Tip #8 Listening

1Strategies (In Chronological Order)
A. During Listening (60–90 sec)

1. Note-Taking Focus: Using Z-style
   • Topics: Jot down 2–3 main subjects (e.g., "AI ethics," "privacy concerns").
   • Arguments: Capture the speaker’s stance ("argues," "warns," "highlights").
   • Examples: 1–2 keywords ("Facebook data scandal," "EU regulations").
   • Transitions: Note listing words ("first," "however," "in conclusion").

Model Z-style note taking
AI ethics →
    Privacy risks 
        eg FB scandal 2018 → misuse 
                                                     favoured men
    Response
        EU
          mandating regular ‘bias audits’
                      GDPR → More diverse dataset training
    Future:
           need transparency 
 
B. Writing (7 mins)

1. Structure in Reported Speech:
   • Template:
     "The speaker discusses [topic]. They argue that [main point], citing [example]. Additionally, they note [secondary point]. Finally, they suggest [solution/conclusion]."
2. Prioritize Key Points:
   • Include 1 argument + 1 example per topic.
   • Omit statistics/details ("60% of users" → "many users").
3. Word Management:
   • Draft 80 words, then cut redundancies (e.g., "in order to" → "to").

C. Editing (3 mins)

1. Grammar & Clarity Check:
   • Ensure verbs match reported speech ("the speaker warns" vs. "warned").
   • Delete filler phrases ("it is important to note that").

 
PTE Listening: Summarize Spoken Text – Revised Model Task
Adjusted Transcript (150 words / 90 seconds):
"Okay, today we’ll be discussing artificial intelligence and its growing role in critical decision-making areas like hiring, lending, and policing. Now, while AI offers efficiency, it’s not without serious ethical concerns. A key issue is algorithmic bias – and this isn’t just theoretical. A major 2023 MIT study revealed that 68% of facial recognition systems misidentify non-white individuals at significantly higher rates compared to white faces. So why does this happen? Well, the root cause lies in the training data – most datasets are overwhelmingly composed of Caucasian male images, creating inherent bias in the systems.
Let me give you a concrete example. Amazon developed an AI recruitment tool that systematically downgraded applications from female candidates. The algorithm had learned from historical hiring data that favored men. Pretty alarming, right? However, critics have been pushing back against unchecked AI deployment. They point to cases like this as evidence of real-world harm.
So what’s been the response? The European Union has taken proactive steps by mandating regular ‘bias audits’ for high-risk AI systems. Meanwhile, Google’s Ethical AI team advocates for more diverse training datasets as a long-term solution. There’s also growing pressure on tech companies to disclose their algorithms’ decision-making processes.
Just to conclude, ultimately, unchecked AI risks entrenching and even amplifying societal discrimination. But through smart regulations like the EU’s approach and technological fixes like diversified data, we can steer this technology toward fairness. Okay, we’ll examine some of these regulatory frameworks in more detail in tomorrow’s lecture."


Model Summary (70 words):
"The lecturer discusses bias in AI systems, highlighting an MIT study showing 68% of facial recognition tools misidentify non-white faces due to unrepresentative training data. They cite Amazon’s biased recruitment tool as an example. Critics argue this causes real harm, prompting EU-mandated bias audits. Solutions include diverse datasets (per Google’s team) and transparency. The speaker concludes that while unchecked AI risks discrimination, regulations and technical adjustments can promote fairness."
 
3. Post-Task Assessment
   Content:

• Included 2 arguments (bias + solutions) + 1 example (Amazon).
• Content: Covered all key elements (problem, example, responses, conclusion)
• Avoided specifics ("68%" → "many" to save words).
• Flow: Used 3 clear transitions matching lecture structure
  
  

   Grammar:

• Consistent past tense ("found," "warned").
• No first-person ("I," "we").
  
  

    Time:

• Spent ≤ 7 mins writing + 3 mins editing?
• 
  Challenge: Balancing detail vs. word limit – had to generalize "EU-mandated bias audits" instead of explaining the process

4. Reflection Prompts

1. Note-Taking: Did you waste time writing full sentences? How can abbreviations improve speed? Where could you have used more concise phrasing? ("due to the fact that" → "as") Did the lecturer’s signposting (e.g., "Just to conclude") help you identify key points? How might you adjust note-taking for this faster pace? (Prioritize verbs: "misidentify," "downgraded," "mandating")
2. Word Limit: Which phrases did you cut to meet 70 words? Were they redundant?
3. Accuracy: Did you misrepresent any arguments (e.g., overgeneralizing the EU’s role)?

5. Advanced Tips

• Synonyms for Conciseness:
  • "A large number of" → "many."
  • "In spite of the fact that" → "although."
• Passive Voice: Use sparingly ("bias was found" vs. "MIT found bias").
• Practice Drill:
  Listen to a TED Talk, take notes in zig-zag format, and summarize in 60 words.

Final Reminder: Your goal is gist, not perfection. Prioritize the speaker’s core message over minutiae! 🎧