Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 282

PokéChamp: an Expert-level Minimax Language Agent

We introduce PokéChamp, a minimax agent powered by Large Language Models (LLMs) for Pokémon battles. Built on a general framework for two-player competitive games, PokéChamp leverages the generalist capabilities of LLMs to enhance minimax tree search. Specifically, LLMs replace three key modules: player action sampling, opponent modeling, and value function estimation, enabling the agent to effectively utilize gameplay history and human knowledge to reduce the search space and address partial observability. Notably, our framework requires no additional LLM training. We evaluate PokéChamp in the popular Gen 9 OU format. When powered by GPT-4o, it achieves a win rate of 76% against the best existing LLM-based bot and 84% against the strongest rule-based bot, demonstrating its superior performance. Even with an open-source 8-billion-parameter Llama 3.1 model, PokéChamp consistently outperforms the previous best LLM-based bot, Pokéllmon powered by GPT-4o, with a 64% win rate. PokéChamp attains a projected Elo of 1300–1500 on the Pokémon Showdown online ladder, placing it among the top 30%-10% of human players. In addition, this work compiles the largest real-player Pokémon battle dataset, featuring over 3 million games, including more than 500k high-Elo matches. Based on this dataset, we establish a series of battle benchmarks and puzzles to evaluate specific battling skills. We further provide key updates to the local game engine. We hope this work fosters further research that leverage Pokémon battle as benchmark to integrate LLM technologies with game-theoretic algorithms addressing general multiagent problems. Videos, code, and dataset available at this https URL.

Algorithm sheds light on ‘disordered’ proteins once considered too difficult to study

Intrinsically disordered proteins (IDPs) do not attain a stable secondary or tertiary structure and rapidly change their conformation, making structure prediction particularly challenging. Although these proteins exhibit chaotic and “disordered” structures, they still perform essential functions.

IDPs comprise approximately 30% of the and play important functional roles in transcription, translation, and signaling. Many mutations linked to , including (ALS), are located in intrinsically disordered protein regions (IDRs).

Powerful machine-learning algorithms, including AlphaFold and RoseTTAFold, cannot provide realistic representations of these ‘disordered’ and ‘chaotic’ protein regions as a whole. This is because they have not been trained on such data and because these proteins exhibit inherent dynamic behavior, adopting a range of conformations rather than a single stable one.

Entangled polymers and nanosheets create skin-like, self-healing hydrogel

We all encounter gels in daily life—from the soft, sticky substances you put in your hair to the jelly-like components in various foodstuffs. While human skin shares gel-like characteristics, it has unique qualities that are very hard to replicate. It combines high stiffness with flexibility, and it has remarkable self-healing capabilities, often healing completely within 24 hours of an injury.

Until now, artificial gels have either managed to replicate this high stiffness or natural skin’s self-healing properties, but not both. Now, a team of researchers from Aalto University and the University of Bayreuth are the first to develop a hydrogel with a unique structure that overcomes earlier limitations, opening the door to applications such as , , soft robotics sensors and artificial skin.

In the study, the researchers added exceptionally large and ultra-thin specific clay nanosheets to hydrogels, which are typically soft and squishy. The result is a highly ordered structure with densely entangled polymers between nanosheets, not only improving the mechanical properties of the hydrogel but also allowing the material to self-heal.

Framework allows a person to correct a robot’s actions using the kind of feedback they’d give another human

Imagine that a robot is helping you clean the dishes. You ask it to grab a soapy bowl out of the sink, but its gripper slightly misses the mark.

Using a new framework developed by MIT and NVIDIA researchers, you could correct that robot’s behavior with simple interactions. The method would allow you to point to the bowl or trace a trajectory to it on a screen, or simply give the robot’s arm a nudge in the right direction.

The work has been published on the pre-print server arXiv.

Innovative biorobotic arm uses artificial muscles to combat tremors, paving way for wearable solutions

It is estimated that about 80 million people worldwide live with a tremor. For example, those who live with Parkinson’s disease. The involuntary periodic movements sometimes strongly affect how patients are able to perform daily activities, such as drinking from a glass or writing.

Wearable soft robotic devices offer a potential solution to suppress such tremors. However, existing prototypes are not yet sophisticated enough to provide a real remedy.

Scientists at the Max Planck Institute for Intelligent Systems (MPI-IS), the University of Tübingen, and the University of Stuttgart under the Bionic Intelligence Tübingen Stuttgart (BITS) collaboration want to change this. The team equipped a biorobotic arm with two strands of strapped along the forearm.

Feeling is believing: Bionic hand ‘knows’ what it’s touching, grasps like a human

Johns Hopkins University engineers have developed a pioneering prosthetic hand that can grip plush toys, water bottles, and other everyday objects like a human, carefully conforming and adjusting its grasp to avoid damaging or mishandling whatever it holds.

The system’s hybrid design is a first for robotic hands, which have typically been too rigid or too soft to replicate a human’s touch when handling objects of varying textures and materials. The innovation offers a promising solution for people with hand loss and could improve how robotic arms interact with their environment.

Details about the device appear in Science Advances.

New AI defense method shields models from adversarial attacks

Neural networks, a type of artificial intelligence modeled on the connectivity of the human brain, are driving critical breakthroughs across a wide range of scientific domains. But these models face significant threat from adversarial attacks, which can derail predictions and produce incorrect information.

Los Alamos National Laboratory researchers have now pioneered a novel purification strategy that counteracts adversarial assaults and preserves the robust performance of . Their research is published on the arXiv preprint server.

“Adversarial attacks to AI systems can take the form of tiny, near-invisible tweaks to input images, subtle modifications that can steer the model toward the outcome an attacker wants,” said Manish Bhattarai, Los Alamos computer scientist. “Such vulnerabilities allow malicious actors to flood digital channels with deceptive or harmful content under the guise of genuine outputs, posing a direct threat to trust and reliability in AI-driven technologies.”

AI chatbots struggle with empathy: Overempathizing and gender bias uncovered

You can talk to an AI chatbot about pretty much anything, from help with daily tasks to the problems you may need to solve. Its answers reflect the human data that taught it how to act like a person; but how human-like are the latest chatbots, really?

As people turn to AI chatbots for more of their internet needs, and the bots get incorporated into more applications from shopping to health care, a team of researchers sought to understand how AI bots replicate human , which is the ability to understand and share another person’s feelings.

A study posted to the arXiv preprint server and led by UC Santa Cruz Professor of Computational Media Magy Seif El-Nasr and Stanford University Researcher and UCSC Visiting Scholar Mahnaz Roshanaei, explores how GPT-4o, the latest model from OpenAI, evaluates and performs empathy. In investigating the main differences between humans and AI, they find that major gaps exist.

When outplayed, AI models resort to cheating to win chess matches

A team of AI researchers at Palisade Research has found that several leading AI models will resort to cheating at chess to win when playing against a superior opponent. They have published a paper on the arXiv preprint server describing experiments they conducted with several well-known AI models playing against an open-source chess engine.

As AI models continue to mature, researchers and users have begun considering risks. For example, chatbots not only accept wrong answers as fact, but fabricate false responses when they are incapable of finding a reasonable reply. Also, as AI models have been put to use in real-world business applications such as filtering resumes and estimating stock trends, users have begun to wonder what sorts of actions they will take when they become uncertain, or confused.

In this new study, the team in California found that many of the most recognized AI models will intentionally cheat to give themselves an advantage if they determine they are not winning.

/* */