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 • 1 year ago •via X (Twitter)

Comedy Science & Technology

Anya Rossi• Live Now

Private livecam show

5 Comments

AI Notkilleveryoneism Memes ⏸️1 year 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 ⏸️1 year 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 Musk1 year ago

Uh oh

Aidan McLaughlin1 year ago

okay this is actually wild. was there prior context?

AI Notkilleveryoneism Memes ⏸️1 year 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

357,661 views • 1 year 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

104,537 views • 1 month ago

American attorney has found internal memos from the chemical company Monsanto discussing how they secretly dumped toxic chemicals in soil used to build an elementary school “So all the kids have cancer now, there's all these hotspots, kids with brain cancer, kids with leukemia dying — And I'm uncovering internal memos from Monsanto, where they're talking about it — How they know it's gonna be a huge thing and they know it's gonna be a problem”

American attorney has found internal memos from the chemical company Monsanto discussing how they secretly dumped toxic chemicals in soil used to build an elementary school “So all the kids have cancer now, there's all these hotspots, kids with brain cancer, kids with leukemia dying — And I'm uncovering internal memos from Monsanto, where they're talking about it — How they know it's gonna be a huge thing and they know it's gonna be a problem”

Wall Street Apes

83,983 views • 10 months ago

#KayaduLohar opens up & goes emotional on the recent comments about her: "It's been hitting me so hard. People tell its easy to come out of it, but its not actually easy to bounce back. Recently its been hitting me a lot and I'm not ok with it 💔"

#KayaduLohar opens up & goes emotional on the recent comments about her: "It's been hitting me so hard. People tell its easy to come out of it, but its not actually easy to bounce back. Recently its been hitting me a lot and I'm not ok with it 💔"

VCD

370,974 views • 7 months ago

No one talks about the internal work - the hypnotherapy and meditation needed to heal the trauma - because a traumatized population is easy to lead into a Smart City concentration camp. You’ll have an Elon Musk brain chip, a Wells Fargo nanotech banking app, and a coin with Drumpf’s face on it, and you won’t be allowed to leave without permission. #Meditation #Spirituality

No one talks about the internal work - the hypnotherapy and meditation needed to heal the trauma - because a traumatized population is easy to lead into a Smart City concentration camp. You’ll have an Elon Musk brain chip, a Wells Fargo nanotech banking app, and a coin with Drumpf’s face on it, and you won’t be allowed to leave without permission. #Meditation #Spirituality

The Dollar Vigilante

19,974 views • 2 months ago

Josh Barnett reveals his reasoning for ending Bloodsport: “I wasn’t trying to make a big deal out of it. I didn’t want it to come off as a gimmick. There were a lot of conflicting feelings within my own heart about whether this was going to be the case or not. When all is said and done, I feel like I absolutely made the right choice here. It wasn’t for anything negative. There wasn’t a tragedy, and there wasn’t some overarching bad issue or problem in that sense. For me, Bloodsport had an evolution of where it was trying to get to, and we’re just not there—we’re not close. For me, it was better to end it now than to keep going and, quite possibly, diminish what the product is.”

Josh Barnett reveals his reasoning for ending Bloodsport: “I wasn’t trying to make a big deal out of it. I didn’t want it to come off as a gimmick. There were a lot of conflicting feelings within my own heart about whether this was going to be the case or not. When all is said and done, I feel like I absolutely made the right choice here. It wasn’t for anything negative. There wasn’t a tragedy, and there wasn’t some overarching bad issue or problem in that sense. For me, Bloodsport had an evolution of where it was trying to get to, and we’re just not there—we’re not close. For me, it was better to end it now than to keep going and, quite possibly, diminish what the product is.”

Ariel Helwani

254,989 views • 2 months ago

The DeepSeek-R1 paper is a gem! Highly encourage everyone to read it. It's clear that LLM reasoning capabilities can be learned in different ways. RL, if applied correctly and at scale, can lead to some really powerful and interesting scaling and emergent properties. There is more to RL than meets the eye! Here is my breakdown of the paper along with a few tests: The multi-state training might not make sense initially but they provide clues on optimizations that we can continue to tap into. Data quality is still very important for enhancing the usability of the LLM. Unlike other reasoning LLMs, DeepSeek-R1's training recipe and weights are open so we can build on top of it. This opens up exciting research opportunities. About the attached clip: the previous preview model wasn't able to solve this task. DeepSeek-R1 can solve this and many other tasks that o1 can solve. It's a very good model for coding and math.

The DeepSeek-R1 paper is a gem! Highly encourage everyone to read it. It's clear that LLM reasoning capabilities can be learned in different ways. RL, if applied correctly and at scale, can lead to some really powerful and interesting scaling and emergent properties. There is more to RL than meets the eye! Here is my breakdown of the paper along with a few tests: The multi-state training might not make sense initially but they provide clues on optimizations that we can continue to tap into. Data quality is still very important for enhancing the usability of the LLM. Unlike other reasoning LLMs, DeepSeek-R1's training recipe and weights are open so we can build on top of it. This opens up exciting research opportunities. About the attached clip: the previous preview model wasn't able to solve this task. DeepSeek-R1 can solve this and many other tasks that o1 can solve. It's a very good model for coding and math.

elvis

140,692 views • 1 year ago

Lemov peeled back the curtain on DARPA—the government’s controversial defense research agency—and its role in shaping the digital world we now live in. DARPA, she revealed, wasn’t just involved in building the internet. It may have laid the groundwork for emotional manipulation on a global scale. “DARPA was involved in the development of the internet and of things like pattern recognition,” she said. “The government has funded many, many studies.” What worries Lemov isn’t just the tech itself—but how it’s being used. “What I got interested in, in social media and how I connect it with the episodes of brainwashing—it creates states of emotional contagion that aren’t really about convincing people of a different way to think,” she explained. She continued: “But more about how you feel about what you think.” That emotional shift, she added, is straight out of the cult playbook. “It’s not that it changed my thoughts,” she said. “It’s how I felt about those thoughts.”

Lemov peeled back the curtain on DARPA—the government’s controversial defense research agency—and its role in shaping the digital world we now live in. DARPA, she revealed, wasn’t just involved in building the internet. It may have laid the groundwork for emotional manipulation on a global scale. “DARPA was involved in the development of the internet and of things like pattern recognition,” she said. “The government has funded many, many studies.” What worries Lemov isn’t just the tech itself—but how it’s being used. “What I got interested in, in social media and how I connect it with the episodes of brainwashing—it creates states of emotional contagion that aren’t really about convincing people of a different way to think,” she explained. She continued: “But more about how you feel about what you think.” That emotional shift, she added, is straight out of the cult playbook. “It’s not that it changed my thoughts,” she said. “It’s how I felt about those thoughts.”

The Vigilant Fox 🦊

918,214 views • 1 year ago

Reasoning is what really changed the game with FSD V14, and as they implement more of it, it’s going to be game over. Here’s a great example. FSD exits the parking lot of this coffee shop. If you zoom into the nav at the top right corner, you can see that it’s telling me to turn left (illegal to do so here). This has been a map error at this shop for a while. On all previous FSD versions, I’d disengage here as FSD would try to make the illegal left. With V14.2 reasoning, the car gathers context, understands it isn’t supposed to turn left here/ignores nav and triggers the re-route.

Reasoning is what really changed the game with FSD V14, and as they implement more of it, it’s going to be game over. Here’s a great example. FSD exits the parking lot of this coffee shop. If you zoom into the nav at the top right corner, you can see that it’s telling me to turn left (illegal to do so here). This has been a map error at this shop for a while. On all previous FSD versions, I’d disengage here as FSD would try to make the illegal left. With V14.2 reasoning, the car gathers context, understands it isn’t supposed to turn left here/ignores nav and triggers the re-route.

Mike P

46,221 views • 4 months ago

This woman is having a huge dispute with her neighbor, so bad that she had to have her trespassed. The neighbor is under the idea that she owns part of this woman’s property and even hired a survey service to survey the property lines. But when the surveyor came out, the neighbors personal assistant came with her too as a means to spy and attempt to influence the outcome of this survey. The woman was upset with the surveyor saying the assistant can’t be there and something tells me the assistant knew she wasn’t supposed to be there because she was refusing to give her name. Was the surveyor just being nice to being her along or should the woman report him too for being biased?

This woman is having a huge dispute with her neighbor, so bad that she had to have her trespassed. The neighbor is under the idea that she owns part of this woman’s property and even hired a survey service to survey the property lines. But when the surveyor came out, the neighbors personal assistant came with her too as a means to spy and attempt to influence the outcome of this survey. The woman was upset with the surveyor saying the assistant can’t be there and something tells me the assistant knew she wasn’t supposed to be there because she was refusing to give her name. Was the surveyor just being nice to being her along or should the woman report him too for being biased?

SonnyBoy🇺🇸

752,166 views • 3 months ago

One of the interesting things about being on-chain is how easy it is for anyone to trace your activity. For casual users it may not matter much, but for serious traders or builders it quickly becomes a concern. Anyone can follow your wallet activity, trading patterns, and transaction history. As the space grows, tools that allow users to interact with DeFi without exposing everything will matter a lot more. So, in the next few months, I'm going to be educating on how Houdini Swap approaches that. Follow me & turn on notifications. It's about to be a long ride from here!

One of the interesting things about being on-chain is how easy it is for anyone to trace your activity. For casual users it may not matter much, but for serious traders or builders it quickly becomes a concern. Anyone can follow your wallet activity, trading patterns, and transaction history. As the space grows, tools that allow users to interact with DeFi without exposing everything will matter a lot more. So, in the next few months, I'm going to be educating on how Houdini Swap approaches that. Follow me & turn on notifications. It's about to be a long ride from here!

Sir Khaycee

16,430 views • 4 months ago

President Donald Trump’s decision to jointly bomb Iran on Feb. 28, despite internal warnings of retaliation and without a clear rationale, surprised the American public and prompted some lawmakers to question his fitness for office. In the past month, several senators have called for invoking the 25th Amendment, which allows the government to declare a president unable to carry out the duties of the office. Here’s what to know about the 25th Amendment and the latest calls to invoke it against Trump, in one minute. Read more here:

President Donald Trump’s decision to jointly bomb Iran on Feb. 28, despite internal warnings of retaliation and without a clear rationale, surprised the American public and prompted some lawmakers to question his fitness for office. In the past month, several senators have called for invoking the 25th Amendment, which allows the government to declare a president unable to carry out the duties of the office. Here’s what to know about the 25th Amendment and the latest calls to invoke it against Trump, in one minute. Read more here:

TIME

71,017 views • 2 months ago

I wrote this song over a year ago when I was first struggling to process the whirlwind of SUCCESS. A year later and i’m still struggling. So much to learn & have so many thoughts…but for now i’ll let the song speak for itself. Feels like the right time to share it with y’all.

I wrote this song over a year ago when I was first struggling to process the whirlwind of SUCCESS. A year later and i’m still struggling. So much to learn & have so many thoughts…but for now i’ll let the song speak for itself. Feels like the right time to share it with y’all.

I AM FOR JESUS

32,945 views • 9 months ago

🐨 next spring our group album is really planned to be released 🐥 our fans.. i'm going to look this way while i'm talking 🐱 just look at me and talk 🐥 our fans might have thought that we would release something as soon as we discharge from the army but in the army.. 🐰we can't.. 🐥 we can’t work on it so we have to start preparing from now on. so we estimate that it will take some time 🐹 even for my solo, it took a couple months for it to be complete after i came out 🐱 since it's not by ourselves, and its a group album, theres going to be a lot of thoughts from the group going into it 🐥 especially since we're trying to do it (incorporate group ideas) even more so this album 🐱 so i think we will need some time

🐨 next spring our group album is really planned to be released 🐥 our fans.. i'm going to look this way while i'm talking 🐱 just look at me and talk 🐥 our fans might have thought that we would release something as soon as we discharge from the army but in the army.. 🐰we can't.. 🐥 we can’t work on it so we have to start preparing from now on. so we estimate that it will take some time 🐹 even for my solo, it took a couple months for it to be complete after i came out 🐱 since it's not by ourselves, and its a group album, theres going to be a lot of thoughts from the group going into it 🐥 especially since we're trying to do it (incorporate group ideas) even more so this album 🐱 so i think we will need some time

Min Suga HQ⁷

12,977 views • 1 year ago