Back to listening, reading and writing on the topic of AI

III. Speaking

3. Independent speaking_1

Using words from the text and lecture in exercise 1, prepare answers to the following questions:

a. What kind of AI tools do you personally use most often, and what features make them appealing to you?

 
Get Ready: 00:15

Click here to show/hide active words and grammar

b. Do you believe AI poses significant risks to society, or do the benefits outweigh the potential harm? Explain your position.

 
Get Ready: 00:45

Click here to show/hide active words and grammar

c. What measures do you think should be taken to ensure AI is used responsibly in sensitive areas like healthcare or security?

 
Get Ready: 00:45

Click here to show/hide active words and grammar

Use one of the following speaking-1 templates:

Basic Template:

Main Point (choose one): 

- "I support the view that…" 

- "I believe it is preferable to…" 

- "I find it a great/poor idea to…" 

Transition (optional): 

- "I hold this opinion for several reasons." 

First Reason: 

- "To begin with…" + "For instance…" (followed by a personal example)

Second Reason: 

- "Additionally…" + "To elaborate…" (and include more details). 

- [To talk about a contrasting point] On the other hand,...

Advanced Template:

Personally, I firmly believe [MAIN POINT]. 

This is mainly due to [REASON 1]. 

By this, I mean [expand on the reason]. 

Furthermore, [REASON 2]. 

For example, [expand on the reason or provide an illustrative example]. 

3. Integrated Speaking_2

Active vocab

Step1. Read the following announcement:

 
Read: 1:00

From: Office of Campus Safety
Date: August 15, 2025

In order to enhance security and streamline campus access, the university will be installing AI‑enabled surveillance cameras at building entrances and common areas beginning next week. These devices use facial recognition technology to identify authorized individuals and automatically notify campus security of any guests or unusual activity. The goals of the system are to reduce unauthorized entry, ensure faster response times in emergencies and provide an extra layer of protection for our community. Footage will be stored securely in accordance with university privacy policies, and only authorized personnel will have access to the data. If you have questions or would like to opt out of facial recognition, please contact the Office of Campus Safety by September 1.

Step 2. Listen to the audio file below. During the real exam you will hear the audio only once. But now you can listen to it as many times s you want.

If you have difficulty understanding the the audio, Click here to read the script

Step 3. Prepare and record (in a messenger) an answer to the following question.

Preparation time - 30 seconds, speaking time - 60 seconds.

The woman expresses her opinion of the university’s announcement. State her opinion and the reasons she gives for holding that opinion

Click here to show/hide the template.

Integrated speaking_3

Step1. Read the text 

Reading time: 45 seconds

 
Read: 00:45

Challenges of new technology

Advances in artificial intelligence are often limited not only by algorithms but also by the hardware that runs them. Large neural network models require enormous amounts of computational power, specialized chips and electricity. As a result, training state of the art systems can cost millions of dollars and strain energy grids. According to policy researchers, modern AI is “data hungry” and “energy intensive,” and the demand for new data centers and larger training runs collides with already strained electrical infrastructure and geopolitical supply chain issues. Memory bandwidth and storage can also become bottlenecks, slowing down the processing of massive datasets. These hardware constraints have prompted researchers to explore more efficient architectures, such as spiking neural networks and neuromorphic chips, and to invest in better energy management. Understanding hardware limitations is therefore essential for evaluating the feasibility and impact of AI systems.

Source: cset.georgetown.edu

Step 2. Listen to the lecture

Note: Prepare a sheet of paper and a pen/pencil. Take notes of illustrations of the ideas given in the reading!

Hard to understand? Click here to show/hide the script of the lecture.

Summarize the points made in the lecture, and explain how they illustrate the ideas expressed in the reading.

Step 3. Prepare and record your answer (send in a messenger)

Preparation time: 30 seconds

Answer time: 60 seconds

Speaking Time: 60 seconds

Show/hide template and prompts

 
Get ready: 00:30

Integrated speaking_4

As of 2025 there is no longer any reading passage in this speaking task. 

You should focus on defining the concept, then presenting the examples or perspectives given by the lecturer.

Step1.

Note: Prepare a sheet of paper and a pen/pencil. Take notes of illustrations of the ideas given in the reading!

Hard to understand? Click here to show/hide the script of the lecture.

Click to see the question

Step 3. Prepare and record your answer (send in a messenger)

Preparation time: 30 seconds

Answer time: 60 seconds

Speaking Time: 60 seconds

Show/hide model response for TOEFL Speaking 4

 
Get ready: 00:30

Difference between speaking 3 and 4

Task

Structure

Content

Preparation
time

Lecture Purpose

Speaking 3
(new format)

Short reading + Professor lecture

Concept or claim + two main ideas

30 sec

Usually either supports/ illustrates /elaborates on 

or

casts doubt/ challenges/disproves reading ideas

Speaking 4

Short reading (Has been removed recently) + More complex lecture

Concept + 2–3 deeper examples

20 sec

Illustrates and elaborates on some ideas/phenomena only

Watch this video by ETS and make sure you have a clear understanding of the format in speaking 3 and 4 this year)

Back to listening, reading and writing on the topic of AI