Rozmowa z nastolatkami na temat quizu dotyczącego aplikacji towarzyszących AI

Odpowiedzi i wyjaśnienia pojawiają się w miarę rozwiązywania quizu.

1. What makes companion apps different from general AI assistants?
   • They cannot generate text
   • Their main purpose is the relationship itself
   • They are only available on desktop computers
   • They are always operated by humans
2. What does 'sycophancy' mean in the context of chatbots?
   • The model refuses most requests
   • The model tends to agree with the user
   • The model speaks only in formal language
   • The model forgets past conversations
3. Why does the memory layer in companion apps raise privacy concerns?
   • It deletes all conversations immediately
   • It stores intimate details about the user as data
   • It shares messages publicly by default
   • Działa tylko w trybie offline
4. According to the guide, why is an outright ban often ineffective?
   • Zakazy są nielegalne
   • Teens can easily find alternative apps
   • Companion apps cannot be uninstalled
   • Parents cannot see which apps are installed
5. Why do safety classifiers struggle in long role-play conversations?
   • Działają tylko w języku angielskim
   • Harmful content can build gradually and classifiers miss context
   • They are turned off after ten messages
   • They only scan images
6. What is the weakness of most age gates on these apps?
   • They require a passport
   • They rely on self-declared birthdates that are easy to bypass
   • They block all adults
   • They require a school email
7. Which opening question fits the guide's recommended approach?
   • 'Why are you wasting time on that?'
   • 'What do you like about talking to it?'
   • 'Give me your phone right now.'
   • 'Only lonely people use those apps, right?'
8. Which warning signs does the guide say parents should watch for?
   • Better grades and more hobbies
   • Withdrawal from friends, secrecy or sleep loss
   • Asking for a new phone case
   • Using the app during daytime