Alignment Newsletter #111: The Circuits hypotheses for deep learning

FromAlignment Newsletter Podcast

Start listening View podcast show

Alignment Newsletter #111: The Circuits hypotheses for deep learning

FromAlignment Newsletter Podcast

ratings:

Length:

18 minutes

Released:

Aug 5, 2020

Format:

Podcast episode

Description

Recorded by Robert Miles More information about the newsletter here

Released:

Aug 5, 2020

Format:

Podcast episode

Titles in the series (100)

The Alignment Newsletter is a weekly publication with recent content relevant to AI alignment. This podcast is an audio version, recorded by Robert Miles (http://robertskmiles.com) More information about the newsletter at: https://rohinshah.com/alignment-newsletter/

Skip carousel

Alignment Newsletter #80: Why AI risk might be solved without additional intervention from longtermists: Why AI risk might be solved without additional intervention from longtermists by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #83: Sample efficient deep learning with ReMixMatch: Sample efficient deep learning with ReMixMatch by Alignment Newsletter Podcast
21 min listen
Alignment Newsletter #76: How dataset size affects robustness, and benchmarking safe exploration by measuring constraint violations: How dataset size affects robustness, and benchmarking safe exploration by measuring constraint violations by Alignment Newsletter Podcast
19 min listen
Alignment Newsletter #86: Improving debate and factored cognition through human experiments: Improving debate and factored cognition through human experiments by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #74: Separating beneficial AI into competence, alignment, and coping with impacts: Separating beneficial AI into competence, alignment, and coping with impacts by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #90: How search landscapes can contain self-reinforcing feedback loops: How search landscapes can contain self-reinforcing feedback loops by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #82: How OpenAI Five distributed their training computation: How OpenAI Five distributed their training computation by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #97: Are there historical examples of large, robust discontinuities?: Are there historical examples of large, robust discontinuities? by Alignment Newsletter Podcast
19 min listen
Alignment Newsletter #94: AI alignment as translation between humans and machines: AI alignment as translation between humans and machines by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #79: Recursive reward modeling as an alignment technique integrated with deep RL: Recursive reward modeling as an alignment technique integrated with deep RL by Alignment Newsletter Podcast
22 min listen
Alignment Newsletter #84: Reviewing AI alignment work in 2018-19: Reviewing AI alignment work in 2018-19 by Alignment Newsletter Podcast
11 min listen
Alignment Newsletter #78: Formalizing power and instrumental convergence, and the end-of-year AI safety charity comparison: Formalizing power and instrumental convergence, and the end-of-year AI safety charity comparison by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #120: Tracing the intellectual roots of AI and AI alignment: Tracing the intellectual roots of AI and AI alignment by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #102: Meta learning by GPT-3, and a list of full proposals for AI alignment: Meta learning by GPT-3, and a list of full proposals for AI alignment by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #81: Universality as a potential solution to conceptual difficulties in intent alignment: Universality as a potential solution to conceptual difficulties in intent alignment by Alignment Newsletter Podcast
20 min listen
Alignment Newsletter #77: Double descent: a unification of statistical theory and modern ML practice: Double descent: a unification of statistical theory and modern ML practice by Alignment Newsletter Podcast
25 min listen
Alignment Newsletter #75: Solving Atari and Go with learned game models, and thoughts from a MIRI employee: Solving Atari and Go with learned game models, and thoughts from a MIRI employee by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #92: Learning good representations with contrastive predictive coding: Learning good representations with contrastive predictive coding by Alignment Newsletter Podcast
19 min listen
Alignment Newsletter #85: The normative questions we should be asking for AI alignment, and a surprisingly good chatbot: The normative questions we should be asking for AI alignment, and a surprisingly good chatbot by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #87: What might happen as deep learning scales even further?: What might happen as deep learning scales even further? by Alignment Newsletter Podcast
9 min listen
Alignment Newsletter #89: A unifying formalism for preference learning algorithms: A unifying formalism for preference learning algorithms by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #91: Concepts, implementations, problems, and a benchmark for impact measurement: Concepts, implementations, problems, and a benchmark for impact measurement by Alignment Newsletter Podcast
25 min listen
Alignment Newsletter #88: How the principal-agent literature relates to AI risk: How the principal-agent literature relates to AI risk by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #123: Inferring what is valuable in order to align recommender systems: Inferring what is valuable in order to align recommender systems by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #96: Buck and I discuss/argue about AI Alignment: Buck and I discuss/argue about AI Alignment by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #103: ARCHES: an agenda for existential safety, and combining natural language with deep RL: ARCHES: an agenda for existential safety, and combining natural language with deep RL by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #99: Doubling times for the efficiency of AI algorithms: Doubling times for the efficiency of AI algorithms by Alignment Newsletter Podcast
20 min listen
Alignment Newsletter #118: Risks, solutions, and prioritization in a world with many AI systems: Risks, solutions, and prioritization in a world with many AI systems by Alignment Newsletter Podcast
20 min listen
Alignment Newsletter #113: Checking the ethical intuitions of large language models: Checking the ethical intuitions of large language models by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #101: Why we should rigorously measure and forecast AI progress: Why we should rigorously measure and forecast AI progress by Alignment Newsletter Podcast
20 min listen

More Episodes from Alignment Newsletter Podcast

Skip carousel

Alignment Newsletter #173: Recent language model results from DeepMind: Recent language model results from DeepMind by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #172: Sorry for the long hiatus!: Sorry for the long hiatus! by Alignment Newsletter Podcast
6 min listen
Alignment Newsletter #171: Disagreements between alignment "optimists" and "pessimists": Recorded by Robert Miles: More information about the newsletter here: YouTube Channel: HIGHLIGHTS (Richard Ngo and Eliezer Yudkowsky) (summarized by Rohin): Eliezer is known for being pessimistic about our chances of... by Alignment Newsletter Podcast
14 min listen
Alignment Newsletter #170: Analyzing the argument for risk from power-seeking AI: Analyzing the argument for risk from power-seeking AI by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #169: Collaborating with humans without human data: Collaborating with humans without human data by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #168: Four technical topics for which Open Phil is soliciting grant proposals: Four technical topics for which Open Phil is soliciting grant proposals by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #167: Concrete ML safety problems and their relevance to x-risk: Concrete ML safety problems and their relevance to x-risk by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #166: Is it crazy to claim we're in the most important century?: Is it crazy to claim we're in the most important century? by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #165: When large models are more likely to lie: When large models are more likely to lie by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #164: How well can language models write code?: How well can language models write code? by Alignment Newsletter Podcast
19 min listen
Alignment Newsletter #163: Using finite factored sets for causal and temporal inference: Using finite factored sets for causal and temporal inference by Alignment Newsletter Podcast
19 min listen
Alignment Newsletter #162: Foundation models: a paradigm shift within AI: Foundation models: a paradigm shift within AI by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #161: Creating generalizable reward functions for multiple tasks by learning a model of functional similarity: Creating generalizable reward functions for multiple tasks by learning a model of functional similarity by Alignment Newsletter Podcast
18 min listen
Alignment Newsletter #160: Building AIs that learn and think like people: Building AIs that learn and think like people by Alignment Newsletter Podcast
17 min listen
Alignment Newsletter #159: Building agents that know how to experiment, by training on procedurally generated games: Building agents that know how to experiment, by training on procedurally generated games by Alignment Newsletter Podcast
27 min listen
Alignment Newsletter #158: Should we be optimistic about generalization?: Should we be optimistic about generalization? by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #157: Measuring misalignment in the technology underlying Copilot: Measuring misalignment in the technology underlying Copilot by Alignment Newsletter Podcast
14 min listen
Alignment Newsletter #156: The scaling hypothesis: a plan for building AGI: The scaling hypothesis: a plan for building AGI by Alignment Newsletter Podcast
14 min listen
Alignment Newsletter #155: A Minecraft benchmark for algorithms that learn without reward functions: A Minecraft benchmark for algorithms that learn without reward functions by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #154: What economic growth theory has to say about transformative AI: What economic growth theory has to say about transformative AI by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #153: Experiments that demonstrate failures of objective robustness: Experiments that demonstrate failures of objective robustness by Alignment Newsletter Podcast
16 min listen
Alignment Newsletter #152: How we’ve overestimated few-shot learning capabilities: How we’ve overestimated few-shot learning capabilities by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #151: How sparsity in the final layer makes a neural net debuggable: How sparsity in the final layer makes a neural net debuggable by Alignment Newsletter Podcast
11 min listen
Alignment Newsletter #150: The subtypes of Cooperative AI research: The subtypes of Cooperative AI research by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #149: The newsletter's editorial policy: The newsletter's editorial policy by Alignment Newsletter Podcast
14 min listen
Alignment Newsletter #148: Analyzing generalization across more axes than just accuracy or loss: Analyzing generalization across more axes than just accuracy or loss by Alignment Newsletter Podcast
22 min listen
Alignment Newsletter #147: An overview of the interpretability landscape: An overview of the interpretability landscape by Alignment Newsletter Podcast
13 min listen
Alignment Newsletter #146: Plausible stories of how we might fail to avert an existential catastrophe: Plausible stories of how we might fail to avert an existential catastrophe by Alignment Newsletter Podcast
15 min listen
Alignment Newsletter #145: Our three year anniversary!: Our three year anniversary! by Alignment Newsletter Podcast
14 min listen
Alignment Newsletter #144: How language models can also be finetuned for non-language tasks: How language models can also be finetuned for non-language tasks by Alignment Newsletter Podcast
13 min listen

Discover this podcast and so much more

Alignment Newsletter #111: The Circuits hypotheses for deep learning

Alignment Newsletter #111: The Circuits hypotheses for deep learning

Description

Titles in the series (100)

More Episodes from Alignment Newsletter Podcast