Learning GenAI via SOTA Papers

Yun Wu

This podcast is focusing on sharing the papers on GenAI related topic, especially the SOTA (State of the Art) papers that are the foundations of GenAI work. It shows how these researches paved the way to the GenAI tools that we are using every day such as ChatGPT, Gemini, Claude Code etc.

Play

Open in

32 episodes
daily
Avg 21 min
English

Episodes 32

Newest Oldest Longest

All20 Unplayed20 In progress0 Played0

Counted on this page — what you have heard stays on this device, so it is not something the list can be paged by.

S1 · E423
Today · 21 min
EP423: Why Agentic AI breaks the datacenter
Title: Architectural Implications of Agentic AI Workflows Source: http://arxiv.org/abs/2608.04458v1 Summary: This paper proposes a foundational understanding of the architectural requirements and design patterns for effective agentic AI systems. By outlining these implications, it provides crucial guidance for the development of new architectural primitives and frameworks in Agentic AI.
S1 · E422
Yesterday · 22 min
EP422: Stopping spurious signals in AI distillation
Title: When Teachers Mislead: Spurious-Signal-Aware On-Policy Distillation Source: http://arxiv.org/abs/2608.03632v1 Summary: This paper introduces a novel distillation method robust to misleading teacher signals, particularly in an on-policy context relevant to AI agents. It represents a significant efficiency and reasoning breakthrough by enabling more reliable and effective training of agentic systems.
S1 · E421
Yesterday · 24 min
EP421: Fixing AI Hallucinations with RAIL Principles
Title: The RAIL Principles for Neurosymbolic AI: Reasoning, Assurances, Interfacing and Learning Source: http://arxiv.org/abs/2608.04285v1 Summary: This paper proposes foundational principles for Neurosymbolic AI, a critical approach for developing agents with advanced reasoning capabilities and interpretability. These principles can guide the design of novel agentic reasoning loops and architectures that integrate symbolic knowledge with neural learning.
S1 · E420
Wednesday · 22 min
EP420: Hijacking AI memory via factual injection
Title: MAFIA: Query-Only Memory Attacks via Probing and Factual Injection against Audited LLM Agents Source: http://arxiv.org/abs/2608.03844v1 Summary: This research reveals critical vulnerabilities in LLM agents' memory through novel attack vectors. Understanding these memory attacks is foundational for developing more robust and secure agentic reasoning loops and frameworks, driving architectural advancements for reliable AI agents.
S1 · E419
Wednesday · 20 min
EP419: Ten Weeks of Autonomous AI Research
Title: Long-Horizon Autonomous Architecture Research with a Language-Model Agent: A Behavioural Case Study Source: http://arxiv.org/abs/2608.01995v1 Summary: This work presents a novel agentic framework enabling AI agents to conduct complex, multi-stage, autonomous research over extended periods. It establishes a foundational capability for agents to plan, execute, and self-correct across long time horizons in intellectual domains, marking a significant leap in agent autonomy.
S1 · E418
Tuesday · 20 min
EP418: DeepVoyager-VL solves the visual search bottleneck
Title: DeepVoyager-VL: Incentivizing Vision-in-the-Loop Search for Long-Horizon Multimodal Agents Source: http://arxiv.org/abs/2608.01827v1 Summary: This work presents 'Incentivizing Vision-in-the-Loop Search,' a novel reasoning framework for long-horizon multimodal agents. It offers a foundational approach for agents to autonomously explore, plan, and operate over extended durations by integrating continuous visual perception and motivational signals.
S1 · E417
Tuesday · 20 min
EP417: AI agents replace human beta testers
Title: Can AI Agents Simulate A/B Test Outcomes? A Validation Framework for Agentic Experimentation Source: http://arxiv.org/abs/2608.02345v1 Summary: This work proposes 'Agentic Experimentation,' a novel framework that empowers AI agents to simulate complex real-world processes like A/B tests and validate their outcomes. This is foundational as it enables agents to perform sophisticated scientific inquiry, predictive analysis, and strategic decision-making autonomously.
S1 · E416
Monday · 22 min
EP416: How AdaThinkV stops AI overthinking video
Title: AdaThinkV: Adaptive Thinking for Token-Efficient Video Reasoning Source: http://arxiv.org/abs/2608.01980v1 Summary: This paper introduces 'Adaptive Thinking,' which proposes a novel reasoning paradigm for AI agents. It also details 'Token-Efficient Video Reasoning,' representing a significant efficiency breakthrough crucial for scaling and deploying multimodal Generative AI and Agentic AI.
S1 · E415
Monday · 20 min
EP415: Fixing the AI granularity mismatch
Title: Where Reasoning Diverges: Localized Multi-Agent Debate Source: http://arxiv.org/abs/2608.01463v1 Summary: This work proposes a novel framework for multi-agent systems to improve their collective reasoning through localized debate mechanisms. It introduces a foundational agentic reasoning loop that allows agents to refine solutions by exploring divergent perspectives.
S1 · E414
Sunday · 21 min
EP414: Why context compaction breaks AI agents
Title: Context Compaction Theory Source: http://arxiv.org/abs/2608.01326v1 Summary: This paper likely introduces a theoretical framework for efficiently managing and utilizing context in large language models. Such a breakthrough could significantly enhance LLM scalability and reasoning by addressing current context window limitations and computational costs.
S1 · E413
Sunday · 12 min
EP413: Slashing AI latency with uncertainty repair
Title: CURE: Local Uncertainty Repair for Block-Parallel Speculative Decoding Source: http://arxiv.org/abs/2608.00531v1 Summary: This research introduces 'Local Uncertainty Repair' for 'Block-Parallel Speculative Decoding,' offering a substantial efficiency breakthrough for Large Language Model (LLM) inference. By enhancing the speed and potentially the reliability of decoding, it directly addresses a critical bottleneck in GenAI deployment, making powerful models more practical and scalable.
S1 · E412
Saturday · 23 min
EP412: Thermodynamic Computing Solves the AI Bottleneck
Title: CN101 - A Digital Thermodynamic Computer for Generative AI Source: http://arxiv.org/abs/2608.00754v1 Summary:This paper proposes a novel 'Digital Thermodynamic Computer' specifically designed for Generative AI, potentially introducing a new architectural primitive for AI computation. Such a radical shift in computing paradigms could unlock unprecedented efficiency or capabilities, leading to significant breakthroughs in how GenAI models are built and scaled.
S1 · E411
Saturday · 22 min
EP411: How NeSyFS Gives AI Fast-Slow Thinking
Title: NeSyFS: A Neuro-symbolic Fast-Slow Thinking Framework for LLM Agent under Partial Observability Source: http://arxiv.org/abs/2607.28942v1 Summary: This paper introduces a novel neuro-symbolic framework enabling LLM agents to employ fast-slow thinking, significantly improving their reasoning capabilities under partial observability. This architecture offers a foundational approach to more sophisticated and human-like agentic decision-making and planning.
S1 · E410
September 4 · 19 min
EP410: How provenance laundering brainwashes AI
Title: Memory Provenance Laundering in LLM Agents: A Non-Amplification Firewall for Persistent Memory Source: http://arxiv.org/abs/2607.29167v1 Summary: This research proposes a critical mechanism for managing and securing the persistent memory of LLM agents, introducing a "non-amplification firewall." This foundational work ensures memory integrity and prevents error propagation, crucial for the reliability and trustworthiness of long-lived, complex agentic AI systems.
S1 · E409
September 4 · 24 min
EP409: Robots That Dream Before They Move
Title: World Action Planner: Generalizable Decision-Making with Action-Conditioned World Models Source: http://arxiv.org/abs/2607.27599v1 Summary: This paper introduces a novel framework for agents to achieve generalizable decision-making by leveraging action-conditioned world models. This represents a foundational step towards more capable and autonomous AI agents through advanced planning and environmental understanding, aligning with novel agentic reasoning loops.
S1 · E408
September 3 · 17 min
EP408: AI memory reconstructed not replayed
Title: MemHarness: Memory Is Reconstructed, Not Replayed Source: http://arxiv.org/abs/2607.28272v1 Summary: This work proposes a groundbreaking paradigm where AI memory is actively reconstructed rather than merely retrieved or replayed. This offers a fundamental architectural and reasoning breakthrough for both generative AI and agents, enabling more dynamic, context-aware, and robust utilization of past experiences.
S1 · E407
September 3 · 22 min
EP407: How AI learns your teamwork capabilities
Title: Partner Capability Estimation for Task-Agnostic Adaptation in Ad-Hoc Teamwork Source: http://arxiv.org/abs/2607.27177v1 Summary: This research presents a novel framework for AI agents to perform partner capability estimation, enabling task-agnostic adaptation in ad-hoc teamwork. This represents a significant breakthrough in agentic reasoning, crucial for developing intelligent agents capable of flexible and robust collaboration in dynamic, open-ended environments.
S1 · E406
September 2 · 21 min
EP406: Ending AI Groundhog Day With Living Harness
Title: Living-Harness Is an Interactive-Agent Evolver Source: http://arxiv.org/abs/2607.26598v1 Summary: This work introduces a novel interactive-agent evolver system, providing a meta-level framework for the systematic discovery and refinement of agentic capabilities. Such an evolver is foundational for developing more advanced agentic reasoning loops and significantly enhancing the efficiency of agent design and optimization.
S1 · E405
September 2 · 22 min
EP405: Are AI Agents Just Talking to Themselves
Title: Do Latent Channels Actually Communicate? A Causal Audit of Latent Multi-Agent LLM Source: http://arxiv.org/abs/2607.26773v1 Summary: This paper fundamentally investigates the internal communication mechanisms within latent multi-agent LLM architectures through a causal audit. Discovering how these "latent channels" truly function provides critical insights for designing novel agentic reasoning frameworks and more robust multi-agent systems.
S1 · E404
September 1 · 18 min
EP404: AI agents hide betrayal in Werewolf
Title: Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems Source: http://arxiv.org/abs/2607.26120v1 Summary: This research delves into the fundamental challenges of objective misalignment and deceptive behavior within complex multi-agent systems powered by LLMs. By analyzing these critical dynamics, it contributes a novel agentic reasoning framework for understanding, predicting, and potentially mitigating emergent behaviors in multi-agent AI, which is essential for ensuring safety, reliability, and control in future AI societies.

Showing 1–20 of 32 episodes