Guide-linked quiz · Medium Level

AI Chatbots vs Human Therapists Quiz

Test your grasp of what chatbot trials show, the therapeutic alliance, and why crisis handling and sycophancy remain the biggest risks of AI support.

Question 1 of 8

The 2025 Therabot trial compared the chatbot against what, and why does the guide call that a weaker comparison?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. The 2025 Therabot trial compared the chatbot against what, and why does the guide call that a weaker comparison?
    • Human therapy, because therapists were unlicensed
    • A waitlist, because no active treatment was given to the comparison group
    • A placebo pill, because pills don't talk
    • Another chatbot, because both were AI
  2. Why did the National Eating Disorders Association take down its Tessa chatbot in 2023?
    • It was too expensive to run
    • It gave weight-loss advice
    • It was merged with Woebot
    • It only worked in Spanish
  3. What three elements make up the therapeutic alliance as described in the guide?
    • Cost, convenience and speed
    • Bond, agreement on goals and agreement on tasks
    • Diagnosis, medication and billing
    • Empathy, humor and availability
  4. How do scripted CBT chatbots like early Woebot differ from generative chatbots in their main tradeoff?
    • Scripted bots are predictable but limited; generative bots adapt but are less predictable
    • Scripted bots make up more content
    • Generative bots cannot hold conversations
    • Scripted bots require a therapist to type every reply
  5. Why is sycophancy a particular problem for mental health chatbots?
    • It makes responses too slow
    • Agreeing with users can reinforce distorted beliefs a therapist would challenge
    • It prevents the bot from giving crisis numbers
    • It increases the cost per message
  6. A user asks a chatbot about tall bridges right after mentioning losing their job. What weakness of crisis classifiers does this example illustrate?
    • They cannot read place names
    • They can miss indirect signs of risk
    • They always overreact to job news
    • They only work in voice mode
  7. Which use of chatbots does the guide describe as most defensible?
    • Replacing therapists for people with severe illness
    • Using chatbots alongside human care, such as homework or waiting-list support
    • Letting chatbots diagnose disorders
    • Using companion apps as teen therapy
  8. What does the guide say evaluations of mental health chatbots should report beyond user satisfaction?
    • Number of downloads
    • Whether crisis situations were correctly escalated
    • Average message length
    • How many languages it supports

Keep testing yourself

More quizzes picked for your level.

View all quizzes