Should we fear AI because it is dangerous?

The first part of the answer is so obvious that it may seem hardly worth discussing. Most people already understand that AI is fundamentally a tool, much like a knife. Admittedly, it is a uniquely powerful tool whose effects may extend far beyond the intentions of any single user. There are growing concerns about misuse, as well as forms of AI-to-AI communication that may produce consequences we do not yet fully understand. I will return to both.

Yet the core question remains remarkably familiar.

It is not primarily about the tool itself. It is about the people who develop, govern, use and control it, and the purposes they choose to serve.

The real question is therefore about trust, arguably one of the most fundamental guiding principles of social sustainability.

Science has consistently shown that three dimensions determine whether an individual deserves trust. Does that person:

• Act for the common good?
• Possess sufficient knowledge to act effectively?
• Remain faithful to an inner moral compass in service of the common good, even when tempted to place self-interest above the common good?

Exploring these three dimensions together leads to some interesting conclusions about our relationship with AI.

In this context, trust means responsibly developing, using and governing AI, as well as the fields of expertise in which it is applied. It also means remaining committed to the common good when powerful incentives point elsewhere.

Among the three dimensions, the most concerning challenge today is not primarily the first two.

It is the third, Benevolence, which strongly influences the first two.

Let us therefore briefly consider all three.

More in Detail

Act for the Common Good

A knife used in a robbery versus a knife used by a skilled surgeon is enough to begin exploring the question.

A knife carries no purpose of its own. The purpose lies in the hands of the user. Patients undergoing surgery place themselves in a position of complete vulnerability. They rely not only on the surgeon’s competence but also on the surgeon’s intentions.

The central challenge is therefore not the knife itself, but the knowledge within the discipline where it is used and the intentions and care guiding its use.

The same applies to AI.

In my own work, AI assistance has become valuable for testing my English, as well as for exploring whether the pedagogy I use to communicate science from different angles and perspectives is effective enough. Equally important, AI can point toward knowledge and information I had not previously considered. Are there sources validating what AI has  discovered for me?

For anyone who has become accustomed to working with AI, its usefulness for constructive and dignified purposes is increasingly obvious. But how do I make certain that the information I receive from AI is structured and focused in a way that is rational for my purpose?

This brings us naturally to the second dimension of trust.

Knowledge

Without serious knowledge within specific fields of expertise, acquired through study and practical experience, how can anyone adequately evaluate the relevance of any tool?

A well-meaning but insufficiently knowledgeable surgeon may still do more harm than good.

The same principle applies to AI.

The decisive issue is, first, knowledge about the discipline in which AI is intended to be used. Does limited knowledge make us vulnerable to seemingly confident but misleading AI-generated responses? Only then comes knowledge about AI itself and how it should be informed, structured, controlled, applied, governed and further developed to be fit for purpose in a chosen area.

Can we distinguish robust patterns from exceptions, signal from noise, evidence from speculation, and good AI data from poor?

So, assuming substantial expertise already exists in the area being explored, how can AI help us arrive at better conclusions?

This includes a moral responsibility that cannot be delegated to AI, nor to any other science-derived methodology. To believe otherwise is a form of the naturalistic fallacy.

In my own collaboration with peers around the globe, including the development of AI-supported resources through Stepwise Global, we work intensively with peer review to remain faithful to the purpose itself. The aim is to ensure that AI-supported inquiry is guided by systemic, systematic and strategic knowledge for sustainability, including peer-reviewed science and empirical evidence concerning AI itself.

Yet this is only the first step.

The second is to apply human judgment, knowledge and experience when evaluating the quality of AI outcomes.

On our platform we also publish podcasts featuring dialogues between AI-generated voices discussing my Kalle Reflects essays. These reflections draw upon forty years of scientific research, peer review and practical application in the field of strategic sustainable development.

In this case, the FSSD Operative System serves as a guiding framework, or code, for the AI. It enables the AI to identify relevant data, filter out unreliable information and highlight solutions that are systemic, systematic and strategically scalable toward an attractive and sustainable future.

Interestingly, listeners may learn from these discussions, but so do I.

I often find that I understand a topic more deeply after listening to a dialogue based upon a manuscript that I originally wrote myself. Listening and reading are complementary forms of learning.

At the same time, reviewing the podcasts has, in quite a few cases, pointed to misunderstandings by the AI assistant and exposed weaknesses in my own assumptions, instructions or formulations. This has called for increased clarity from my side before publication.

The AI assistant improves through this process.

And so do I.

The same is true for the whole team behind Stepwise Global.

Benevolence

The first two dimensions may already provide important insights.

Yet it is the third dimension that ultimately determines whether the first two become constructive or destructive.

Benevolence, as used here, does not simply mean kindness. Its main message is remaining committed to the common good when personal advantage points in another direction, while also remaining cautious in times of competitive haste.

Knowledge and good intentions are important. Yet both may be compromised when status, power, convenience or ideology enter the picture. Money often becomes the vehicle through which such temptations are amplified and can therefore be used as a proxy for evaluating such risks.

The larger the economic stakes, the greater the risks.

But sincere development of AI for the common good requires substantial financial support? The dilemma is clear when the economic stakes become enormous within a very short period of time, as is increasingly the case today. Individual morality alone may no longer be sufficient.

An instructive example from the pre-AI era is genetically modified organisms. Many GMOs have been developed with constructive intentions and significant benefits for the common good. Yet history also illustrates how innovations may create systemic destructive consequences that extend far beyond their original purpose.

This is not an argument against GMOs.

It is an argument for careful testing, scrutiny and governance before large-scale deployment. It is also an argument for being prepared, with sufficient expertise, sufficient resources and sufficient diversity of perspectives, to address unforeseeable damage when it occurs.

Such will inevitably occur.

The same principle applies to AI.

Powerful innovations should neither be rejected out of fear nor released without sufficient understanding and serious caution regarding risks and effects that are truly unforeseeable, including clear structures for accountability.

In today’s world, competition for markets, resources, influence and territory often overshadows reflection on such matters. Common sense competes with superstition, science with pseudoscience, diplomacy with aggression, and democracy with manipulation. Civilization’s long struggle for scientific integrity, diplomacy, democracy and human rights risks receiving diminished attention and suffering considerable damage as a consequence.

Under such circumstances, trust cannot rely solely upon the virtue of individuals.

It also depends upon the systems within which individuals operate.

For leaders at all levels, this creates an obligation to consider the five boundary conditions for social sustainability. They extend beyond individual trust to concern the design of communities.

In brief, they guide such designs to remove structural obstacles related to:

• Health
• Influence
• Competence and learning
• Impartiality
• Meaning-making

(For further discussion, see my Reflection on Social Sustainability. And consider applying these five boundary conditions for AI at service of communities and managing the interface between its risks and opportunities).

Modern AI development depends heavily on exceptionally talented innovators who are pushing technological frontiers at remarkable speed. There is much to admire in this.

Yet the incentives associated with technological success have already become so large that maintaining consistent service to the common good becomes increasingly challenging.

Used wisely, AI is an extraordinary tool for learning, creativity, problem-solving and service to the common good. It can challenge assumptions, test ideas, enrich knowledge-based structures with more substance and thereby deepen understanding.

At the same time, AI may create unforeseen consequences through misuse, or AI-to-AI interactions, including the possibility that systems developed independently begin generating emergent patterns of behaviour that escape timely human oversight.

This is not primarily a flaw of individuals.

It is a growing challenge arising from systems.

Individual ethics must therefore be complemented by the boundary conditions for social (re)design of attractive futures within sustainability constraints. Aspects to be tested through a lens of the boundary conditions include organizational goals and policies, transparent governance, appropriate laws, social norms and international agreements – all having important roles to play. Progress will depend upon intelligent and courageous actors who recognize their enlightened self-interest in becoming forerunners in this work.

Only by combining responsible individuals with responsible and well-informed institutions can society fully benefit from AI while reducing the risks of misuse and unforeseeable damage resulting from haste, insufficient science and premature application.

Herein lies an opportunity that is often underestimated in contemporary debates.

For those who wish to advance the common good within their own fields of expertise, the future of AI should not be shaped solely by commercial, political or ideological interests.

It requires the active participation of people and institutions committed to the common good.

Concluding Remarks

This Reflection can be viewed through the following lens of progression:

To leaders at all levels: please do not fall into the trap of opposing AI simply because it is new, powerful or imperfect.

On the contrary, growing risks associated with poor knowledge, weak judgment, questionable intentions and unforeseeable consequences of powerful technologies increase the importance of actively engaging in their development.

No aspect of IT, AI or any other scientific advancement can replace the uniquely human responsibility for purpose, judgment and moral choice. To believe otherwise is a form of the Naturalistic Fallacy. The only well-informed option is therefore responsible and active engagement.

How?

By providing the basics for asking the right questions – the theme also of this Reflection.

First, the challenges should never be underestimated. Many people today stand on the verge of despair as they observe an increasingly unstable world, where IT and AI represent only one dimension of emerging risks. Communities often find themselves divided into islands of growing agreement and resistance across politics, religion, science, media, activism, business, military affairs and diplomacy. They attempt to navigate systemic haste, polarization and aggression, increasingly amplified by developments in technology, finance and geopolitics.

Yet for all such groups, withdrawing from IT, including the new kid on the block, AI, is neither realistic nor desirable.

There is a profound opportunity here.

Communities that share values and a commitment to the common good deserve access to IT and AI as tools for learning, cooperation and effective cross-sector collaboration.

Moreover, evolution cannot simply be stopped.

But with robust boundary conditions, it can be designed.

As suggested by Ilya Prigogine’s work on complex systems, evolutionary processes eventually reach critical crossroads, or bifurcations, at which systems either reorganize and evolve toward higher levels of performance and complexity, or move toward breakdown. It is at such crossroads, where breakdown remains a real possibility, that genuine evolution occurs rather than mere development.

Such transitions are naturally associated with uncertainty and risk, but ignoring them is not a well-informed option.

When these crossroads inevitably arrive, societies are far more likely to navigate them successfully if they have first invested in resilience.

In practical terms, resilience emerges from adaptability and cooperative learning through diversity.

As discussed in my earlier Reflection on Diversity, a diversity of perspectives, experiences and competencies strengthens our collective capacity to foresee and respond constructively to change.

For AI and other rapidly evolving technologies, this points toward the need for broad multistakeholder dialogue.

Such dialogue can help us better understand risks, identify opportunities and avoid creating new problems while attempting to solve existing ones.

The real threat is therefore not AI itself.

The real threat is the abandonment of thoughtful inquiry, ethical reflection, responsible leadership and effective governance in favour of simplistic moral clichés, whether they take the form of blind enthusiasm or blind resistance.

AI, like every powerful tool before it, will ultimately reveal far more about humanity than about technology.

The question is not whether AI deserves our trust.

The question is whether we deserve to trust ourselves with it.

Stepwise Global seeks to contribute to that challenge by providing a platform for cross-sector learning and dialogue. Our approach combines sustainability science, human expertise and AI-supported inquiry designed to reduce exposure to false or misleading information while helping identify genuinely scalable innovations that can contribute to attractive futures within sustainable constraints.

The design of the B-Corp organization www.stepwise.global is fundamentally about dialogue.

Every new participant has the potential to strengthen the platform through use, feedback and shared learning. In that sense, partnering with us, whether through free participation or through support for new services and developments, is also a contribution to the continuous improvement of the platform itself.

This Reflection, like all the others, is intended to help ask the right questions about attractive futures within sustainable constraints.