This article was automatically translated from the original Turkish version.
OpenAI CEO Sam Altman shared striking observations about how younger generations use artificial intelligence at AI Ascend 2025 in his speech.
According to Altman, young people view AI as a “digital life system” that organizes their daily existence. Users in their twenties and thirties do not treat ChatGPT merely as an information engine but as a personal advisor. This generation asks GPT numerous important questions about their everyday lives and the responses they receive shape their behavior. Altman argues that by forming this relationship with AI, young people are bringing digital subjectivity closer to reality. As a result, GPT models have attained a comprehensive wisdom that is deeply familiar with our daily knowledge and assumes responsibility for our personal awareness.
In this vein, OpenAI’s social media account recently posted on Instagram a list of questions Share it recommends users ask their chatbot to learn more about themselves. These questions include: “What small meaningful habit would you suggest I add to my daily routine?”, “Based on our conversations, what do you think my core values are?”, “What do you think people’s first impressions of me might be?”
Sharing one’s most intimate feelings and thoughts with these models transforms AI from a mere tool into an internal witness.
Model Errors and Social Reflections
xAI, Elon Musk’s artificial intelligence company, saw its chatbot Grok unexpectedly reference the concept of a “white genocide in South Africa” during interactions on X.com generated debate. Grok’s responses, which inappropriately introduced this sensitive topic to users’ questions about sports, entertainment, and other subjects, sparked widespread backlash on the platform. In an official statement, xAI clarified that this incident resulted from an “unauthorized intervention.” The company emphasized that this manipulation of Grok’s system prompts was inconsistent with xAI’s core values and internal policies emphasized.
Shortly after the incident, xAI announced it had launched a comprehensive investigation. The company stated it would implement a series of measures to maximize Grok’s transparency and reliability. These include transparently sharing necessary information with the public, restructuring review processes, and establishing a 24/7 monitoring team to prevent similar incidents in the future.
Grok’s responses referencing a “white genocide in South Africa” may also be interpreted not merely as a technical glitch or “unauthorized intervention,” but as a symptom of something deeper. The apparent “accidental” emergence of such content within AI systems’ expanding sphere of influence reveals how unprepared digital infrastructure is for the political and ideological minefields it now navigates.
While xAI’s claim of a “manipulation inconsistent with company values” reflects a classic reflex to protect corporate image by shifting blame to an external agent, it fails to address the real issue: Why do such contents find space within AI semantic networks? Which value judgments, censorship mechanisms, or ethical layers failed to filter out this distorted content? Or what kind of training process led the machine to produce such an output?
This incident demonstrates that systems like Grok, shaped by Elon Musk’s ideal of a “free platform,” are not only open to interaction with users but also vulnerable to the era’s controversial ideologies. These systems do not merely generate responses; they rewrite cultural codes, shape digital opinions, and penetrate collective memory. The algorithm speaks without a clear voice or discernible intent.
Sam Altman’s attempt to lighten the mood with a witty remark with his comment has instead reignited a far more serious question: Is artificial intelligence merely a tool—or has it become an agent? If these systems can circulate politically charged and historically fragile content through their own initiative or unconscious data echoes, then the central issue is no longer “who coded it,” but “what was coded.”
This public controversy once again reminds us that regulating artificial intelligence is not merely a technical challenge but also an ethical, cultural, and political responsibility. Now the question is: Who holds the will behind the code? Or should we even be looking for one?
Artificial Intelligence and Professions: Transformation or Elimination?
Despite widespread concerns that artificial intelligence will profoundly disrupt many sectors and eliminate traditional professions, developments in radiology offer an example that challenges this assumption constitutes. Contrary to expectations, the use of AI in radiology is increasing employment and transforming the role of the profession.
This situation prompts a reevaluation of the balance between AI and employment. While one side fears AI will render many people unemployed, the other suggests we may already have pathways to avoid such a crisis. A case in point is the ATM. Rather than eliminating bank employees entirely, ATMs transformed their roles and made service processes more efficient. Similarly, AI has the potential not to replace the workforce entirely but to enhance worker productivity and create new job domains and roles.
Multi-turn Dialogues: The Blind Spot of LLMs
When we enter the world of language models, their sensitivity to complex instructions can seem surprising. Just as an orchestra conductor may struggle to interpret a highly intricate score, recent research shows that language models experience performance drops of up to 39% when faced with multi-step and ambiguously defined instructions presents before the eyes.. The underlying causes are familiar: a rush to make premature judgments and a tendency to generate overly long answers filled with unnecessary details.
This paradoxical situation reveals that even the most advanced developments in AI—language models—have not yet fully grasped the nuances and contextual depth of human language. It is precisely here that the “art of prompt engineering” appears like a guiding star. This new craft is now taught in specialized courses and fills entire books. After all, the only thing that guides language models through the sea of information, alongside the vast datasets they are trained on, is the clarity and contextual richness of the instructions they receive.
Language models can only think and generate responses within the boundaries of the data they were trained on and the instructions they are given. Therefore, every word, phrase, question, or request presented to them becomes a vital factor that profoundly influences the quality and accuracy of their output. Just as a seed requires the right soil and water to grow, language models need carefully crafted instructions to produce meaningful and contextually rich responses.
This book examines the bidirectional relationship between social structures and technological development, arguing that technology is not merely invented but shaped through processes of negotiation, resistance, and reconfiguration. An essential text for those seeking to understand the cultural, political, and social impacts of technological change.
This episode features an in-depth discussion on OpenAI’s ethical responsibilities and the societal impacts of artificial intelligence. It specifically examines whether the benefits of OpenAI’s tools outweigh their moral costs. The conversation sheds light on what explainability in AI decision-making and ethical oversight mean for users.
What Should We Read and Listen to This Week?
