You must log in or register to comment.
Didn’t openAI said we need to be ok with AI stealing “intellectual property”?
Anyway
Oh no!
There’s no link. Did they? Was it trained on ChatGPT answers? Because a lot of small team LLMs are trained on ChatGPT answers.
Yeah okay so they’re literally just talking about using ChatGPT answers in the training set. Everyone does this, it’s completely fair game. Hell, the R1 Qwen that everyone is using now to run ‘R1’ on their own device is Qwen distilled with R1.
(Not that I believe in copyright to begin with).
Exactly and it’s just stunning hypocrisy on the part of OpenAI who have been sucking up all data with complete disregard for ownership. Also agree that copyright wouldn’t even be necessary in a socialist society, while it only ends up being abused under capitalism.
I would argue that a new socialist country should retain copyright, but drastically shorten its duration (back to the original 7 years or less). The reason being that copyright is the mechanism that protects the copyleft. Without copyright, the GPL would no longer be enforceable. I’m no fan of copyright and would love to see it eventually abolished, but it has its uses today.
Yeah, capitalist relations need to be abolished first.
Yeah the real problem with this is that ChatGPT is going to give a lot of garbage answers when it comes to political questions, and if DeepSeek takes in too much of that, it will also start spitting out very bad takes. It’s like copying off of your classmate in a test in school, but that classmate is only known for getting good grades in some classes but not so good grades in others. You need to be careful to only copy on the subjects you know they’re actually competent at.
Lol, what’s ClosedAI going to do? Hide ChatGPT’s answers behind a summary (like with the CoT) 🤣
What is CoT?
Notable OpenAI’s O1 cot is hidden with only a summary available sometimes
From Deepseek:
Chain of Thought (CoT) in LLMs refers to a prompting technique that guides large language models to articulate intermediate reasoning steps when solving a problem, mimicking human-like logical progression. Here’s a concise breakdown:
-
Purpose: Enhances performance on complex tasks (e.g., math, logic, commonsense reasoning) by breaking problems into sequential steps, reducing errors from direct, unstructured answers.
-
Mechanism:
- Prompting Strategy: Users provide examples (few-shot) or explicit instructions (zero-shot, e.g., “Let’s think step by step”) to encourage step-by-step explanations.
- Output Structure: The model generates tokens sequentially, simulating a reasoned pathway to the answer, even though internal processing remains parallel (via transformer architecture).
-
Benefits:
- Accuracy: Improves results on multi-step tasks by isolating and addressing each component.
- Transparency: Makes the model’s “thinking” visible, aiding debugging and trust.
-
Variants:
- Few-Shot CoT: Examples with detailed reasoning are included in the prompt.
- Zero-Shot CoT: Direct instructions trigger step-by-step output without examples.
- Self-Consistency: Aggregates answers from multiple CoT paths to select the most consistent one.
-
Effectiveness: Particularly impactful for tasks requiring structured reasoning, while less critical for simple queries. Research shows marked accuracy gains in benchmarks like math word problems.
In essence, CoT leverages the model’s generative capability to externalize reasoning, bridging the gap between opaque model decisions and interpretable human problem-solving.
Example:
-
