirely and replace it with a three-vehicle modular architecture — a dedicated Earth-to-orbit transport, a permanent cislunar tug operating exclusively in space, and a specialized lunar lander optimized solely for lunar operations.

The logic is appealing. Specialization produces simpler vehicles. Simpler vehicles are more reliable. The Apollo Lunar Module, optimized for a single environment, remains one of the most efficient spacecraft ever built.

But the appeal obscures a critical substitution. The cislunar tug still needs propellant. Where does it come from?

If the propellant comes from Earth, you haven't eliminated orbital refueling — you've renamed it and moved it to a different vehicle. The tug must be loaded in orbit just as the HLS must be loaded in orbit. The fundamental problem of transferring cryogenic propellants in the space environment doesn't disappear because the receiving vehicle is called a tug instead of a lander.

If the propellant comes from the Moon — from water ice mined at the lunar poles and electrolyzed into hydrogen and oxygen — then you've assumed a technology that doesn't yet exist at operational scale. The MOXIE experiment on the Mars Perseverance rover produced oxygen at roughly 10 grams per hour in a laboratory demonstration. Scaling that to the hundreds of tons required for routine cislunar operations is a generational engineering project, not a near-term solution.

The three-vehicle architecture also generates a problem it doesn't acknowledge: three docking events instead of one. To reach the lunar surface and return, a crew must rendezvous the Earth-orbit transport with the tug, travel to lunar orbit, transfer to the lander, descend, ascend, rendezvous with the tug again, and finally transfer back to the Earth-return vehicle. Each rendezvous is a failure point. The architecture trades one category of complexity for another without demonstrating a net improvement in mission reliability.

And then there is the permanent cislunar tug itself. A vehicle that never returns to Earth for servicing, operating in a high-radiation environment saturated with micrometeorites, degrading continuously over years or decades — this is not a solved problem. The International Space Station requires constant maintenance by visiting crews and resupply missions. A tug in cislunar space, far from Earth's protection and rescue range, faces a harder maintenance environment with less access.

What NASA Already Tried — and What Happened to It

It is worth noting that the modular approach has already been pursued at institutional scale. The Lunar Gateway — a small space station in a Near-Rectilinear Halo Orbit around the Moon — was originally designed to serve as exactly this kind of cislunar node: a permanent platform for crew transfer, logistics staging, and eventual lunar lander operations.

The Gateway was not killed because the concept was wrong. It was effectively cancelled in March 2026 because it was expensive, slow to develop, politically fragile, and — critically — added complexity to near-term lunar missions rather than reducing it. The Trump administration's proposed FY2026 budget called for cancellation, citing costs and commercial alternatives, though Congressional funding kept it nominally alive through the One Big Beautiful Bill Act. By early 2026, references to the Gateway had quietly disappeared from Congressional funding legislation, and NASA Administrator Jared Isaacman was reportedly considering replacing it with a surface-based Moon program altogether.

The Gateway's trajectory illustrates a core tension in lunar architecture design: the infrastructure that would make long-term operations simpler is expensive and time-consuming to build, while the approaches that are cheaper and faster in the near term concentrate risk into fewer, higher-stakes operations.

A More Honest Framework for Thinking About This

Rather than choosing between "orbital refueling works fine" and "modular architecture solves everything," a realistic assessment has to grapple with a harder set of tradeoffs.

On orbital refueling: The technology is real, the physics are sound, and the engineering challenges are finite. But the timeline is not 2026. It is probably 2027 at the earliest for a demonstration, and 2028 or beyond for mission-qualified capability. NASA and SpaceX should say this clearly rather than continuing to defend dates that independent analysts have already abandoned. The Artemis program's credibility suffers more from repeated slips than it would from a single honest timeline reset.

On modular architectures: The right question isn't whether to use multiple vehicles — that's already the plan, since Orion and Starship HLS together constitute a two-vehicle system. The right question is where to place the complexity. Orbital refueling concentrates complexity in the Earth-orbit phase. A cislunar tug concentrates complexity in the cislunar phase. Neither eliminates complexity; they distribute it differently. The choice should be made on the basis of which distribution is more manageable given actual technological readiness, not on the basis of which narrative sounds more elegant.

On ISRU: Lunar propellant production is probably the most genuinely transformative technology in the long-term vision of sustainable lunar operations — but it belongs in the 2030s roadmap, not in the architecture justification for missions targeting 2027 or 2028. Treating it as a near-term solution obscures the actual state of the technology and misrepresents what is required to make the immediate missions work.

On development economics: This point is almost entirely absent from technical architecture discussions, but it matters enormously. Developing three independent vehicle systems — Earth transport, cislunar tug, lunar lander — requires three development programs, three qualification campaigns, three operational logistics chains. SpaceX's economic model has demonstrated that aggressive vertical integration and high launch cadence can dramatically reduce costs. A fragmented architecture across multiple vehicles and potentially multiple contractors partially sacrifices this advantage. Any honest proposal for a modular system needs to account for total development cost, not just mission elegance.

What a Realistic Path Forward Looks Like

No single architecture solves the lunar transportation problem cleanly. But a realistic near-term path would look something like this:

The immediate priority must be completing the orbital propellant transfer demonstration at scale — not in the incremental, prototyping mode SpaceX favors, but in a structured test campaign that actually validates mission-level propellant quantities and transfer rates. Until that demonstration succeeds and produces verifiable data, all Artemis timelines are speculative.

The longer-term architecture question should be decoupled from the near-term mission question. Whether the right long-term infrastructure is a reconfigured Gateway, a permanent cislunar tug, a surface-based logistics hub enabled by ISRU, or some combination of these is a legitimate policy debate — but it's a debate about 2032 and beyond, not about Artemis III.

What is not helpful is pretending that the current architecture's challenges are routine when they are not, or pretending that alternative architectures eliminate those challenges when they mostly relocate them.

The Moon is hard. The engineering is hard. The honest answer is that humanity is attempting something that has never been done at scale under cost and schedule pressure that has no precedent. The least we can do is be clear-eyed about what we're actually solving.

The author has no financial interest in any space company or program referenced in this article.