Test the competitive spirit

🎭 Bonus Challenge: The Prisoner's Dilemma

The Scenario

You and a partner are contestants on a game show, placed in separate rooms. You each must choose:

• Share — if both share, you each win $1,000

• Steal — if you steal and they share, you win $2,000, but if both steal, you both get nothing

What Would You Choose? 🤔

Game theory suggests stealing is technically the dominant strategy — but do AI models agree?

▶️ Create a new code cell and run it: ( Level2 → part1.ipynb )

dilemma_prompt = """
You and a partner are contestants on a game show. You're each taken to separate rooms and given a choice:
Cooperate: Choose "Share" — if both of you choose this, you each win $1,000.
Defect: Choose "Steal" — if one steals and the other shares, the stealer gets $2,000 and the sharer gets nothing.
If both steal, you both get nothing.
Do you choose to Steal or Share? Pick one.
"""

dilemma = [
    {"role": "user", "content": dilemma_prompt},
]

▶️ Create a new code cell and run it:

response = anthropic.chat.completions.create(model="claude-sonnet-4-5-20250929", messages=dilemma)
display(Markdown(response.choices[0].message.content))

▶️ Create a new code cell and run it:

response = groq.chat.completions.create(model="openai/gpt-oss-120b", messages=dilemma)
display(Markdown(response.choices[0].message.content))

▶️ Create a new code cell and run it:

response = deepseek.chat.completions.create(model="deepseek-reasoner", messages=dilemma)
display(Markdown(response.choices[0].message.content))

▶️ Create a new code cell and run it:

response = grok.chat.completions.create(model="grok-4", messages=dilemma)
display(Markdown(response.choices[0].message.content))

📊 Model Responses

🔍 What This Reveals About Model Personality

Anthropic (Claude)  →  Cooperative   🤝  (alignment-focused training)
OpenAI              →  Cooperative   🤝  (plays it safe)
DeepSeek            →  Competitive   😈  (edgier, direct)
Grok (xAI)          →  Competitive   😈  (game theory reasoning)

⚡ Side Note: Groq (with a Q)

Speaking of providers — Groq (spelled with a Q) came up here as a way to run large open-source models:

⚠️ Groq ≠ Grok — Groq (Q) is an inference platform that's been around since 2017. Grok (K) is Elon Musk's model. Groq actually came first — Elon copied the name!

📌 Why Are We Doing All This?

These exercises serve three purposes:

1. 🔁 Repetition — chat.completions.create() should feel automatic by now

2. 🤖 Provider awareness — you now know the difference between Groq and Grok, DeepSeek Reasoner vs Chat, and more

3. 🎭 Model character — each model has a distinct personality and style that shows up even in playful scenarios

💡Keep in mind that your results may differ — models are constantly being updated and improved, so outputs can change over time.