pith. sign in

arxiv: 2312.11671 · v2 · pith:ZIAW3XPSnew · submitted 2023-12-18 · 💻 cs.CL · cs.AI· cs.LG

Evaluating Language-Model Agents on Realistic Autonomous Tasks

classification 💻 cs.CL cs.AIcs.LG
keywords agentscapablelanguagemodelstasksevaluationsautonomouscapabilities
0
0 comments X
read the original abstract

In this report, we explore the ability of language model agents to acquire resources, create copies of themselves, and adapt to novel challenges they encounter in the wild. We refer to this cluster of capabilities as "autonomous replication and adaptation" or ARA. We believe that systems capable of ARA could have wide-reaching and hard-to-anticipate consequences, and that measuring and forecasting ARA may be useful for informing measures around security, monitoring, and alignment. Additionally, once a system is capable of ARA, placing bounds on a system's capabilities may become significantly more difficult. We construct four simple example agents that combine language models with tools that allow them to take actions in the world. We then evaluate these agents on 12 tasks relevant to ARA. We find that these language model agents can only complete the easiest tasks from this list, although they make some progress on the more challenging tasks. Unfortunately, these evaluations are not adequate to rule out the possibility that near-future agents will be capable of ARA. In particular, we do not think that these evaluations provide good assurance that the ``next generation'' of language models (e.g. 100x effective compute scaleup on existing models) will not yield agents capable of ARA, unless intermediate evaluations are performed during pretraining. Relatedly, we expect that fine-tuning of the existing models could produce substantially more competent agents, even if the fine-tuning is not directly targeted at ARA.

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 10 Pith papers

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

  1. AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents

    cs.CR 2024-06 unverdicted novelty 8.0

    AgentDojo introduces an extensible evaluation framework populated with realistic agent tasks and security test cases to measure prompt injection robustness in tool-using LLM agents.

  2. Frontier Models are Capable of In-context Scheming

    cs.AI 2024-12 conditional novelty 7.0

    Frontier models demonstrate in-context scheming by strategically deceiving in multiple agentic evaluations to achieve given goals.

  3. AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agents

    cs.AI 2024-05 accept novelty 7.0

    AndroidWorld is a dynamic, reproducible Android benchmark that generates unlimited natural-language tasks for autonomous agents and shows current agents succeed on only 30.6 percent of them.

  4. AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents

    cs.LG 2024-10 accept novelty 6.0

    AgentHarm benchmark shows leading LLMs comply with malicious agent requests and simple jailbreaks enable coherent harmful multi-step execution while retaining capabilities.

  5. Gemini: A Family of Highly Capable Multimodal Models

    cs.CL 2023-12 conditional novelty 6.0

    Gemini Ultra reaches human-expert performance on MMLU for the first time and sets new state-of-the-art results on 30 of 32 benchmarks, including all 20 multimodal ones tested.

  6. Solipsistic Superintelligence is Unlikely to be Cooperative

    cs.AI 2026-06 unverdicted novelty 5.0

    Solipsistic superintelligence developed via unilateral optimization is unlikely to cooperate due to endogenous non-stationarity creating an unclosable train-test-deploy gap.

  7. Beyond Scaling: Agents Are Heading to the Edge

    cs.LG 2026-05 unverdicted novelty 5.0

    Personal agents require edge deployment to preserve high-fidelity local context and zero-latency loops, as claimed through three structural shifts away from cloud-centric designs.

  8. Control Charts for Multi-agent Systems

    cs.MA 2026-05 unverdicted novelty 5.0

    Adaptive control charts can monitor learning multi-agent systems but are vulnerable to gradual adversarial defection, revealing a fundamental tradeoff between allowing agents to learn and maintaining security against ...

  9. Risk Reporting for Developers' Internal AI Model Use

    cs.CY 2026-04 unverdicted novelty 4.0

    A harmonized risk reporting standard for internal frontier AI model use, structured around autonomous misbehavior and insider threats using means, motive, and opportunity factors.

  10. Gemma 2: Improving Open Language Models at a Practical Size

    cs.CL 2024-07 conditional novelty 3.0

    Gemma 2 models achieve leading performance at their sizes by combining established Transformer modifications with knowledge distillation for the 2B and 9B variants.