Ego Development and AI

It was a recent article by Svenja Hofert that got me thinking: as we now move into the age of AI, how does this relate to the ego development stages defined by Jane Loevinger.

I did some research and found that there is no evidence on this yet. So the following remarks are not established findings, but purely reflections that still need to be discussed and evaluated.

Loevinger distinguishes 9 stages of ego development: E1 to E9. From E4 onward, the stages relate to adulthood, with E4 to E6 referred to as the conventional range and E7 to E9 as post-conventional. Thomas Binder popularized the model in Germany, and Svenja Hofert recently added a neurobiological perspective.

Let's look at the characteristics of the stages and how they potentially interact with the introduction of AI:

E4 – Conformist: The core theme here is belonging to a group. Behavior is subordinated to this. Thinking in terms of "we" and "they" — in-group and out-group — is also pronounced. Belonging, clear structures, and authority provide security.

Belonging to a group is probably completely irrelevant to AI systems. They act independently of it. Moreover, large language models have the ability to reflect a variety of viewpoints. This can be unsettling for people at this stage, or the statements of an AI are relativized and rejected. It then simply belongs to the out-group.

However, since large language models are trained to affirm users rather than unsettle them (people pleasers), there is a chance that linguistically careful broadening of perspectives could open up additional development opportunities for people at this stage, more easily than through a human conversation partner, since LLMs tend not to be perceived as a threat.

E5 – Self-Aware: In this phase, autonomy and competence take center stage. Belonging to a group recedes somewhat into the background.

Since experiencing competence contributes to identity and differentiation plays a role, this phase brings both a vulnerability through AI and an opportunity to strengthen one's own position. For people who have learned to use AI tools to expand and showcase their own competence, these tools are a great aid. On the other hand, they are also a danger, because AI's competence profile keeps becoming more unbeatable. If people enter into competition with AI, especially in work contexts, this can lead to a deep devaluation of their personal role, since humans can only lose against AI.

E6 – Conscientious: The experience of competence here expands to include the perspective of efficiency. It's about achieving goals and being successful.

AI has a double-edged effect here. On the one hand, artificial intelligence is currently the technology that supports efficiency most strongly. On the other hand, at this stage too there is a threat that humans and machines will end up in competition with one another. And in the world of work, this contrast is currently becoming apparent. It can turn into a race that leads to overload and exhaustion. One way out can be a more nuanced view of AI, reflecting on human roles, values, and the meaning of work and life. Remaining stuck in efficiency logic, on the other hand, becomes a trap.

E7 – Individualistic: Thinking becomes more flexible and multifaceted, and human qualities such as understanding and empathy for others are more prominent. People become more tolerant of ambiguity, better able to endure multiple meanings.

At this stage, something that still seems difficult at E6 should get easier: perceiving human qualities and AI's capabilities as complementary rather than competing with it. By looking at oneself, at others, and at AI, it becomes possible to shape work environments that are AI-supported yet still humane. In the best case, artificial intelligence loses its threatening character, since individualistic people are able to recognize its limits and fill the space AI cannot reach because it is purely human: creating meaning, connectedness, understanding of the world, judgment, empathy.

E8 – Autonomous: The big picture as well as connections and interdependencies are perceived. Thinking is original, multi-layered, and strategic, but not cold and purely rational, since there is also an understanding of other people's differences.

Most people at this stage probably perceive AI as one element within a complex world: capable, but not all-powerful. Strategic thinking makes it possible to integrate even highly advanced systems into overall concepts without losing human control. On the contrary, human qualities are perceived and cultivated as essential factors for a successful life.

E9 – Integrated: Loevinger herself described this final stage as barely attainable. It is reserved for the few people who manage to come to terms with themselves to such an extent that they can let go of all the stresses of life. Instead, there is serenity and a need to understand the world ever more deeply.

As at stage E8, the perception of AI is probably rather relaxed and interested. The need to make AI useful for oneself or an organization is likely to be quite weak, except perhaps for a curiosity to see what further insights can be gained from it.  

Conversely, from AI's point of view, it probably has to admit that it is completely left out of the picture. It would take different architectures and a highly advanced machine consciousness to be able to enter an E9 world.

Loevinger's system and its further developments make clear, in a way not previously discussed, that there are areas of human life and consciousness that AI cannot reach, at least not in the foreseeable future. This can take some of the drama out of the AI discussion.

It also becomes clear that actively shaping a world with AI is essential and possible in order to keep it humane. Only a few people develop to stage E7 or higher. The majority of humanity moves within the conventional range. And here there is indeed a risk that central components of the ego could be jeopardized by AI.

All these reflections on how people, viewed through the lens of the ego development model, can deal with AI are speculative. No one knows whether people at E8 actually have fewer concerns about AI, or whether they are simply better able to handle it. So far, there is no research data available.

Nevertheless, I would like to spark the discussion. Until now, there has only been a comparison between human and artificial intelligence. Loevinger's model points beyond that, by bringing genuinely human qualities to the surface.

Only if we know and value them, we will succeed in shaping a world with AI in which people can do what only people can do.