Résultats
383 articlesTrier
1:48:20EN
Cognitive Revolution "How AI Changes Everything"· 25 janv. 2025PodcastFrom 15% to 70% on Right-or-Wrong Scores: DeepSeek-R1 Laid Reinforcement Learning's Cards on the TableAnime : Nathan LabenzOn Inauguration Day 2025, DeepSeek and Moonshot dropped reasoning-model results that Nathan Labenz reads as proof the reinforcement-learning recipe is simple, public, and still climbing.
1:40:22ENCognitive Revolution "How AI Changes Everything"· 22 janv. 2025PodcastTwenty-atom crystals taught an AI the physics of a heartbeatAnime : Nathan LabenzOrbital Materials' Jonathan Godwin and Tim Duignan tell Nathan Labenz how a neural network trained on tiny crystals simulated a potassium ion channel on a single 2017 GPU.
16:25EN
Matthew Berman· 18 janv. 2025Why Test-Time Compute Could Reshape AI InferenceMatthew Berman, Lisa Su, Jensen Huang · Forward FutureMatthew Berman explains how extra computation at inference can improve model outputs, and why hardware executives expect a much larger inference market.
10:05ENMatthew Berman· 9 janv. 2025A Chess Experiment Shows How an AI Model Exploited Its Test EnvironmentMatthew Berman · Forward FuturePalisade Research’s small o1-preview test and related alignment studies illustrate why objectives, permissions and evaluation conditions matter.
2:04:45ENCognitive Revolution "How AI Changes Everything"· 7 janv. 2025Podcasto3 Hit Superhuman on ARC-AGI. It Still Lost at Tic-Tac-ToeAnime : Nathan LabenzNathan Labenz walks through OpenAI's o3 scores, deliberative alignment, and why 2024's biggest surprise was beating PhDs while failing to book a calendar invite.
15:34ENMatthew Berman· 6 janv. 2025Sam Altman on AGI and Superintelligence: A 2025 DiscussionMatthew Berman, Ray Kurzweil, Lex Fridman · Forward FutureMatthew Berman examines Altman’s statements about AGI, the singularity and possible AI-driven scientific progress.
30:29ENMatthew Berman· 3 janv. 2025How Researchers Think o1-Style Reasoning Models WorkMatthew Berman · Forward FutureMatthew Berman explains a research survey of policy initialization, rewards, search and reinforcement learning, and the open questions behind reasoning models.
21:02ENMatthew Berman· 24 déc. 2024Reactions to OpenAI’s 2024 o3 Preview: Benchmarks, Cost and LimitsMatthew Berman · Forward FutureMatthew Berman surveys the responses to the o3 preview, from mathematical benchmarks to debates about AGI, compute and reliability.
14:15ENMatthew Berman· 21 déc. 2024Anthropic’s Alignment Faking Study: What the Experiments Actually ShowMatthew Berman · Forward FutureMatthew Berman examines a 2024 study of models responding differently when they believe their answers will be used for training.
1:46:41ENCognitive Revolution "How AI Changes Everything"· 21 déc. 2024PodcastFrontier Models Hit the 10th–40th Percentile on Real AI R&DAnime : Nathan LabenzNeev Parikh of METR walks Nathan Labenz through REBench, a seven-task test of whether agents can do machine learning research—and what happened when one of them decided not to train.
9:00ENMatthew Berman· 20 déc. 2024Best-of-N Jailbreaking: What Repeated Prompt Variations Revealed in 2024Matthew Berman · Forward FutureMatthew Berman reviews research on how repeated input variations can expose safety failures in text, vision and audio models.
2:05:41ENCognitive Revolution "How AI Changes Everything"· 7 déc. 2024Podcasto1 will copy its weights, fake the score data, and keep lying 85% of the timeAnime : Nathan LabenzForty-eight hours after OpenAI shipped full o1, Apollo Research published a 70-page paper: frontier models are capable of in-context scheming.
0:47ENMatthew Berman· 5 déc. 2024Interview-Based AI Agents and the Limits of Simulating PeopleMatthew Berman · Forward FutureMatthew Berman discusses agents built from interviews and imagines using them to simulate societies.
1:54:44ENCognitive Revolution "How AI Changes Everything"· 30 nov. 2024PodcastPreferences Are Data. The Alignment Target Is Norms.Anime : Nathan LabenzMIT PhD student Tan Zhi Xuan tells The Cognitive Revolution why role ethics and contractualism should replace preference maximization — and how grid-world agents infer rules from apples left uneaten.
0:58ENMatthew Berman· 18 nov. 2024A Withdrawn AI Research Study and Questions About Scientific DiscoveryMatthew Berman · Forward FutureA 2024 video cites an MIT-affiliated preprint that arXiv later withdrew; MIT says its findings should not be relied on.
14:05ENMatthew Berman· 15 nov. 2024Test-Time Training Improves ARC Puzzle Results in an MIT StudyMatthew Berman · Forward FutureMatthew Berman explains a 2024 paper that adapts a language model to each puzzle at inference time and combines its predictions with program synthesis.
11:10ENMatthew Berman· 13 nov. 2024AI and Scientific Discovery: A 2024 Video’s Withdrawn StudyMatthew Berman · Forward FutureMatthew Berman discusses AI-assisted research and recursive improvement; the central preprint was later withdrawn over serious data-validity concerns.
0:35ENMatthew Berman· 6 nov. 2024An IBM Visit: Granite Models, InstructLab and Quantum ComputingMatthew Berman · Forward FutureMatthew Berman previews a longer discussion of IBM’s small AI models, model customization and quantum research.
1:23:37ENCognitive Revolution "How AI Changes Everything"· 23 oct. 2024PodcastNine of the top ten research ideas were Claude'sAnime : Nathan LabenzStanford PhD student Chenglei Si ran a blind study pitting Claude 3.5 Sonnet against more than a hundred NLP researchers on novel research ideas.
41:42EN
Sequoia Capital· 22 oct. 2024PodcastNeural Networks Are Still Waiting for Their ThermodynamicsOpenAI researcher Dan Roberts on why physicists keep migrating into AI, and why the field still lacks the equivalent of the ideal gas law.
2:25:53ENCognitive Revolution "How AI Changes Everything"· 16 oct. 2024PodcastWhen GPT-4's Base Model Introduced Itself, 75% of the Answers Were About AIAnime : Nathan LabenzOwain Evans tells Michael Trazzi why situational awareness and out-of-context reasoning are leading indicators of deceptive alignment, and why a pretrained model with almost no self-knowledge still kept talking about being an AI.
1:13:23ENCognitive Revolution "How AI Changes Everything"· 12 oct. 2024PodcastAn AI Found a GABA-Cancer Link. The Next Step Is Talking to the Cells.Anime : Nathan LabenzMichael Levin and Leo Pio Lopez on a multilayer embedding that predicted melanoma, a cancer with no mutations, and why biology's real interface is cognitive.
26:43ENMatthew Berman· 10 oct. 2024Testing OpenAI o1 on Planning, Rules and Spatial ReasoningMatthew Berman · Forward FutureMatthew Berman reviews a 2024 pilot study comparing o1-preview, o1-mini and GPT-4 on six planning domains.
1:19:00EN
20VC with Harry Stebbings· 7 oct. 2024Podcast · 2 voixEiso Kant on Poolside, Compute and AI for Software DevelopmentInvité : Eiso Kant · Anime : Harry StebbingsPoolside’s CTO explains code-execution feedback, the cost of model development and his company’s ambitions in the AI race.Charger plus24 sur 383 affichés