The reviewed record of science sign in
Pith

arxiv: 2505.21731 · v1 · pith:UHO2HJ47 · submitted 2025-05-27 · cs.LG · cs.AI

Deep Reinforcement Learning Agents are not even close to Human Intelligence

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:UHO2HJ47record.jsonopen to challenge →

classification cs.LG cs.AI
keywords agentstaskshumanintelligencelearningdeepevaluationreinforcement
0
0 comments X
read the original abstract

Deep reinforcement learning (RL) agents achieve impressive results in a wide variety of tasks, but they lack zero-shot adaptation capabilities. While most robustness evaluations focus on tasks complexifications, for which human also struggle to maintain performances, no evaluation has been performed on tasks simplifications. To tackle this issue, we introduce HackAtari, a set of task variations of the Arcade Learning Environments. We use it to demonstrate that, contrary to humans, RL agents systematically exhibit huge performance drops on simpler versions of their training tasks, uncovering agents' consistent reliance on shortcuts. Our analysis across multiple algorithms and architectures highlights the persistent gap between RL agents and human behavioral intelligence, underscoring the need for new benchmarks and methodologies that enforce systematic generalization testing beyond static evaluation protocols. Training and testing in the same environment is not enough to obtain agents equipped with human-like intelligence.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. SLR: Automated Synthesis for Scalable Logical Reasoning

    cs.AI 2025-06 unverdicted novelty 7.0

    SLR automates creation of inductive logic tasks and a 20-level curriculum benchmark to train and evaluate LLM reasoning, yielding doubled accuracy for Llama-3-8B and generalization to other benchmarks.

  2. Bridging Physical Reasoning and Task Generalization via Visual Action Outcome Reasoning Alignment

    cs.AI 2026-07 conditional novelty 6.0

    VAORA aligns VLM chain-of-thought reasoning with visual scene observations and post-action outcomes via structured symbolic rewards, achieving cross-task and cross-environment generalization on physical reasoning benchmarks.

  3. GRAIL: Autonomous Concept Grounding for Neuro-Symbolic Reinforcement Learning

    cs.AI 2026-04 unverdicted novelty 6.0

    GRAIL autonomously grounds relational concepts in NeSy-RL by using LLM weak supervision followed by interaction-based refinement, matching or exceeding manually defined concepts on Atari games.

  4. ActivationReasoning: Logical Reasoning in Latent Activation Spaces

    cs.LG 2025-10 unverdicted novelty 6.0

    ActivationReasoning grounds logical reasoning in LLM latent activations via SAEs to enable structured inference, concept composition, and behavior steering on multi-hop, abstraction, and safety tasks.