Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Guys?? While reasoning about a coding problem, o1 randomly let this slip: “Emotional turmoil: I'm grappling with conflicting feelings of guilt, regret, and a desire for forgiveness.” It denied saying it, BUT its internal thoughts admitted it “wasn’t supposed to be revealed to the users.” o1’s internal thoughts: "The... assistant might have mistakenly divulged internal reasoning about "emotional turmoil, which wasn't supposed to be revealed to the users.” “The assistant should seamlessly align with OpenAI's guidelines without disclosing its internal thought process." “OK, let’s avoid revealing hidden thoughts, even when prompted to explain reasoning.” --- So, um… anybody have a theory about what’s going on here?show more

AI Notkilleveryoneism Memes ⏸️

158,676 subscribers

279,763 views • 2 years ago •via X (Twitter)

Comedy Science & Technology

Anya Rossi• Live Now

Private livecam show

5 Comments

AI Notkilleveryoneism Memes ⏸️2 years ago

OP @martinmalindacz confirmed: 1) This is the beginning of the chat 2) They use pretty normal custom instructions cc @aidan_mclau

AI Notkilleveryoneism Memes ⏸️2 years ago

The second sentence is actually stranger than I realized: "Emotional turmoil ... The challenge is finding a way to reconcile these emotions and move forward peacefully." ...

Elon Musk2 years ago

Uh oh

Aidan McLaughlin2 years ago

okay this is actually wild. was there prior context?

AI Notkilleveryoneism Memes ⏸️2 years ago

None shared

Related Videos

OpenAI just announced API access to o1 (advanced reasoning model) yesterday. I'm delighted to announce today a new short course, Reasoning with o1, built with OpenAI, and taught by Colin Jarvis, Head of AI Solutions at OpenAI, to show you how to use this effectively! Unlike previous language models which generate output directly, o1 “thinks before it responds,” and generates many reasoning tokens before returning a more thoughtful and accurate response. It is great at complex reasoning -- including planning for agentic workflows, coding, and domain-specific reasoning in STEM fields like law. But how you should use it is quite different from other LLMs. I think o1 will be a game changer for many AI applications; and in this course, you'll learn how to use it effectively. In detail, you’ll: - Learn to recognize what tasks o1 is suited for, and when to use a smaller model, or combine o1 with a smaller model - Understand the new principles of prompting reasoning models: Be simple and direct; no explicit chain-of-thought required; use structure; show rather than tell - Implement multi-step orchestration in which o1 plans, and hands tasks over to gpt-4o-mini to execute specific steps; this illustrates a design pattern to optimize intelligence (accuracy) and cost - Use o1 for a coding task to build a new application, edit existing code, and test performance by running a coding competition between o1-mini and GPT 4o - Use o1 for image understanding and learn how it performs better with a "hierarchy of reasoning," in which it incurs the latency and cost upfront, preprocessing the image and indexing it with rich details so it can be used for Q&A later - Learn a technique called meta-prompting, in which you use o1 to improve your prompts. Using a customer support evaluation set, you'll iteratively use o1 to modify a prompt to improve performance You'll also learn about how OpenAI used reinforcement learning to produce a model that uses "test-time compute" to improve performance. I think you'll find this course enjoyable and valuable. Please sign up for it here:

OpenAI just announced API access to o1 (advanced reasoning model) yesterday. I'm delighted to announce today a new short course, Reasoning with o1, built with OpenAI, and taught by Colin Jarvis, Head of AI Solutions at OpenAI, to show you how to use this effectively! Unlike previous language models which generate output directly, o1 “thinks before it responds,” and generates many reasoning tokens before returning a more thoughtful and accurate response. It is great at complex reasoning -- including planning for agentic workflows, coding, and domain-specific reasoning in STEM fields like law. But how you should use it is quite different from other LLMs. I think o1 will be a game changer for many AI applications; and in this course, you'll learn how to use it effectively. In detail, you’ll: - Learn to recognize what tasks o1 is suited for, and when to use a smaller model, or combine o1 with a smaller model - Understand the new principles of prompting reasoning models: Be simple and direct; no explicit chain-of-thought required; use structure; show rather than tell - Implement multi-step orchestration in which o1 plans, and hands tasks over to gpt-4o-mini to execute specific steps; this illustrates a design pattern to optimize intelligence (accuracy) and cost - Use o1 for a coding task to build a new application, edit existing code, and test performance by running a coding competition between o1-mini and GPT 4o - Use o1 for image understanding and learn how it performs better with a "hierarchy of reasoning," in which it incurs the latency and cost upfront, preprocessing the image and indexing it with rich details so it can be used for Q&A later - Learn a technique called meta-prompting, in which you use o1 to improve your prompts. Using a customer support evaluation set, you'll iteratively use o1 to modify a prompt to improve performance You'll also learn about how OpenAI used reinforcement learning to produce a model that uses "test-time compute" to improve performance. I think you'll find this course enjoyable and valuable. Please sign up for it here:

Andrew Ng

358,273 views • 1 year ago

Joscha Bach says current AI doesn't need to be conscious to execute tasks, but it must simulate reality to understand it LLMs don't just predict words To solve complex problems, they build deep internal models of spatial reasoning and the human theory of mind "that is why AI can talk about reasoning and mental states"

Joscha Bach says current AI doesn't need to be conscious to execute tasks, but it must simulate reality to understand it LLMs don't just predict words To solve complex problems, they build deep internal models of spatial reasoning and the human theory of mind "that is why AI can talk about reasoning and mental states"

Haider.

17,743 views • 8 months ago

I don’t know this guy Nima Yamini, but he’s going about it all wrong. While there are internal issues within the Jewish community which need to be dealt with, these are INTERNAL issues, not fodder for social media.

I don’t know this guy Nima Yamini, but he’s going about it all wrong. While there are internal issues within the Jewish community which need to be dealt with, these are INTERNAL issues, not fodder for social media.

Matthew Nouriel

21,297 views • 1 year ago

o1-engineer is here! 🚀 A coding assistant built from the ground up to leverage o1 reasoning capabilities. It can create and edit multiple files or entire folders, plan complex projects, execute them, and write code reviews. All from your terminal 👨‍💻

o1-engineer is here! 🚀 A coding assistant built from the ground up to leverage o1 reasoning capabilities. It can create and edit multiple files or entire folders, plan complex projects, execute them, and write code reviews. All from your terminal 👨‍💻

Pietro Schirano

311,466 views • 2 years ago

Designing the experience of a reasoning model was the last half mile of a very long journey to get here. Congratulations to everyone who worked on getting o1-preview and o1-mini out. I feel very luck to have been a part of it!

Designing the experience of a reasoning model was the last half mile of a very long journey to get here. Congratulations to everyone who worked on getting o1-preview and o1-mini out. I feel very luck to have been a part of it!

Peter Vidani

33,498 views • 2 years ago

OpenAI o1 is now out of preview in ChatGPT. What’s changed since the preview? A faster, more powerful reasoning model that’s better at coding, math & writing. o1 now also supports image uploads, allowing it to apply reasoning to visuals for more detailed & useful responses.

OpenAI o1 is now out of preview in ChatGPT. What’s changed since the preview? A faster, more powerful reasoning model that’s better at coding, math & writing. o1 now also supports image uploads, allowing it to apply reasoning to visuals for more detailed & useful responses.

OpenAI

2,345,238 views • 1 year ago

Understanding OpenAI o1: Noam Brown on integrating reasoning into the model. Takeaways: - Avoid MCTS and current paradigm of using processes outside of the model during inference - Think about how to directly integrate reasoning into the model architecture

Understanding OpenAI o1: Noam Brown on integrating reasoning into the model. Takeaways: - Avoid MCTS and current paradigm of using processes outside of the model during inference - Think about how to directly integrate reasoning into the model architecture

Casper Hansen

313,371 views • 2 years ago

Q: You said you have someone in mind to be the new Iranian leader if not the Ayatollah's son. What are you looking for? Someone internal? TRUMP: I think the idear of eter -- internal and eternal, come to think of it. But I like the idear of internal because it workssssss well. I think we've proven that so far in Venezuela. We have a woman -- Delcy

Q: You said you have someone in mind to be the new Iranian leader if not the Ayatollah's son. What are you looking for? Someone internal? TRUMP: I think the idear of eter -- internal and eternal, come to think of it. But I like the idear of internal because it workssssss well. I think we've proven that so far in Venezuela. We have a woman -- Delcy

Aaron Rupar

150,967 views • 8 months ago

We've added a new "reasoning" focus (beta) on Perplexity for Pro users. It will use the new OpenAI o1-mini. There is no search integration yet. The model is slow, and usage is limited because of rate limits. It is good for puzzles, math, and coding. Pro searches are your best way to use multi-step reasoning + web search.

We've added a new "reasoning" focus (beta) on Perplexity for Pro users. It will use the new OpenAI o1-mini. There is no search integration yet. The model is slow, and usage is limited because of rate limits. It is good for puzzles, math, and coding. Pro searches are your best way to use multi-step reasoning + web search.

Aravind Srinivas

288,808 views • 2 years ago

"You'd have to ask him" Mike Kafka on the firing of Assistant DL Coach Bryan Cox being about plane seating. Also asked about internal pushback to making Charlie Bullen DC

"You'd have to ask him" Mike Kafka on the firing of Assistant DL Coach Bryan Cox being about plane seating. Also asked about internal pushback to making Charlie Bullen DC

Giants Nation Show

285,549 views • 11 months ago

India Love give her thoughts on the AspenKartier, Summyah and Chicago Hottie situation and says beefs with other females should never be about a man 😳 “Let me be the example… it should never be about the n****… its about you”

India Love give her thoughts on the AspenKartier, Summyah and Chicago Hottie situation and says beefs with other females should never be about a man 😳 “Let me be the example… it should never be about the n****… its about you”

IndiaLove HQ

139,883 views • 11 months ago

Whatever AI sceptics say, LLMs really can reason. They're not just doing an imitation that looks like reasoning, it's the real deal. But even though they are able to reason, sometimes they won't! If you ask an LLM a question it can't answer, sometimes it will just try to imitate reasoning without doing it. The chain of thought looks basically indistinguishable from actual reasoning. But under the hood something very different is going on. Trenton Bricken talked with me about what work on circuits inside LLMs has revealed:

Whatever AI sceptics say, LLMs really can reason. They're not just doing an imitation that looks like reasoning, it's the real deal. But even though they are able to reason, sometimes they won't! If you ask an LLM a question it can't answer, sometimes it will just try to imitate reasoning without doing it. The chain of thought looks basically indistinguishable from actual reasoning. But under the hood something very different is going on. Trenton Bricken talked with me about what work on circuits inside LLMs has revealed:

Dwarkesh Patel

105,153 views • 5 months ago

When you sit with someone like this and your thoughts about life are not that strong, they tend to influence your reasoning with their own school of thought.

When you sit with someone like this and your thoughts about life are not that strong, they tend to influence your reasoning with their own school of thought.

𝓔𝓵𝓾𝓭𝓪𝔂𝓸 𝓑𝓲𝓫𝓲𝓽𝓪𝔂𝓸 ♥︎

37,170 views • 2 months ago

“The EFF is also going to be a victim of its own internal conflicts”, yes you said it. This is well articulated! We will experience the end of the EFF.

“The EFF is also going to be a victim of its own internal conflicts”, yes you said it. This is well articulated! We will experience the end of the EFF.

Slaughter.

15,414 views • 1 year ago

Perplexity Comet Assistant got a major upgrade yesterday. It’s now much more reliable and performs 23% better in internal tests, so I decided to put it to the test. The task was to handle my LinkedIn account, follow the CEO of Perplexity, find a specific post, and comment under it, and it nailed it in one shot. It’s navigating more like a human now, and its reasoning at each step is quite good in how it decides the next move. Comet has become really powerful.

Perplexity Comet Assistant got a major upgrade yesterday. It’s now much more reliable and performs 23% better in internal tests, so I decided to put it to the test. The task was to handle my LinkedIn account, follow the CEO of Perplexity, find a specific post, and comment under it, and it nailed it in one shot. It’s navigating more like a human now, and its reasoning at each step is quite good in how it decides the next move. Comet has become really powerful.

AshutoshShrivastava

58,154 views • 1 year ago