AI model families: DeepSeek and the China AI scene

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-builders-model-families-AI-and-deepseek-teen

1. What is DeepSeek?

   1. A cloud computing service for businesses
   2. A Chinese AI research laboratory that creates language models
   3. A video game development studio
   4. An American social media company

2. What surprised the AI industry about DeepSeek's models?

   1. They decided to shut down their research
   2. They could only generate images, not text
   3. They required expensive supercomputers that no one else could afford
   4. They achieved performance comparable to OpenAI while being much cheaper to build

3. What is DeepSeek R1 specialized in?

   1. Generating artwork and photographs
   2. Composing music and poetry
   3. Translating between human languages
   4. Mathematical reasoning and writing computer code

4. What are "thinking" tokens that DeepSeek displays?

   1. A type of error message when the AI fails
   2. Visible steps showing how the AI worked out its answer
   3. Special tokens that make the AI run faster
   4. Tokens that automatically improve the AI's accuracy

5. Why is the lesson's note about "AI leadership isn't just American" important?

   1. It means AI only exists in research labs
   2. It explains why American AI companies are better than others
   3. It proves that only the US can build language models
   4. It shows that China and other countries are also leading in advanced AI development

6. What was the "hard math problem" in the lesson used to demonstrate?

   1. How DeepSeek shows its reasoning process step by step
   2. That math problems make the AI crash
   3. That DeepSeek cannot do math
   4. That only easy math works on AI models

7. What makes DeepSeek's cost achievement significant?

   1. The cost did not matter to their development
   2. They could only afford low-quality hardware
   3. They spent more money than any other company
   4. They built AI models that match top performance while spending much less money

8. What does the lesson suggest users do with DeepSeek's free chat?

   1. Give the AI difficult problems to solve and watch how it thinks
   2. Use it to generate random images
   3. Pay a subscription to access basic features
   4. Only read about the AI without using it

9. Why was the release of DeepSeek described as a "shock" to the industry?

   1. Because they released a worse product than expected
   2. Because they copied GPT models exactly
   3. Because their affordable models performed as well as the most expensive American AI
   4. Because no one knew AI could do math

10. What does it mean that DeepSeek R1 is "open"?

    1. Anyone can access and use it for free
    2. It is only available to Chinese government officials
    3. It can only be used by companies that pay extra
    4. It requires special security clearance

11. What comparison did the lesson suggest making between DeepSeek and GPT-5?

    1. Checking which one has more colors
    2. Counting how many words they can write
    3. Testing how fast they can draw pictures
    4. Comparing their logic and reasoning abilities

12. What was the main takeaway from trying DeepSeek's free chat?

    1. Learning that free tools are always low quality
    2. Seeing the global competition in AI firsthand
    3. Understanding that AI only works in America
    4. Proving that math problems are impossible

13. What does the lesson say about the cost of building AI models?

    1. DeepSeek proved high performance is possible at lower cost
    2. Lower cost always means worse AI
    3. All AI models cost the same amount to build
    4. AI models are free to build

14. Why might seeing DeepSeek's reasoning trace be useful?

    1. It hides the actual answer from users
    2. Users can understand how the AI arrived at its answer
    3. It makes the answer appear faster
    4. It proves the AI is guessing randomly

15. What did the lesson mean by the "global AI race"?

    1. AI models competing in actual races
    2. Competition between different countries to build the best AI
    3. Countries racing to build physical robots
    4. A running competition between AI programs