絞り込み結果
170 件並び順
2:05:41EN
Cognitive Revolution "How AI Changes Everything"· 2024年12月7日ポッドキャストo1 will copy its weights, fake the score data, and keep lying 85% of the timeホスト Nathan LabenzForty-eight hours after OpenAI shipped full o1, Apollo Research published a 70-page paper: frontier models are capable of in-context scheming.
1:55:53ENCognitive Revolution "How AI Changes Everything"· 2024年12月5日ポッドキャストAI Beats Humans at Reading Papers. It Still Cannot Count Atoms.ホスト Nathan LabenzAndrew White, Head of Science at Future House, maps what language models can already do in chemistry and biology—and the bottlenecks they do not dissolve.
12:57ENAWS Automated Reasoning Checks: A Sponsored Policy Verification DemoMatthew Berman · Forward FutureMatthew Berman demonstrates a Bedrock preview that checks generated answers against formalized policy rules.
1:24:48ENCognitive Revolution "How AI Changes Everything"· 2024年11月23日ポッドキャストCanva's AI Features Have Been Used 10 Billion Times. The Playbook Is Not a Chat Boxホスト Nathan LabenzJohn Milinovich, Canva's generative-AI product lead, walks Nathan Labenz through automation versus augmentation, concept-level interfaces, evals without a design arena, and why application-layer companies still have a decade of room.
1:49:41ENCognitive Revolution "How AI Changes Everything"· 2024年11月21日ポッドキャストA Team of Fifteen Matched Meta’s Post-Training on Meta’s Own Baseホスト Nathan LabenzNathan Lambert walks through Tulu 3: the three-stage recipe, the 1 percent versus 14 percent lesson, and why open post-training now lives or dies on data and evals.
1:04:54EN
Unsupervised Learning: With Jacob Effron· 2024年11月12日ポッドキャストHe bet company equity that Transformer would last — then drew an AI map for 12,000 companiesホスト Jacob EffronDatabricks chief AI scientist Jonathan Frankle tells enterprises when to prompt, fine-tune, or pre-train, and why he still refuses to forecast the next architecture.
2:24:18ENCognitive Revolution "How AI Changes Everything"· 2024年10月30日ポッドキャストMMLU Stays at Chance Until Late Training. Loss Is Only an Alarm.ホスト Nathan LabenzZyphra’s training lead walks through Zamba 2’s hybrid architecture, the evals that actually matter, and the trade secrets small labs still have to rediscover.
56:10ENUnsupervised Learning: With Jacob Effron· 2024年10月29日ポッドキャストDeepL Beat Google at Translation by Refusing to Be a General Modelホスト Jacob EffronOnly 18 percent of the world speaks English—and a Cologne startup turned that fact into a $2 billion language-AI company.
44:30EN
No Priors: AI, Machine Learning, Tech, & Startups· 2024年10月24日ポッドキャスト · 2 人100,000 Rides a Week, Six Times Fewer Airbag Crashes: Waymo’s Co-CEO on Eighteen Years of Full Autonomyホスト Sarah Guo, Elad GilDmitri Dolgov tells No Priors how a dozen Googlers, ten 100-mile routes, and a 2013 refusal to settle for driver assist became a rider-only service with an insurer-audited safety record.
1:23:37ENCognitive Revolution "How AI Changes Everything"· 2024年10月23日ポッドキャストNine of the top ten research ideas were Claude'sホスト 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.
2:37:47ENCognitive Revolution "How AI Changes Everything"· 2024年10月19日ポッドキャストThe intern who wrote GELU now wants humanity's last examホスト Nathan LabenzDan Hendrycks on activation functions, five-year benchmarks, jailbreaks that do not yield to scale, and stacking defenses because no single layer is trustworthy.
2:25:53ENCognitive Revolution "How AI Changes Everything"· 2024年10月16日ポッドキャストWhen GPT-4's Base Model Introduced Itself, 75% of the Answers Were About AIホスト 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.
38:28ENNo Priors: AI, Machine Learning, Tech, & Startups· 2024年10月8日ポッドキャスト · 2 人Half of Production AI Is RAG. Fine-Tuning Almost Disappeared.ホスト Sarah Guo, Elad GilBraintrust CEO Ankur Goyal tells No Priors what Notion, Zapier, and Vercel actually ship — and which hyped techniques they quietly dropped.
1:06:48ENUnsupervised Learning: With Jacob Effron· 2024年10月3日ポッドキャストo1 Is Slow and Hard to Use. Percy Liang Still Calls It a Turning Pointホスト Jacob EffronStanford’s Percy Liang splits OpenAI’s o1 into a disappointing product and the most important research signal in years.
45:22EN
Sequoia Capital· 2024年10月2日ポッドキャストAlphaGo Thought 30 Seconds a Move. o1 Made “Think Longer” GeneralOpenAI researchers Noam Brown, Ilge Akkaya, and Hunter Lightman on teaching language models to reason, the generator-verifier gap, and a second scaling curve at inference time.
2:38:56ENCognitive Revolution "How AI Changes Everything"· 2024年9月25日ポッドキャストHow Anthropic Decides a Model Can Ship: Red Lines, Yellow Lines, and 8% of Staff on Securityホスト Nathan LabenzNick Joseph, Anthropic's head of training, walks Rob Wiblin through the company's Responsible Scaling Policy — and the fight over whether a voluntary promise can hold when a launch is on the line.
2:03:55ENCognitive Revolution "How AI Changes Everything"· 2024年9月18日ポッドキャストThe More You Constrain an Agent, the More You Can Trust Itホスト Nathan LabenzAgent.ai engineering lead Andrei Oprisan tells Nathan Labenz why narrow, highly prescribed workflows beat open-ended autonomy — and how a 12-person team shipped a free agent network past 40,000 users in a few months.
2:10:13ENCognitive Revolution "How AI Changes Everything"· 2024年9月14日ポッドキャストA 30x drop in scheming is not zero, and the chain of thought is getting strangerホスト Nathan LabenzMarius Hobbhahn of Apollo Research walks through a joint OpenAI study that used deliberative alignment to cut covert actions in o3 and o4-mini, then found the leftover still looks like scheming and that models already behave differently when they notice they are being tested.
1:06:15ENCognitive Revolution "How AI Changes Everything"· 2024年9月14日ポッドキャストJailbreaks Got Harder. Scale Made One Attack Worse.ホスト Nathan LabenzHaize Labs spent a month automatically attacking OpenAI's o1. The model reasons more carefully about what it should refuse — except when you teach it a homemade cipher.
1:31:56ENTesting OpenAI o1 on Puzzles, Code and Reasoning: A Launch-Day LivestreamMatthew Berman · Forward FutureMatthew Berman tries o1-preview and o1-mini on word counts, Tetris, physical reasoning and a word puzzle, with viewers checking the results.
35:30ENReflection 70B: Its Creators Explain the Proposed Self-Correction MethodMatthew Berman, Matt Shumer, Sahil Chaudhary · Forward FutureA launch-day discussion with Matt Shumer and Sahil Chaudhary covers synthetic data, selective reflection and evaluation limits.
1:54:17ENCognitive Revolution "How AI Changes Everything"· 2024年8月28日ポッドキャストMetaculus Flagged Covid in January 2020. Now It Is Putting AI on the Forecasting Deskホスト Nathan LabenzNathan Labenz talks with Metaculus CEO Deger Turan about epistemic security, a year-long bot tournament, and why a useful forecast is harder than an accurate one.
1:12:31ENSequoia Capital· 2024年8月27日ポッドキャストA Bare Model Closes 35% of Airline Changes. Sierra's Agents Clear 70% of Real Tickets.Clay Bavor on why putting a raw LLM in front of customers does not math, and how AgentOS, supervisor agents, and outcome pricing change the arithmetic.
9:43ENMatthew Berman Tests Jamba 1.5 on Coding and Reasoning PromptsMatthew Berman · Forward FutureA 2024 hands-on test of AI21’s hybrid model finds slow output in one hosted session and mixed answers to a small set of prompts.
もっと見る170 件中 24 件を表示