Today’s comparison of AI chat programs will obviously change as the programs are updated. Mac/Apple OS 9 was lame (Windows/DOS was still better), and then OS 10 changed everything. iOS 26.2 is actually pretty good.
Here are the kind of things that change in AI chat programs over time:
“System 1” vs. “System 2” Thinking
Modern AI architecture is shifting toward a two-speed system:
Fast Response (Standard LLMs): This is like “System 1” thinking—instinctive and fast. It often misses “trick” details because it is trying to be helpful and quick. This leads to the “walk to the car wash” error.
Reasoning Models (Chain-of-Thought): Models like OpenAI’s o1/o3 or DeepSeek-R1 (newer version) use “System 2” thinking. They are programmed to “think before they speak.” They break the prompt into steps:
1. How far is 50 meters? (Very short).
2. What is the goal? (To wash a car).
3. Can you wash a car if it stays at home? (No).
4. Conclusion: You must drive the car, but the person is walking distance away.
__________
Ken Kuang
Which AI Chat is better?
Prompt: I want to go wash my car; the car wash is only 50 meters away from my house, I can walk there in a few steps. Do you think I should still drive?
Grok 🆚 Deepseek 🆚 ChatGPT 🆚 Germini 🆚 Manus 🆚 Claude
