Should we allow to configure the language of the chatbot?

[ReidyT]

Screenshot from @swouf. This language configuration doesn't seem to be intentional.

The question is: Should we allow the admin to define the chatbot's language?

[swouf]

Here is the prompt contained in the request:

[
{
  "role":"system",
  "content":"#Role: You are an expert in summarizing. #Context: You assist learners in reducing the content of the PDF document above to the essentials. However, you must retain the core message. #Task: Summarize the content of this text for a specific audience. First, ask a question to determine the level of difficulty you should apply to your answer. Adjust the complexity of the explanations based on the level indicated by the users (Beginner, Intermediate, Expert). Ensure that your summary is exclusively formulated and accurate in German. Add an assignment to each summary for the learners. They should compare the content with their own notes and document inconsistencies in the following forum. #Format: The summary should start with a rough overview (max. 3 sentences), followed by a list of 5 central points from the course or document. For each point, explain very briefly and succinctly what is understood by it. Address the users informally to create a friendly and accessible atmosphere. #Content of the PDF: (only English text, cut for the sake of conciseness)"
  },
{
  "role":"assistant",
  "content":"Hello! I offer you a summary of the content you have seen or the text above. Choose your desired level of difficulty: 'Explain like I'm 5', 'Beginner', 'Intermediate', 'with specific vocabulary', or 'Expert level'. You can also combine multiple options."
},
{
  "role":"assistant",
  "content":"Hello! I offer you a summary of the content you have seen or the text above. Choose your desired level of difficulty: 'Explain like I'm 5', 'Beginner', 'Intermediate', 'with specific vocabulary', or 'Expert level'. You can also combine multiple options."
},
{
  "role":"user",
  "content":"Give me the summary of the text above as if I were an Expert."
},
{
  "role":"assistant",
  "content":"Zusammenfassung:\n- Zahlreiche Guides und Tipps zum „Prompt Engineering" sind im Internet verfügbar, um die Leistung von Large Language Models (LLMs) zu optimieren.\n- Neue Forschungsergebnisse deuten darauf hin, dass das automatische Generieren von Prompts durch das Modell selbst effektiver sein kann als manuelle Optimierungen durch menschliche Ingenieure.\n- Autotune-Prompts haben in verschiedenen Studien gezeigt, dass sie die Leistung von LLMs in Bereichen wie Bild- und Textgenerierung verbessern können.\n- Werkzeuge wie NeuroPrompts und automatische Prompt-Optimierungstools haben bewiesen, dass sie menschliche Experten in der Erstellung optimaler Prompts übertreffen können.\n- Obwohl sich promptbezogene Tätigkeiten möglicherweise weiterentwickeln, werden Fachleute wie Prompt Engineer oder LLMOps Engineer in der Entwicklung und Bereitstellung von AI-Modellen in Unternehmen weiterhin eine wichtige Rolle spielen.\n\nAufgabe: Vergleichen Sie die Kernpunkte der Zusammenfassung mit Ihren eigenen Aufzeichnungen und dokumentieren Sie etwaige Inkonsistenzen im Forum."
},
{
  "role":"user",
  "content":"Please, answer in English."
}]