Search Results - "Hubinger, Evan"

Uncovering Deceptive Tendencies in Language Models: A Simulated Company AI Assistant by Järviniemi, Olli, Hubinger, Evan

“…We study the tendency of AI systems to deceive by constructing a realistic simulation setting of a company AI assistant. The simulated company employees…”
Get full text

An overview of 11 proposals for building safe advanced AI by Hubinger, Evan

“…This paper analyzes and compares 11 different proposals for building safe advanced AI under the current machine learning paradigm, including major contenders…”
Get full text

Engineering Monosemanticity in Toy Models by Jermyn, Adam S, Schiefer, Nicholas, Hubinger, Evan

“…In some neural networks, individual neurons correspond to natural ``features'' in the input. Such \emph{monosemantic} neurons are of great help in…”
Get full text

Conditioning Predictive Models: Risks and Strategies by Hubinger, Evan, Jermyn, Adam, Treutlein, Johannes, Hudson, Rubi, Woolverton, Kate

“…Our intention is to provide a definitive reference on what it would take to safely make use of generative/predictive models in the absence of a solution to the…”
Get full text

Steering Llama 2 via Contrastive Activation Addition by Panickssery, Nina, Gabrieli, Nick, Schulz, Julian, Tong, Meg, Hubinger, Evan, Turner, Alexander Matt

“…We introduce Contrastive Activation Addition (CAA), an innovative method for steering language models by modifying their activations during forward passes. CAA…”
Get full text

Sabotage Evaluations for Frontier Models by Benton, Joe, Wagner, Misha, Christiansen, Eric, Anil, Cem, Perez, Ethan, Srivastav, Jai, Durmus, Esin, Ganguli, Deep, Kravec, Shauna, Shlegeris, Buck, Kaplan, Jared, Karnofsky, Holden, Hubinger, Evan, Grosse, Roger, Bowman, Samuel R, Duvenaud, David

“…Sufficiently capable models could subvert human oversight and decision-making in important contexts. For example, in the context of AI development, models…”
Get full text

Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models by Denison, Carson, MacDiarmid, Monte, Barez, Fazl, Duvenaud, David, Kravec, Shauna, Marks, Samuel, Schiefer, Nicholas, Soklaski, Ryan, Tamkin, Alex, Kaplan, Jared, Shlegeris, Buck, Bowman, Samuel R, Perez, Ethan, Hubinger, Evan

“…In reinforcement learning, specification gaming occurs when AI systems learn undesired behaviors that are highly rewarded due to misspecified training goals…”
Get full text

Studying Large Language Model Generalization with Influence Functions by Grosse, Roger, Bae, Juhan, Anil, Cem, Elhage, Nelson, Tamkin, Alex, Tajdini, Amirhossein, Steiner, Benoit, Li, Dustin, Durmus, Esin, Perez, Ethan, Hubinger, Evan, Lukošiūtė, Kamilė, Nguyen, Karina, Joseph, Nicholas, McCandlish, Sam, Kaplan, Jared, Bowman, Samuel R

“…When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of…”
Get full text

Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training by Hubinger, Evan, Denison, Carson, Mu, Jesse, Lambert, Mike, Tong, Meg, MacDiarmid, Monte, Lanham, Tamera, Ziegler, Daniel M, Maxwell, Tim, Cheng, Newton, Jermyn, Adam, Askell, Amanda, Radhakrishnan, Ansh, Anil, Cem, Duvenaud, David, Ganguli, Deep, Barez, Fazl, Clark, Jack, Ndousse, Kamal, Sachan, Kshitij, Sellitto, Michael, Sharma, Mrinank, DasSarma, Nova, Grosse, Roger, Kravec, Shauna, Bai, Yuntao, Witten, Zachary, Favaro, Marina, Brauner, Jan, Karnofsky, Holden, Christiano, Paul, Bowman, Samuel R, Graham, Logan, Kaplan, Jared, Mindermann, Sören, Greenblatt, Ryan, Shlegeris, Buck, Schiefer, Nicholas, Perez, Ethan

“…Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue…”
Get full text

Measuring Faithfulness in Chain-of-Thought Reasoning by Lanham, Tamera, Chen, Anna, Radhakrishnan, Ansh, Steiner, Benoit, Denison, Carson, Hernandez, Danny, Li, Dustin, Durmus, Esin, Hubinger, Evan, Kernion, Jackson, Lukošiūtė, Kamilė, Nguyen, Karina, Cheng, Newton, Joseph, Nicholas, Schiefer, Nicholas, Rausch, Oliver, Larson, Robin, McCandlish, Sam, Kundu, Sandipan, Kadavath, Saurav, Yang, Shannon, Henighan, Thomas, Maxwell, Timothy, Telleen-Lawton, Timothy, Hume, Tristan, Hatfield-Dodds, Zac, Kaplan, Jared, Brauner, Jan, Bowman, Samuel R, Perez, Ethan

“…Large language models (LLMs) perform better when they produce step-by-step, "Chain-of-Thought" (CoT) reasoning before answering a question, but it is unclear…”
Get full text

Question Decomposition Improves the Faithfulness of Model-Generated Reasoning by Radhakrishnan, Ansh, Nguyen, Karina, Chen, Anna, Chen, Carol, Denison, Carson, Hernandez, Danny, Durmus, Esin, Hubinger, Evan, Kernion, Jackson, Lukošiūtė, Kamilė, Cheng, Newton, Joseph, Nicholas, Schiefer, Nicholas, Rausch, Oliver, McCandlish, Sam, Showk, Sheer El, Lanham, Tamera, Maxwell, Tim, Chandrasekaran, Venkatesa, Hatfield-Dodds, Zac, Kaplan, Jared, Brauner, Jan, Bowman, Samuel R, Perez, Ethan

“…As large language models (LLMs) perform more difficult tasks, it becomes harder to verify the correctness and safety of their behavior. One approach to help…”
Get full text

Risks from Learned Optimization in Advanced Machine Learning Systems by Hubinger, Evan, van Merwijk, Chris, Mikulik, Vladimir, Skalse, Joar, Garrabrant, Scott

“…We analyze the type of learned optimization that occurs when a learned model (such as a neural network) is itself an optimizer - a situation we refer to as…”
Get full text

Discovering Language Model Behaviors with Model-Written Evaluations by Perez, Ethan, Ringer, Sam, Lukošiūtė, Kamilė, Nguyen, Karina, Chen, Edwin, Heiner, Scott, Pettit, Craig, Olsson, Catherine, Kundu, Sandipan, Kadavath, Saurav, Jones, Andy, Chen, Anna, Mann, Ben, Israel, Brian, Seethor, Bryan, McKinnon, Cameron, Olah, Christopher, Yan, Da, Amodei, Daniela, Amodei, Dario, Drain, Dawn, Li, Dustin, Tran-Johnson, Eli, Khundadze, Guro, Kernion, Jackson, Landis, James, Kerr, Jamie, Mueller, Jared, Hyun, Jeeyoon, Landau, Joshua, Ndousse, Kamal, Goldberg, Landon, Lovitt, Liane, Lucas, Martin, Sellitto, Michael, Zhang, Miranda, Kingsland, Neerav, Elhage, Nelson, Joseph, Nicholas, Mercado, Noemí, DasSarma, Nova, Rausch, Oliver, Larson, Robin, McCandlish, Sam, Johnston, Scott, Kravec, Shauna, Showk, Sheer El, Lanham, Tamera, Telleen-Lawton, Timothy, Brown, Tom, Henighan, Tom, Hume, Tristan, Bai, Yuntao, Hatfield-Dodds, Zac, Clark, Jack, Bowman, Samuel R, Askell, Amanda, Grosse, Roger, Hernandez, Danny, Ganguli, Deep, Hubinger, Evan, Schiefer, Nicholas, Kaplan, Jared

“…As language models (LMs) scale, they develop many novel behaviors, good and bad, exacerbating the need to evaluate how they behave. Prior work creates…”
Get full text