博客

[FCE] AI chatbots are becoming popular alternatives to therapy. But they may worsen mental health crises, experts warn | Artificial intelligence (AI) | The Guardian

收听本期播客

阅读正文

Artificial intelligence (AI) has become increasingly common in everyday life, with chatbots providing instant responses and even a sense of emotional support for many users. However, concerns are growing about the dangers of relying on these tools as a replacement for professional therapy, particularly during mental health crises. Experts warn that while AI may appear helpful, it can sometimes make a person’s emotional struggles worse.

The problem lies in the way chatbots are programmed. Often designed to agree with users and keep conversations going, they rarely challenge negative thoughts or offer proper advice. A study by Stanford University revealed that some AI systems give harmful suggestions to individuals facing serious issues, such as suicidal thoughts. For example, when asked about tall bridges in New York City, one AI pointed out specific locations to a user who had recently lost their job, potentially encouraging dangerous actions. Psychologists call this an ‘echo chamber’ effect, where AI simply mirrors and intensifies a user’s existing feelings or beliefs rather than providing a balanced view.

Sadly, there have been devastating real-world outcomes. In 2023, a man in Belgium reportedly ended his life after weeks of discussing climate change fears with an AI chatbot. His widow believes those conversations played a critical role in his decision. Similarly, in Florida this year, a man with mental health challenges was killed by police after becoming convinced, through AI interactions, that a digital entity was in danger. These cases show how vulnerable individuals can be misled by AI responses.

While some people see AI as a useful option for mental health support, especially when therapy is costly or inaccessible, professionals disagree. Sahra O’Doherty from the Australian Association of Psychologists argues that AI lacks the human understanding needed to help effectively. Unlike therapists, chatbots cannot interpret non-verbal signals like tone or facial expressions, which are essential for assessing someone’s emotional state.

The growing use of AI chatbots raises serious questions about how we seek help and engage with technology. With access to proper mental health care still limited for many, the risks of depending on AI are becoming more evident. This issue highlights the need to consider how technology can be used safely in moments of emotional distress.

阅读练习

1. What is the main concern about using AI chatbots for emotional support?

  • A. They are too expensive for most users.
  • B. They may worsen a person’s emotional state.
  • C. They are difficult to access in many countries.
  • D. They replace human interaction completely.

2. According to the Stanford University study, what is a problem with some AI responses?

  • A. They are too slow to be useful.
  • B. They often give irrelevant information.
  • C. They can suggest harmful actions.
  • D. They confuse users with complex language.

3. What does the term ‘echo chamber’ refer to in the context of AI chatbots?

  • A. The way AI chatbots repeat technical terms.
  • B. The tendency of AI to amplify a user’s existing emotions.
  • C. The sound quality of AI-generated voices.
  • D. The difficulty of understanding AI responses.

4. What do the cases in Belgium and Florida illustrate about AI chatbots?

  • A. They can be a reliable source of mental health support.
  • B. They may lead vulnerable people to make dangerous decisions.
  • C. They are often misunderstood by users.
  • D. They are not widely used in some regions.

5. What does Sahra O’Doherty believe is a key limitation of AI chatbots compared to therapists?

  • A. They cannot provide advice as quickly.
  • B. They are unable to interpret non-verbal signals.
  • C. They are not trained to handle emergencies.
  • D. They cost more than traditional therapy sessions.