AI Papers of the Week
Every paper worth reading in AI, hand-picked one week at a time.
Beyond Individual Intelligence
A multi-agent systems survey covering 200+ papers, organized along three axes: collaboration mechanisms, failure attribution, and self-evolution. Each axis is treated as a distinct research line. The self-evolution chapter maps how memory, meta-learning, and procedure-editing approaches intersect.
AutoTTS
AutoTTS reframes test-time scaling as a search problem. Instead of designing branching, pruning, and stopping heuristics directly, the user constructs a discovery environment in which TTS strategies are searched automatically. Width-depth TTS is recast as controller synthesis over pre-collected reasoning trajectories and probe signals, so candidate controllers can be evaluated without repeated LLM calls.
AI Co-Mathematician
Google DeepMind presents AI Co-Mathematician, an agentic research workbench for mathematicians. The system is an asynchronous, stateful environment that supports ideation, literature discovery, computational analysis, theorem verification, and knowledge development across long sessions. It reaches 48% on FrontierMath Tier 4, a new high among AI systems evaluated.
AEvo
AEvo separates the iterative self-improvement loop into two roles: a candidate-proposer that generates the next attempt, and a meta-agent that observes traces and edits the procedure used to propose future candidates. Past runs (candidates, feedback, traces, failures) function as memory the meta-agent reads from when revising the procedure. AEvo reports a 26% relative gain over the strongest evolution baseline on agentic and reasoning benchmarks, and SOTA on three open-ended optimization tasks under the same iteration budget. The work demonstrates one way to operationalize accumulated agentic search logs as input to procedure-level updates rather than discarding them after each run.
The Memory Curse in LLM Agents
A study of how long histories affect LLM agent behavior. Across 7 LLMs and 4 social dilemma games over 500 rounds, expanding accessible history degraded cooperation in 18 of 28 model-game combinations. Lexical analysis of 378,000 reasoning traces shows the mechanism is erosion of forward-looking intent rather than increased suspicion: long histories pull the model toward reasoning about past interactions rather than future payoffs. A LoRA adapter trained only on forward-looking traces mitigates the decay and transfers zero-shot to new games. Memory sanitization, which keeps prompt length fixed but swaps in synthetic cooperative records, restores cooperation, indicating the trigger is content rather than length. Ablating explicit chain-of-thought often reduces the collapse, suggesting deliberation amplifies the effect. The paper provides a diagnostic plus interventions for long-running agent systems where history quality, not just history length, drives behavior.
Token Superposition Training
Nous Research's second pretraining paper of the week. Token Superposition Training (TST) is a modification to the standard LLM pretraining loop that produces a 2 to 3x wall-clock speedup at matched FLOPs without changing the model architecture, optimizer, tokenizer, or training data. During the first third of training, the model reads and predicts contiguous bags of tokens, averaging their embeddings on the input side and predicting the next bag with a modified cross-entropy on the output side. For the remainder of the run, training reverts to standard next-token prediction. The inference-time model is identical to one produced by conventional pretraining. TST was validated at 270M, 600M, and 3B dense scales, and at a 10B-A1B mixture-of-experts model where it reaches a lower final loss while consuming 4,768 B200-GPU-hours versus the baseline's 12,311. Together with Lighthouse Attention, this is the second pretraining-loop modification from the same lab this week reporting substantial speedups without architecture changes.
HeavySkill
One of the cleaner takes on agentic harness design released this year. The paper argues that what actually drives harness performance is not the orchestration code, but a single inner skill: parallel reasoning followed by deliberation. Internalize that pattern into the model and most of the surrounding scaffolding becomes optional. HeavySkill systematizes the idea as a two-stage pipeline you can run beneath any harness, then trains it as a learnable skill via RLVR. The result is a harness win that looks more like a model win.
Conductor
Sakana AI's ICLR 2026 paper introduces a 7B Conductor model that hits SOTA on GPQA-Diamond and LiveCodeBench by orchestrating other LLMs instead of solving problems itself. The Conductor is trained with RL to do two things simultaneously: design communication topologies between worker agents (open or closed source) and prompt-engineer focused instructions to each worker so it leverages individual strengths. The orchestrator becomes a learnable policy, not a wrapper around one.
Self-Improving Pretraining
Most LLM safety, factuality, and reasoning fixes get bolted on at post-training. By then the patterns have already set. This Meta FAIR paper moves those behaviors into pretraining itself. The team uses a strong post-trained model as both a rewriter and a judge: it rewrites pretraining suffixes toward higher-quality, safer continuations, then scores model rollouts against the original suffix and the rewrite to drive RL during pretraining. Instead of next-token prediction, the policy learns sequence generation from the start, with rewards for quality, safety, and factuality.
Connect Four AlphaZero from Scratch
This paper proposes a new way to evaluate coding agents: hand them a minimal task description, give them a tight budget, and ask them to autonomously rebuild a famous ML breakthrough end-to-end. Connect Four plus AlphaZero is the first instance. It is small enough to run on a laptop and hard enough to require a real research engineering loop. Claude Opus 4.7 implemented the full pipeline (MCTS, neural value and policy nets, self-play, training schedule) in three hours on consumer hardware, then beat the Pascal Pons solver 7 of 8 as first-mover. No other frontier coding agent tested cleared 2 of 8.
Coordination as Architecture
Multi-agent LLM systems fail in production at rates between 41% and 87%, and the majority of those failures are coordination defects, not base-model capability. Most published comparisons of multi-agent architectures cannot even tell you whether the gain came from coordination or from one configuration just having more context. This paper argues coordination should be treated as a configurable architectural layer, separable from agent logic and information access, then backs the position with an information-controlled experiment.
Horizon Generalization
Microsoft Research runs a controlled study where the only variable is task horizon length. Same decision rules, same reasoning structure, different sequence length to the goal. The main finding: horizon alone is a training bottleneck. As goal distance grows, exploration explodes combinatorially and credit assignment gets ambiguous. Models that learn cleanly on short horizons fall apart on long ones, even when the underlying reasoning is identical. The fix is not more compute, it is horizon reduction.
1,000 Synthetic Computers
Microsoft Research builds 1,000 synthetic computers, each with realistic directory structures, documents, and artifacts, then runs long-horizon simulations on top of them. One agent plays the user and sets productivity goals; another executes the work. Each simulation runs over 8 hours of agent runtime and 2,000+ turns on average, roughly a month of human work compressed into one trace. Training on this experiential data drives significant improvements on both in-domain and out-of-domain productivity evaluations.
Contextual Agentic Memory is a Memo
Most agent memory today is not memory, it is closer to a memo. Vector stores, RAG buffers, and scratchpads implement lookup, not consolidation. The paper draws on neuroscience's Complementary Learning Systems theory: biological intelligence pairs fast hippocampal storage with slow neocortical consolidation, and current AI agents only implement the first half (fast write, similarity recall, no abstraction step). The authors prove a generalization ceiling on compositionally novel tasks: as long as memory stays retrieval-only, the agent cannot apply abstract rules to inputs that do not already look like something in the store, and it remains permanently exposed to memory poisoning. If you are building long-running agents and treating memory as a vector index, this paper is a clean diagnosis of what you are missing.
Agentic-imodels
The entire interpretability literature is built around human readers. As more analysis gets delegated to agents, the right target of interpretability shifts. Microsoft Research introduces Agentic-imodels, an autoresearch loop where a coding agent (Claude Code, Codex) iteratively evolves scikit-learn-compatible regressors that are simultaneously accurate AND readable by other LLMs. Interpretability is measured by whether a small LLM can simulate the fitted model's behavior just by reading its string representation, predictions, feature effects, and counterfactuals from the __str__ output alone. Across 65 tabular datasets, the discovered models push the Pareto frontier past every classical interpretable baseline (decision trees, GAMs, sparse linear), and improve four downstream agentic data-science systems on the BLADE benchmark by 8% to 73%.
Skills as Verifiable Artifacts
If you ship agent skills, your runtime is treating signed-and-cleared skills as trusted by default. This paper argues a skill is untrusted code until it is verified, and the runtime should enforce that default rather than infer trust from origin. Without skill verification, HITL has to fire on every irreversible call, which degrades into rubber-stamping at any non-trivial scale. With verification as a separate gated process, HITL fires only for what is unverified. Skills are now first-class deployment artifacts, and we have decades of supply-chain lessons on what happens when trust is inferred from a signature. This is the right ask for SKILL.md before agent skill libraries become the next attack surface.
Agentic Harness Engineering
Most coding-agent harnesses are still tuned by hand or kept alive through brittle trial-and-error self-evolution. This paper introduces Agentic Harness Engineering (AHE), a framework that makes harness evolution observable and falsifiable. AHE separates the system into three layers: components stored as revertible files, experience condensed from millions of trajectory tokens into structured evidence, and decisions written as predictions that get checked against task outcomes. Every edit becomes a contract you can verify or revert.
AgenticQwen-30B-A3B
Alibaba shows that a 30B MoE model with only 3B active parameters can match Qwen3-235B on real tool-use workloads. AgenticQwen-30B-A3B scores 50.2 average on TAU-2 plus BFCL-V4 Multi-Turn, while AgenticQwen-8B scores 47.4. Both more than double their vanilla Qwen baselines and close most of the gap to a 235B model. The recipe is built around two reinforcement learning flywheels that run in parallel, with simulated users actively trying to mislead the agent.
Agentic World Modeling
A massive 40-author survey lands the cleanest taxonomy of world models in agent research released so far. The paper proposes a "levels by laws" framework spanning three capability levels and four law regimes, then synthesizes 400+ works and 100+ representative systems across model-based RL, video generation, web and GUI agents, multi-agent simulation, and scientific discovery. As agents shift from chatbots to goal-accomplishers, the bottleneck moves from language to environment, and this is the first paper that gives builders a shared vocabulary across communities that have been working in isolation.
RecursiveMAS
Multi-agent systems usually pass full text messages between agents at every step, which causes token bloat, latency, and context dilution that all grow with team size. RecursiveMAS asks a different question: what if agents collaborated through recursive computation in a shared latent space instead of through text? The system treats a multi-agent team as a recursive computation where each agent acts like an RLM layer, iteratively passing latent representations to the next and forming a looped interaction process. Less talking, more thinking.
OneManCompany
If you are building multi-agent systems, you are probably wiring static org charts. This paper argues they should look more like a labor market. OneManCompany (OMC) replaces fixed teams with "Talents," portable agent identities that bundle skills and tools, and a "Talent Market" where agents get recruited dynamically per task. An Explore-Execute-Review tree search decomposes work hierarchically and aggregates results back up. On PRDBench, OMC reaches 84.67% success, +15.5 points over prior SOTA, and the framework generalizes across the case studies the authors run.
From Skill Text to Skill Structure
SKILL.md files entangle invocation interface, execution flow, and tool side effects in a single blob of natural language. That makes downstream discovery and risk review brittle as skill registries scale. This paper proposes SSL, a three-layer typed JSON representation drawn from Schank and Abelson's classical work on scripts, MOPs, and conceptual dependency. An LLM-based normalizer converts existing SKILL.md files into the structure, so adoption does not require rewriting registries by hand.
Latent Agents
Multi-agent debate makes models reason better. It also burns tokens generating long transcripts before any answer comes out. Latent Agents distills the entire debate into a single LLM through a two-stage fine-tuning pipeline: the model first learns debate structure, then internalizes it through dynamic reward scheduling and length clipping. The internalized model matches or beats explicit multi-agent debate while using up to 93% fewer tokens, which makes debate-quality reasoning practical at production scale.
OCR-Memory
Most agent memory systems compress trajectories into text summaries and hope the model remembers what matters, which is exactly where the information loss hides. OCR-Memory renders the agent's interaction history as images with indexed visual anchors, then retrieves via a locate-and-transcribe pipeline: the model scans visual memory, predicts the index of the relevant region, and the original text is fetched verbatim from a database. Older trajectories are stored as low-resolution thumbnails with active-recall up-sampling, and the method reaches SOTA on Mind2Web and AppWorld under strict context limits.