Paper page - Does Scaling Toward AGI Depend on What System-1 Lets System-2 Execute?

Source: https://huggingface.co/papers/2610.39497 Published on Oct 29

·

Submitted byhttps://huggingface.co/yunqu

Yun Quon Nov 2

#3 Paper of the day Authors:

,

,

,

,

,

,

,

,

,

,

Abstract

Co-designing a resource-heavy reasoning system with a lightweight predictive evaluator can reduce execution costs by filtering interactions, with efficiency governed by the density of useful interactions.

Progress toward artificial general intelligence (AGI) has largely relied on scaling a resource-intensive executive,System-2, through larger models, more data and compute, longer reasoning, and more agents. We ask whether the next stage may come instead from co-designing it withSystem-1, a lightweight, semantically aware predictor that evaluates candidate interactions before execution, so thatSystem-2acts only on those worth their cost. Our answer is a qualified yes. We studyconditional interaction, a predict--gate--execute--update loop:System-1estimates each candidate’s utility,System-2executes those admitted by anadaptive threshold, and the outcomes updateSystem-1. The loop spans training, test-time reasoning, adaptive decision-making, and agentic tool use, with utility and cost defined by the application. Under ideal screening and equal execution costs, execution compute falls by a factor approaching 1/ρ_u, where ρ_u is the density of utility-relevant interactions (cost-weighted when costs differ); prediction errors, exploration, and distribution shift reduce the saving. We boundselection regretby predictor error and theutility margin. These results motivate our central hypothesis: ρ_u is a measurable coordinate ofscaling efficiencyand makesinformative computea better description of scaling than total compute.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2610\.39497

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2610.39497 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.39497 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2610.39497 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.