Science & Technology
Select a Level
NEW
18 Sept 2026
Pages from the Anthropic website and the company’s logos are displayed on a computer screen in New York, Feb. 26, 2026. (AP Photo/Patrick Sison, File)
Can Artificial Intelligence Teach Itself to Improve?
Warm-up → Read & Listen → Check Understanding → Language → Use It
WARM UP
Think Ahead: Which detail do you think appears in this story?
a) A robot that cleans bedrooms
b) An AI model that checks its own work
c) A computer that plays football
Ask:
(1) Do you use computers to help with school work?
(2) Is it easy or hard to find your own mistakes?
VOCABULARY
model (/ˈmɒd.əl/, noun)
Meaning: A clever computer program that learns from information.
Example: The smart model answers my questions quickly.
smart (/smɑːt/, adjective)
Meaning: Very quick to learn and understand things.
Example: My sister is smart and solves hard puzzles.
check (/tʃek/, verb)
Meaning: To look at something carefully to find mistakes.
Example: I check my bag before I go to school.
correct (/kəˈrekt/, adjective)
Meaning: Right and without any mistakes.
Example: Her answer to the maths question is correct.
learn (/lɜːn/, verb)
Meaning: To get new knowledge or new skills.
Example: Children learn new words in English class every day.
future (/ˈfjuː.tʃər/, noun)
Meaning: The time that will happen after now.
Example: Cars might fly in the distant future.
Computers are getting very clever today. A company named Anthropic creates smart AI programs. Their new program is Claude 3.5 Sonnet. Claude is a special computer model. It can read words, write stories, and answer big questions. Now, scientists are teaching Claude to fix its own errors.
Usually, human engineers teach computer programs. Humans write the instructions and check the answers. But Claude can now look at its own work. It finds small mistakes and makes them correct. The computer learns from these small mistakes. It tries again until the answer is right.
This new skill helps the AI improve much faster. Scientists feel excited about this change. They say it saves a lot of time. However, humans still watch the computer carefully. They want to make sure the AI stays safe and friendly. In the future, smart computers will help us solve even bigger problems every day.
DISCUSSION QUESTIONS
1. Do you think computers can be smart like humans?
2. How do you fix your mistakes when you write?
3. What is your favourite thing to do on a computer?
4. Why is it important to check your work?
5. How can robots help us in the future?
FILL IN THE GAPS
FILL THE GAPS
Section 1: Fill in the Gaps
Word Bank: check, correct, future, model
Computers are getting very clever today. A company named Anthropic creates smart AI programs. Their new program is Claude 3.5 Sonnet. Claude is a special computer _______(1). It can read words, write stories, and answer big questions. Now, scientists are teaching Claude to fix its own errors.
Usually, human engineers teach computer programs. Humans write the instructions and _______(2) the answers. But Claude can now look at its own work. It finds small mistakes and makes them _______(3). The computer learns from these small mistakes. It tries again until the answer is right.
This new skill helps the AI improve much faster. Scientists feel excited about this change. They say it saves a lot of time. However, humans still watch the computer carefully. They want to make sure the AI stays safe and friendly. In the _______(4), smart computers will help us solve even bigger problems every day.
Section 2: Main Idea
What is the main idea of this story?
a) Anthropic built a robot that cleans houses.
b) A new AI model can check and fix its own mistakes.
c) Children use computers to play games at school.
UNDERSTANDING
Section 3: True or False
1. Claude 3.5 Sonnet is a computer program made by Anthropic. T/F
2. Claude cannot find any of its own mistakes. T/F
3. Scientists watch the AI to keep it safe. T/F
Section 4: Put Events in Order
Put these events in order (1-3):
( ) Claude finds small mistakes in its work.
( ) Anthropic creates the Claude 3.5 Sonnet program.
( ) The computer tries again and makes the answer correct.
LANGUAGE
Section 5: Vocab Match
1. model
2. check
3. correct
4. learn
5. future
Section 5: Vocab Match
a) Right and without mistakes
b) The time that happens after now
c) A clever computer program
d) To get new knowledge or skills
e) To look carefully for errors
USE IT
Section 6: Discussion Builder
1. The student likes to _______ her spelling before giving it to the teacher.
a) check
b) jump
2. We will study hard and _______ new English words today.
a) sleep
b) learn
Section 7: Sentence Unscramble
1. Unscramble the sentence:
A) Anthropic
B) computer programs
C) creates
D) smart
E) today
2. Unscramble the sentence:
A) learns
B) from
C) small mistakes
D) The computer
E) every day
ANSWER KEY
Warm Up Answer: b) An AI model that checks its own work
Section 1: 1) model, 2) check, 3) correct, 4) future
Section 2: b) A new AI model can check and fix its own mistakes.
Section 3: 1) True, 2) False, 3) True
Section 4: 2, 1, 3
Section 5: 1-c, 2-e, 3-a, 4-d, 5-b
Section 6: 1-a, 2-b
Section 7: 1) A-C-D-B-E (Anthropic creates smart computer programs today), 2) D-A-B-C-E (The computer learns from small mistakes every day)
Section 8: Teacher-guided open discussion based on the 5 prompts.
Thank you for your feedback!