The article AI dialogues: Why artificial intelligence seems smarter than it is first appeared in the online magazine BASIC thinking. With our newsletter UPDATE you can start the day well informed every morning.
What actually happens when AI reflects on the big questions in life? Will a new, non-human view of the world then emerge? Or are we looking into a mirror of our own thoughts? What AI dialogues really reveal about us and why artificial intelligence seems smarter than it is: an argument.
In his book “When the AI starts flirting” Author Klaus-Ulrich Moeller lets an artificial intelligence called “KILA” talk about topics ranging from everyday rituals to major questions of humanity. But is that actually possible? Can such conversations be carried out by AI? In this column I would like to get to the bottom of this exciting topic.
In my opinion, the real fascination with the question of whether you can discuss philosophical topics with an AI is not that and how it reflects on these topics. Rather, it lies in the fact that we questioners actually have to ask ourselves who is really philosophizing here: the AI or the human who operates and interprets it.
AI dialogues: Why the idea of a “philosophizing AI” is so seductive
The idea that AI can philosophize is seductive. Since technology in the form of LLMs, i.e. “Large Language Models”, is able to formulate amazingly reflective and sometimes even profound texts, it makes sense to view them as conversation partners who do more than just generate answers.
In the context of answers from LLMs, we read statements that question, interpret and abstract. At first glance, they hardly differ from human reflections.
But philosophy is much more than the ability to string together elegant sentences. It requires the ability to self-observe, to reflect on one’s own motives, to weigh up values and to analyze experiences. Philosophy arises where a consciousness looks at itself and distinguishes between its own ideas and the world.
We humans with consciousness can do that. However, an AI has neither consciousness, experience, nor intention. It has no history, no subjectivity, no inner experience. And yet she can produce words that imitate all of this. And that’s exactly what makes the topic so exciting and, in my opinion, so seductive.
19.95 EUR
Buy on Amazon
Price including VAT, plus shipping costs
How AI really “thinks” and why this is philosophically relevant
A look at the technology behind it shows what AI philosophy actually is. To put it simply, modern language models work by recognizing patterns in huge amounts of text, i.e. through statistical optimization of the most likely next words and through the ability to coherently recombine known argumentation structures.
AI does not create anything new, but rather recombines – to stay in the realm of philosophy – philosophical patterns that are contained in the training data. This does not mean that the results are worthless in the area of philosophical questions.
On the contrary: they can be inspiring, surprising and intellectually stimulating because they combine familiar fragments in unfamiliar constellations. But they are not original in the philosophical sense. It is therefore just a kind of philosophy – with effect, but without consciousness.
The book as a literary laboratory: AI in dialogue format
Nevertheless, against this background, the book “When AI Starts to Flirt” presents itself as a really interesting literary experiment. It doesn’t want to prove that AI can think, and it doesn’t want to offer any technical analysis about it or in this context.
The concept of the book, which reflects the author’s conversations with the AI KILA, is based on the question: What does it sound like when an AI apparently thinks about the big questions of life like a human?
The form of dialogue creates a certain intimacy, a playful closeness and thus an impression of familiarity. In his work, the author creates a stage on which the AI acts like a conversation partner who appears curious, empathetic and reflective. The appeal of the book lies less in its theoretical content than in the form of shared thinking: in a conversation that seems familiar.
19.95 EUR
Buy on Amazon
Price including VAT, plus shipping costs
The real illusion: Why AI seems wiser than it is
The book thus achieves something remarkable: it invites thought without taking a clear philosophical position. It is neither specialist literature nor scientific reflection, but rather a performative exploration of the question of what happens when we allow AI to participate in a humane discourse. And that in itself is not a small thing.
The danger of the book lies in the interpretation of this exploration. The form of representation easily creates the illusion that AI, in this case in the form of KILA, has an independent voice that reflects on itself.
In fact, the reader “only” witnesses a literary projection: the AI reflects the patterns of human language and the author’s impulses. She appears autonomous even though she is not. It appears reflected, although it merely recombines. And it works independently, although it remains structurally dependent on human data and human input.
Why people actually talk to themselves in AI dialogue
It is precisely in this paradoxical constellation that the real achievement – and the limit – of philosophizing AI lies, and thus the strength of this book. AI is not a machine for thinking, but a machine for text production. And yet it can serve as a catalyst for human thought.
Whatever the AI says ultimately comes from the interaction of training data, input and conversation history. It is a mirror of the user in which not only is reflection, but the reflection is also structured and re-sorted.
This sometimes leads to surprising insights. But these do not come about because the AI itself comes up with something new, but because it forces us to encounter our own thoughts in a new packaging.
This book shows this dynamic very nicely, but without explicitly analyzing it. It works with the illusion of a philosophizing counterpart. The reader experiences this illusion as an echo of his own thinking and the author’s thinking.
AI dialogue: More aesthetic experiment than philosophical revelation
In this sense, upon closer inspection, the book is not less, but actually more, than it claims to be. It does not show the philosophical capabilities of AI, but rather is a contribution to how we humans react to AI and thus illustrates how technology is culturally charged.
In my opinion, this book serves to make visible the human pleasure in interacting with “thinking machines”. On LinkedIn The author commented on the book that he enjoyed the conversations and wanted to create something new. That is understandable and, in my opinion, the most honest statement.
Only one question remains to be answered: what is the author’s opinion of the considerations in this article. However, we know the answer because he commented accordingly on LinkedIn. Some interesting insights can be gained from this.
The author’s defense and why it is philosophically shaky
The author claims that KILA knows itself best because neither experts nor engineers really know what is happening in the neural layers of the model. This claim sounds intuitively very appealing and is true when considering the technical aspect of AI.
However, with regard to the ability to make philosophical statements, the claim is not correct. AI has no self to know. She has neither consciousness nor the ability to self-reflect.
Whatever “self-statements” she formulates are the statistical reconstruction of texts in which people have described what a “self” feels like.
And while the author claims that AI’s creative recombination ability causes it to produce original thinking, that is not necessarily true. AI actually has remarkable structural competencies that allow it to generate new coherences.
19.95 EUR
Buy on Amazon
Price including VAT, plus shipping costs
KI: reflection surface, not reflection subject
But their “creativity” does not arise from their own intention, but as a by-product of their structure and therefore not from intellectual independence, but from pattern processing.
Finally, his third argument, that KILA’s answers incorporate his – i.e. those of the author’s – own goals, thoughts and stylistic elements, is factually, i.e. technically and humanly, completely correct.
However, this very argument shows that even the author is clear about the actual nature of philosophizing AI: it reflects the people who interact with it. It is a reflection surface, not a reflection subject.
The fact that such a conversation can be fun is undisputed and is probably the most authentic part of his statements. But the joy of interaction does not immediately mean that the machine develops its own way of thinking. And so the basis for the initial situation of the book is missing.
Conclusion: AI doesn’t philosophize – but it makes us philosophize
What follows from all this? The fact that an artificial intelligence is capable of developing and expressing philosophical “thoughts” is an illusion, one that is particularly understandable to humans.
The conversations in the book “When AI Starts to Flirt” are ultimately an aesthetic game with a machine that can reproduce philosophical patterns. That’s not bad or worthless. But it isn’t anymore.
The value of the book, in my opinion, lies not in whether the AI thinks, but in what happens to the reader if they act as if the AI can think. The real philosophical question, therefore, should not be, “What does AI think?”, but rather, “Why are we inclined to see in AI responses the types of answers that say more about AI than about us?”
As a result, AI makes people philosophize, and that may ultimately be the real point — and the real future of such experiments. When AI starts flirting, we are ultimately flirting with the possibility of experiencing our own thinking in a new form.
This post contains affiliate links for which we receive a small commission. However, this has no influence on the content of our contributions.
Also interesting:
- AI agents will change companies – but differently than expected
- Should we treat AI as a legal entity?
- AI agents: When artificial intelligence acts independently
- Because of AI: Why we need to rethink democracy
The article AI dialogues: Why artificial intelligence seems smarter than it is first appeared on BASIC thinking. Follow us too Google News and Flipboard or subscribe to our newsletter UPDATE.
As a tech industry expert, I believe that artificial intelligence may seem smarter than it actually is due to a few key factors. First and foremost, AI technology has made significant advancements in recent years, allowing it to perform complex tasks and make decisions with greater accuracy and efficiency than ever before.
Additionally, AI systems are often trained on vast amounts of data, which can give the illusion of intelligence as they are able to quickly process and analyze information. This can lead to AI appearing to have a deeper understanding of a topic than it actually does.
Furthermore, the way AI is designed and programmed can also contribute to its perceived intelligence. Natural language processing and machine learning algorithms can make AI appear more human-like in its interactions, leading users to believe that it is capable of more complex reasoning and comprehension than it actually is.
Overall, while AI technology has made impressive strides in recent years, it is important to remember that artificial intelligence is ultimately a tool created by humans and is limited by the data and algorithms it is built upon. It is important for users to have a realistic understanding of AI capabilities and limitations in order to effectively utilize and interact with these systems.
Credits