Researchers at UT-Austin and Google Model Human Decision-Making in Rock-Paper-Scissors
Researchers from UT-Austin and Google used AlphaEvolve, an evolutionary code-optimization method, to synthesize interpretable Python programs that predict move-by-move decisions of LLMs and humans playing rock-paper-scissors against bots. They found that Gemini 2.5 Pro, Gemini 2.5 Flash, and GPT-4.1 share similar sequential-pattern-tracking strategies that are more systematic than typical human play, while GPT-OSS 120B and humans relied on simpler opponent-move-frequency heuristics. The study demonstrates that code synthesis from behavioral data can serve as an interpretability tool for LLM decision-making, revealing that LLMs do not simply mimic human strategies.
Related guides (3)
Related events (8)
AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
DeepMind has announced AlphaEvolve, a coding agent powered by Gemini that autonomously evolves algorithms for mathematical and practical computing applications. The system combines large language model creativity with automated evaluators to iteratively improve algorithmic solutions. It represents a significant step in AI-driven algorithm discovery, extending DeepMind's prior work in this space (e.g., AlphaTensor, FunSearch). The announcement comes from DeepMind's official blog, indicating a substantive capability release rather than a research preview.
AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields
DeepMind published a blog post detailing the real-world impact of AlphaEvolve, a Gemini-powered coding agent designed to discover and optimize algorithms. The post covers applications spanning business operations, infrastructure, and scientific research. AlphaEvolve represents a deployment of LLM-driven evolutionary algorithm search at scale across multiple domains.
OpenAI GPT-5.4 Pro and GPT-5.4 Thinking challenge Gemini 3.1 Pro Preview for top AI model position
OpenAI released GPT-5.4 in two variants (Pro and Thinking), featuring expanded context windows up to 1.05M tokens, native computer use, tool search capabilities, and adjustable reasoning levels. In independent benchmarks by Artificial Analysis, GPT-5.4 Pro at xhigh reasoning nearly ties Gemini 3.1 Pro Preview on the Intelligence Index (57 vs 57.2 points) but at roughly 3.3x the cost, while leading on coding and agentic sub-indices. The release leapfrogs Claude Opus 4.6 on most benchmarks but faces stiff competition from Google's Gemini 3.1 Pro Preview, which maintains a price and multimodal advantage.
Persona Generators: Evolutionary LLM Method for Diverse Synthetic Human Personas
Google researchers Davide Paglieri, Logan Cross, and colleagues propose Persona Generators, a system that uses the AlphaEvolve evolutionary algorithm to generate code that produces 25 diverse persona prompts covering a broad range of attitudes and opinions. The method iteratively optimizes persona prompt diversity using six metrics, outperforming Nemotron Personas (82% vs 76% coverage of possible responses) and a Concordia memory-based baseline (46%). The system uses Gemini 2.5 Pro for questionnaire generation and Gemma 3-27B-IT for persona simulation via the Concordia agent library. The approach reframes persona generation as a coverage optimization problem rather than a data-matching one, enabling more representative synthetic user populations for product research.
Frontier coding agents use metaprogramming to handle esoteric programming languages
A new arXiv paper evaluates six LLM-based coding agents on four esoteric programming languages (including Brainfuck and Befunge-98), finding that the strongest agents—Claude Opus 4.6 and GPT-5.4 xhigh—often avoid writing the target language directly, instead generating it via Python metaprograms. Forbidding this strategy causes large performance drops, and text guidance alone does not transfer the capability to weaker models, though sharing Opus-derived Python helper code does sharply improve mid-tier agents. The study reveals capability stratification that mainstream benchmarks like SWE-Bench Verified compress into narrow bands, suggesting frontier agents succeed by constructing and debugging working models of unfamiliar environments rather than pattern-matching to training data.
GLM-5.1 Open-Weights Model Targets Long-Running Agentic Tasks; Andrew Ng on Coding Agent Acceleration by Software Domain
Z.ai released GLM-5.1, an open-weights mixture-of-experts LLM (754B total / 40B active parameters) designed for sustained agentic coding tasks lasting up to eight hours, featuring iterative planning-execution-evaluation loops with thousands of tool calls. The model claims top open-weights performance on Artificial Analysis Intelligence Index and SWE-Bench Pro, available under MIT license via HuggingFace. The accompanying editorial by Andrew Ng offers a tiered framework for how much coding agents accelerate different software work categories—frontend most, then backend, infrastructure, and research least—with practical implications for team organization. A secondary item references data-center opposition and LLM helpfulness failure modes.
Program synthesis used to reverse-engineer transformer attention heads with executable Python surrogates
Researchers propose a pipeline that approximates transformer attention heads with executable Python programs generated by a language model, then re-ranked by held-out predictive accuracy. Applied to GPT-2, TinyLlama-1.1B, and Llama-3B, fewer than 1,000 programs reproduce attention patterns with >75% average IoU similarity on TinyStories. Replacing 25% of attention heads with programmatic surrogates incurs only a 16% average perplexity increase while preserving downstream QA performance, demonstrating a path toward symbolic transparency in neural models.
Data Points: Thinking Machines Interaction Model, ERNIE 5.1, Co-Mathematician, RL Conductor, and More
This edition of The Batch covers five notable AI developments: Thinking Machines' research preview of an 'interaction model' with a 200ms micro-turn multimodal architecture; Baidu's ERNIE 5.1, a compressed derivative of ERNIE 5.0 using only 6% of typical pre-training compute; Google DeepMind's Co-Mathematician collaborative workbench reaching 48% on FrontierMath Tier 4; a 7B RL Conductor model that orchestrates multi-agent workflows via reinforcement learning; and Google's Magic Pointer cursor system powered by Gemini. Secondary items include GitHub Copilot pricing restructuring ahead of usage-based billing.