pith. machine review for the scientific record. sign in

arxiv: 2605.12966 · v1 · submitted 2026-05-13 · 💻 cs.AI

Recognition: 2 theorem links

· Lean Theorem

Position: Agentic AI System Is a Foreseeable Pathway to AGI

Junwei Liao , Shuai Li , Muning Wen , Jun Wang , Weinan Zhang
Authors on Pith no claims yet

Pith reviewed 2026-05-14 20:05 UTC · model grok-4.3

classification 💻 cs.AI
keywords agentic AIAGIdirected acyclic graphgeneralizationsample efficiencymonolithic scalingmixture of experts
0
0 comments X

The pith

Agentic AI systems using DAG topologies achieve exponentially superior generalization and sample efficiency compared to monolithic models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper challenges the idea that scaling up a single AI model is enough to reach AGI. It proposes Agentic AI, where multiple specialized agents are organized in a directed acyclic graph, as a better way to handle varied real-world tasks. Through theoretical comparisons, it shows these systems have much higher sample efficiency and generalization ability because they avoid the optimization bottlenecks of monolithic learners. This shift matters because real tasks come from complex, heterogeneous distributions that single models struggle with. The work also links this to mixture-of-experts approaches and urges more research into agentic frameworks.

Core claim

By contrasting the optimization constraints of monolithic learners against Agentic systems and progressing from simple routing to general DAG topologies, the authors demonstrate that Agentic AI achieves exponentially superior generalization and sample efficiency. This positions agentic structures as a necessary paradigm for mastering complex task distributions toward AGI.

What carries the argument

Directed Acyclic Graph (DAG) topologies for organizing specialized AI components, enabling efficient routing and composition that monolithic single-model architectures lack.

If this is right

  • Agentic AI will generalize better to new tasks with fewer examples due to modular structure.
  • Monolithic scaling hits fundamental limits in optimization for heterogeneous data.
  • Multi-agent systems can be stabilized by adopting general DAG topologies rather than ad-hoc designs.
  • Greater research investment in agentic AI will accelerate progress toward AGI.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the derivations hold, combining agentic DAGs with scaled models could create more efficient hybrid AGI pathways.
  • This framework suggests testable predictions on sample complexity for multi-task benchmarks.
  • Implications for AI safety include easier modularity and interpretability in agentic systems.

Load-bearing premise

That the optimization constraints of monolithic learners are fundamentally more limiting than those of agentic DAG systems.

What would settle it

A direct empirical comparison on a heterogeneous task distribution where a scaled monolithic model matches or exceeds the generalization and sample efficiency of an optimized agentic DAG system.

Figures

Figures reproduced from arXiv: 2605.12966 by Jun Wang, Junwei Liao, Muning Wen, Shuai Li, Weinan Zhang.

Figure 1
Figure 1. Figure 1: Agentic AI expands the range of usable tasks and im￾proves performance compared to monolithic models. While mono￾lithic models exhibit narrow performance peaks only on specific tasks they are trained for, Agentic AI demonstrates multi-peak performance across a broader spectrum. This expands usable capa￾bilities, approaching and even surpassing the altitude and breadth of human intelligence. multi-peak perf… view at source ↗
Figure 2
Figure 2. Figure 2: A demonstration of the Average Trap. The monolithic optimum is pulled towards the sharp task, illustrating the curvature￾induced bias described in Proposition 3.3. global optimum, the system exploits the geometric decom￾posability of the task mixture. We formalize the routed agentic hypothesis by assuming the target function f can be factorized through a routing mechanism π and a set of local maps: fR-Agen… view at source ↗
read the original abstract

Is monolithic scaling the only path to AGI? This paper challenges the dogma that purely scaling a single model is sufficient to achieve Artificial General Intelligence. Instead, we identify Agentic AI as a necessary paradigm for mastering the complex, heterogeneous distribution of real-world tasks. Through rigorous theoretical derivations, we contrast the optimization constraints of monolithic learners against the efficiency of Agentic systems, progressing from simple routing mechanisms to general Directed Acyclic Graph (DAG) topologies. We demonstrate that Agentic AI achieves exponentially superior generalization and sample efficiency. Finally, we discuss the connection to Mixture-of-Experts, reinterpret the instability of current multi-agent frameworks, and call for greater research focus on Agentic AI.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper argues that monolithic scaling of single models is insufficient for AGI and positions Agentic AI systems structured as Directed Acyclic Graphs (DAGs) as a necessary paradigm. It claims to provide rigorous theoretical derivations contrasting the optimization constraints of monolithic learners with the efficiency of agentic routing and specialization mechanisms, demonstrating exponentially superior generalization and sample efficiency. The work also reinterprets Mixture-of-Experts models and instabilities in multi-agent frameworks while calling for increased research focus on agentic approaches.

Significance. If the claimed exponential improvements in generalization and sample efficiency can be rigorously derived and validated, the paper would have substantial significance in redirecting AGI research away from pure scaling toward modular, agentic architectures capable of handling heterogeneous real-world tasks. It offers a conceptual bridge between current MoE systems and more general DAG-based agentic designs that could inform future system-level innovations.

major comments (2)
  1. [Abstract] Abstract: The central claim of 'rigorous theoretical derivations' showing that Agentic AI achieves 'exponentially superior generalization and sample efficiency' is unsupported, as the manuscript supplies no equations, complexity bounds, PAC-style analyses, or explicit comparisons between monolithic constraints and agentic DAG topologies. This absence makes the exponential (as opposed to polynomial) improvement an assertion rather than a derived result and is load-bearing for the paper's primary thesis.
  2. [Main text] Main argument: The contrast between 'optimization constraints of monolithic learners' and 'efficiency of Agentic systems' risks circularity, as the superiority appears defined relative to assumptions internal to the agentic DAG framing (e.g., routing and specialization enabling exponential gains) without independent external benchmarks, information-theoretic arguments, or falsifiable predictions to establish the claimed gap.
minor comments (1)
  1. The discussion of connections to Mixture-of-Experts would be strengthened by citing specific prior works on MoE scaling laws or routing mechanisms to ground the reinterpretation of multi-agent instabilities.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. We address each major comment below and outline revisions that will be incorporated into the next version of the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim of 'rigorous theoretical derivations' showing that Agentic AI achieves 'exponentially superior generalization and sample efficiency' is unsupported, as the manuscript supplies no equations, complexity bounds, PAC-style analyses, or explicit comparisons between monolithic constraints and agentic DAG topologies. This absence makes the exponential (as opposed to polynomial) improvement an assertion rather than a derived result and is load-bearing for the paper's primary thesis.

    Authors: We agree that the abstract overstates the formality of the arguments. The manuscript is a position paper whose core contribution is a conceptual contrast between monolithic and agentic optimization landscapes, supported by qualitative reasoning and connections to existing MoE results rather than formal PAC bounds or complexity derivations. In revision we will replace 'rigorous theoretical derivations' with 'theoretical arguments' and remove the specific claim of 'exponentially superior' generalization, replacing it with 'substantially improved' to reflect the level of support actually provided. These wording changes will be made throughout the abstract and introduction. revision: yes

  2. Referee: [Main text] Main argument: The contrast between 'optimization constraints of monolithic learners' and 'efficiency of Agentic systems' risks circularity, as the superiority appears defined relative to assumptions internal to the agentic DAG framing (e.g., routing and specialization enabling exponential gains) without independent external benchmarks, information-theoretic arguments, or falsifiable predictions to establish the claimed gap.

    Authors: We acknowledge the risk of circularity in the current framing. To mitigate this, the revised manuscript will (1) cite information-theoretic results on task decomposition and modular representations (e.g., from the literature on hierarchical Bayesian models and compositional generalization), (2) reference empirical scaling trends observed in Mixture-of-Experts systems as external evidence, and (3) add a short subsection listing concrete, testable predictions (such as sample-efficiency gains on heterogeneous task suites when routing is introduced). These additions will ground the comparison in independent literature and observable outcomes rather than solely in the DAG framing itself. revision: yes

Circularity Check

0 steps flagged

No circularity: position paper asserts derivations without self-referential reduction

full rationale

The paper is a position piece that claims 'rigorous theoretical derivations' progressing from routing to DAG topologies and demonstrating 'exponentially superior generalization and sample efficiency' for agentic systems. No equations, parameter fits, self-citations, or ansatzes are supplied in the provided text that reduce this superiority claim to its own inputs by construction. The contrast between monolithic constraints and agentic efficiency is presented as a demonstration rather than a fitted or self-defined result, and no load-bearing step collapses to a prior self-citation or renaming. The derivation chain is therefore self-contained as an assertion within the position framing, with no exhibited circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on unstated assumptions about optimization landscapes in monolithic versus agentic systems and the exponential nature of efficiency gains, with no free parameters or invented entities explicitly listed but implicit in the theoretical contrast.

axioms (2)
  • domain assumption Monolithic scaling faces inherent optimization constraints that limit generalization on heterogeneous tasks
    Invoked to establish the baseline inferiority of single-model approaches.
  • ad hoc to paper Agentic DAG topologies enable exponentially better sample efficiency through routing and specialization
    Core to the claimed superiority without external validation.

pith-pipeline@v0.9.0 · 5414 in / 1258 out tokens · 73373 ms · 2026-05-14T20:05:18.066719+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

90 extracted references · 90 canonical work pages · 2 internal anchors

  1. [1]

    Langley , title =

    P. Langley , title =. Proceedings of the 17th International Conference on Machine Learning (ICML 2000) , address =. 2000 , pages =

    work page 2000

  2. [2]

    T. M. Mitchell. The Need for Biases in Learning Generalizations. 1980

    work page 1980

  3. [3]

    M. J. Kearns , title =

    work page

  4. [4]

    Machine Learning: An Artificial Intelligence Approach, Vol. I. 1983

    work page 1983

  5. [5]

    R. O. Duda and P. E. Hart and D. G. Stork. Pattern Classification. 2000

    work page 2000

  6. [6]

    Suppressed for Anonymity , author=

    work page

  7. [7]

    Newell and P

    A. Newell and P. S. Rosenbloom. Mechanisms of Skill Acquisition and the Law of Practice. Cognitive Skills and Their Acquisition. 1981

    work page 1981

  8. [8]

    A. L. Samuel. Some Studies in Machine Learning Using the Game of Checkers. IBM Journal of Research and Development. 1959

    work page 1959

  9. [9]

    and Macready, W.G

    Wolpert, D.H. and Macready, W.G. , journal=. No free lunch theorems for optimization , year=

    work page

  10. [10]

    Stone , title =

    Charles J. Stone , title =. The Annals of Statistics , number =. 1982 , doi =

    work page 1982

  11. [11]

    Upper bounds on the

    Jin, Ying , booktitle=. Upper bounds on the. 2023 , volume=

    work page 2023

  12. [12]

    Multiclass Learnability and the

    Daniely, Amit and Sabato, Sivan and Ben-David, Shai and Shalev-Shwartz, Shai , booktitle =. Multiclass Learnability and the. 2011 , editor =

    work page 2011

  13. [13]

    Proceedings of the 26th International Conference on Neural Information Processing Systems - Volume 1 , pages =

    Daniely, Amit and Sabato, Sivan and Shalev-Shwartz, Shai , title =. Proceedings of the 26th International Conference on Neural Information Processing Systems - Volume 1 , pages =. 2012 , publisher =

    work page 2012

  14. [14]

    and Sethna, James P

    Ramesh, Rahul and Mao, Jialin and Griniasty, Itay and Yang, Rubing and Teoh, Han Kheng and Transtrum, Mark K. and Sethna, James P. and Chaudhari, Pratik , title =. Proceedings of the 40th International Conference on Machine Learning , articleno =. 2023 , publisher =

    work page 2023

  15. [15]

    , title =

    Tsybakov, Alexandre B. , title =. 2008 , isbn =

    work page 2008

  16. [16]

    Tohoku Mathematical Journal , number =

    Kazuoki Azuma , title =. Tohoku Mathematical Journal , number =. 1967 , doi =

    work page 1967

  17. [17]

    Probability Inequalities for Sums of Bounded Random Variables , urldate =

    Wassily Hoeffding , journal =. Probability Inequalities for Sums of Bounded Random Variables , urldate =

    work page

  18. [18]

    2020 , eprint=

    Scaling Laws for Neural Language Models , author=. 2020 , eprint=

    work page 2020

  19. [19]

    DOI: https://doi.org/10.1016/S0004-3702(99)00052-1

    Richard S. Sutton and Doina Precup and Satinder Singh , keywords =. Between. Artificial Intelligence , volume =. 1999 , issn =. doi:https://doi.org/10.1016/S0004-3702(99)00052-1 , url =

    work page doi:10.1016/s0004-3702(99)00052-1 1999

  20. [20]

    , title =

    Dietterich, Thomas G. , title =. J. Artif. Int. Res. , month = nov, pages =. 2000 , issue_date =

    work page 2000

  21. [21]

    Natarajan, B. K. , title =. Mach. Learn. , month = oct, pages =. 1989 , issue_date =. doi:10.1023/A:1022605311895 , abstract =

    work page doi:10.1023/a:1022605311895 1989

  22. [22]

    Vapnik, V. N. and Chervonenkis, A. Ya. , title =. Theory of Probability & Its Applications , volume =. 1971 , doi =

    work page 1971

  23. [23]

    1957 , publisher=

    Dynamic Programming , author=. 1957 , publisher=

    work page 1957

  24. [24]

    An empirical analysis of compute-optimal large language model training , volume =

    Hoffmann, Jordan and Borgeaud, Sebastian and Mensch, Arthur and Buchatskaya, Elena and Cai, Trevor and Rutherford, Eliza and de Las Casas, Diego and Hendricks, Lisa Anne and Welbl, Johannes and Clark, Aidan and Hennigan, Thomas and Noland, Eric and Millican, Katherine and van den Driessche, George and Damoc, Bogdan and Guy, Aurelia and Osindero, Simon and...

    work page

  25. [25]

    Resolving Discrepancies in Compute-Optimal Scaling of Language Models , volume =

    Porian, Tomer and Wortsman, Mitchell and Jitsev, Jenia and Schmidt, Ludwig and Carmon, Yair , booktitle =. Resolving Discrepancies in Compute-Optimal Scaling of Language Models , volume =. doi:10.52202/079017-3189 , editor =

    work page doi:10.52202/079017-3189

  26. [26]

    Reconciling

    Tim Pearce and Jinyeop Song , journal=. Reconciling. 2024 , url=

    work page 2024

  27. [27]

    On the method of bounded differences , booktitle=

    McDiarmid, Colin , year=. On the method of bounded differences , booktitle=

    work page

  28. [28]

    Approximation Rate of the

    Jiang, Haotian and Li, Qianxiao , booktitle =. Approximation Rate of the. doi:10.52202/079017-2202 , editor =

    work page doi:10.52202/079017-2202

  29. [29]

    Chulhee Yun and Srinadh Bhojanapalli and Ankit Singh Rawat and Sashank Reddi and Sanjiv Kumar , booktitle=. Are. 2020 , url=

    work page 2020

  30. [30]

    Error bounds for approximations with deep

    Dmitry Yarotsky , keywords =. Error bounds for approximations with deep. Neural Networks , volume =. 2017 , issn =. doi:https://doi.org/10.1016/j.neunet.2017.07.002 , url =

    work page doi:10.1016/j.neunet.2017.07.002 2017

  31. [31]

    , journal=

    Barron, A.R. , journal=. Universal approximation bounds for superpositions of a sigmoidal function , year=

    work page

  32. [32]

    2023 , url=

    Shunyu Yao and Jeffrey Zhao and Dian Yu and Nan Du and Izhak Shafran and Karthik R Narasimhan and Yuan Cao , booktitle=. 2023 , url=

    work page 2023

  33. [33]

    2023 , url=

    Noah Shinn and Federico Cassano and Ashwin Gopinath and Karthik R Narasimhan and Shunyu Yao , booktitle=. 2023 , url=

    work page 2023

  34. [34]

    2023 , url=

    Timo Schick and Jane Dwivedi-Yu and Roberto Dessi and Roberta Raileanu and Maria Lomeli and Eric Hambro and Luke Zettlemoyer and Nicola Cancedda and Thomas Scialom , booktitle=. 2023 , url=

    work page 2023

  35. [35]

    Weiwen Liu and Xu Huang and Xingshan Zeng and xinlong hao and Shuai Yu and Dexun Li and Shuai Wang and Weinan Gan and Zhengying Liu and Yuanqing Yu and Zezhong WANG and Yuxian Wang and Wu Ning and Yutai Hou and Bin Wang and Chuhan Wu and Wang Xinzhi and Yong Liu and Yasheng Wang and Duyu Tang and Dandan Tu and Lifeng Shang and Xin Jiang and Ruiming Tang a...

    work page 2025

  36. [36]

    The Thirteenth International Conference on Learning Representations , year=

    Robust Function-Calling for On-Device Language Model via Function Masking , author=. The Thirteenth International Conference on Learning Representations , year=

    work page

  37. [37]

    Guangming Sheng and Chi Zhang and Zilingfeng Ye and Xibin Wu and Wang Zhang and Ru Zhang and Yanghua Peng and Haibin Lin and Chuan Wu , year =

    work page

  38. [38]

    AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning

    Wei Fu and Jiaxuan Gao and Xujie Shen and Chen Zhu and Zhiyu Mei and Chuyi He and Shusheng Xu and Guo Wei and Jun Mei and Jiashu Wang and Tongkai Yang and Binhang Yuan and Yi Wu , year=. 2505.24298 , archivePrefix=

    work page internal anchor Pith review arXiv

  39. [39]

    Jian Hu and Xibin Wu and Zilin Zhu and Xianyu and Weixun Wang and Dehao Zhang and Yu Cao , journal=

    work page

  40. [40]

    arXiv preprint arXiv:2411.14033 , year =

    Yingxuan Yang and Qiuying Peng and Jun Wang and Ying Wen and Weinan Zhang , year=. 2411.14033 , archivePrefix=

    work page arXiv

  41. [41]

    MARFT: multi-agent reinforcement fine-tuning.CoRR, abs/2504.16129, 2025

    Junwei Liao and Muning Wen and Jun Wang and Weinan Zhang , year=. 2504.16129 , archivePrefix=

    work page arXiv

  42. [42]

    AgentNet: Decentralized Evolutionary Coordination for

    Yingxuan Yang and Huacan Chai and Shuai Shao and Yuanyi Song and Siyuan Qi and Renting Rui and Weinan Zhang , booktitle=. AgentNet: Decentralized Evolutionary Coordination for. 2025 , url=

    work page 2025

  43. [43]

    2024 , eprint=

    Large Language Model based Multi-Agents: A Survey of Progress and Challenges , author=. 2024 , eprint=

    work page 2024

  44. [44]

    Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Xiao Bi and Haowei Zhang and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo , year=. 2402.03300 , archivePrefix=

    work page internal anchor Pith review Pith/arXiv arXiv

  45. [45]

    O’Brien, Carrie J

    Park, Joon Sung and O'Brien, Joseph and Cai, Carrie Jun and Morris, Meredith Ringel and Liang, Percy and Bernstein, Michael S. , title =. 2023 , isbn =. doi:10.1145/3586183.3606763 , booktitle =

    work page doi:10.1145/3586183.3606763 2023

  46. [46]

    Multi-Agent Evolve:

    Yixing Chen and Yiding Wang and Siqi Zhu and Haofei Yu and Tao Feng and Muhan Zhang and Mostofa Patwary and Jiaxuan You , year=. Multi-Agent Evolve:. 2510.23595 , archivePrefix=

    work page arXiv

  47. [47]

    MultiAgentBench : Evaluating the collaboration and competition of LLM agents

    Zhu, Kunlun and Du, Hongyi and Hong, Zhaochen and Yang, Xiaocheng and Guo, Shuyi and Wang, Zhe and Wang, Zhenhailong and Qian, Cheng and Tang, Robert and Ji, Heng and You, Jiaxuan. M ulti A gent B ench : Evaluating the Collaboration and Competition of LLM agents. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volu...

    work page doi:10.18653/v1/2025.acl-long.421 2025

  48. [48]

    Reflective Multi-Agent Collaboration based on Large Language Models , url =

    Bo, Xiaohe and Zhang, Zeyu and Dai, Quanyu and Feng, Xueyang and Wang, Lei and Li, Rui and Chen, Xu and Wen, Ji-Rong , booktitle =. Reflective Multi-Agent Collaboration based on Large Language Models , url =. doi:10.52202/079017-4397 , editor =

    work page doi:10.52202/079017-4397

  49. [49]

    2024 , note =

    OpenAI , title =. 2024 , note =

    work page 2024

  50. [50]

    2024 , note =

    Anthropic , title =. 2024 , note =

    work page 2024

  51. [51]

    2024 , note =

    Manus , title =. 2024 , note =

    work page 2024

  52. [52]

    2024 , note =

    Cursor , title =. 2024 , note =

    work page 2024

  53. [53]

    The Twelfth International Conference on Learning Representations , year=

    Gr. The Twelfth International Conference on Learning Representations , year=

    work page

  54. [54]

    2025 , note =

    Anthropic , title =. 2025 , note =

    work page 2025

  55. [55]

    2024 , url=

    Carlos E Jimenez and John Yang and Alexander Wettig and Shunyu Yao and Kexin Pei and Ofir Press and Karthik R Narasimhan , booktitle=. 2024 , url=

    work page 2024

  56. [56]

    and Zhang, Kaiqing and Kim, Joo-Kyung

    Park, Chanwoo and Han, Seungju and Guo, Xingzhi and Ozdaglar, Asuman E. and Zhang, Kaiqing and Kim, Joo-Kyung. MAP o RL : Multi-Agent Post-Co-Training for Collaborative Large Language Models with Reinforcement Learning. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025. doi:10.18653/v1/20...

    work page doi:10.18653/v1/2025.acl-long.1459 2025

  57. [57]

    2510.11062 , archivePrefix=

    Yujie Zhao and Lanxiang Hu and Yang Wang and Minmin Hou and Hao Zhang and Ke Ding and Jishen Zhao , year=. 2510.11062 , archivePrefix=

    work page arXiv

  58. [58]

    Agentic Web: Weaving the Next Web with

    Yang, Yingxuan and Ma, Mulei and Huang, Yuxuan and Chai, Huacan and Gong, Chenyu and Geng, Haoran and Zhou, Yuanjian and Wen, Ying and Fang, Meng and Chen, Muhao and others , journal=. Agentic Web: Weaving the Next Web with

    work page

  59. [59]

    2508.13787 , archivePrefix=

    Zihan Guo and Yuanjian Zhou and Chenyi Wang and Linlin You and Minjie Bian and Weinan Zhang , year=. 2508.13787 , archivePrefix=

    work page arXiv

  60. [60]

    Rao Surapaneni and Miku Jha and Michael Vakoc and Todd Segal , title =

    work page

  61. [61]

    and Kumar, Sanjiv , title =

    Yun, Chulhee and Chang, Yin-Wen and Bhojanapalli, Srinadh and Rawat, Ankit Singh and Reddi, Sashank J. and Kumar, Sanjiv , title =. Proceedings of the 34th International Conference on Neural Information Processing Systems , articleno =. 2020 , isbn =

    work page 2020

  62. [62]

    , title =

    Cybenko, G. , title =. Mathematics of Control, Signals and Systems , year =. doi:10.1007/BF02551274 , url =

    work page doi:10.1007/bf02551274

  63. [63]

    Multilayer feedforward networks are universal approximators , journal =

    Kurt Hornik and Maxwell Stinchcombe and Halbert White , keywords =. Multilayer feedforward networks are universal approximators , journal =. 1989 , issn =. doi:https://doi.org/10.1016/0893-6080(89)90020-8 , url =

    work page doi:10.1016/0893-6080(89)90020-8 1989

  64. [64]

    Roumeliotis, and Manoj Karkee

    Sapkota, Ranjan and Roumeliotis, Konstantinos I. and Karkee, Manoj , year=. doi:10.1016/j.inffus.2025.103599 , journal=

    work page doi:10.1016/j.inffus.2025.103599 2025

  65. [65]

    2019 , eprint=

    On the Measure of Intelligence , author=. 2019 , eprint=

    work page 2019

  66. [66]

    2007 , eprint=

    Universal Intelligence: A Definition of Machine Intelligence , author=. 2007 , eprint=

    work page 2007

  67. [67]

    Transactions on Machine Learning Research , issn=

    A Generalist Agent , author=. Transactions on Machine Learning Research , issn=. 2022 , url=

    work page 2022

  68. [68]

    2025 , eprint=

    Humanity's Last Exam , author=. 2025 , eprint=

    work page 2025

  69. [69]

    and Mao, Huanzhi and Cheng-Jie Ji, Charlie and Yan, Fanjia and Suresh, Vishnu and Stoica, Ion and E

    Patil, Shishir G. and Mao, Huanzhi and Cheng-Jie Ji, Charlie and Yan, Fanjia and Suresh, Vishnu and Stoica, Ion and E. Gonzalez, Joseph , booktitle=. The

    work page

  70. [70]

    ICLR 2025 Workshop on Building Trust in Language Models and Applications , year=

    Why Do Multiagent Systems Fail? , author=. ICLR 2025 Workshop on Building Trust in Language Models and Applications , year=

    work page 2025

  71. [71]

    LLM -Coordination: Evaluating and Analyzing Multi-agent Coordination Abilities in Large Language Models

    Agashe, Saaket and Fan, Yue and Reyna, Anthony and Wang, Xin Eric. LLM -Coordination: Evaluating and Analyzing Multi-agent Coordination Abilities in Large Language Models. Findings of the Association for Computational Linguistics: NAACL 2025. 2025. doi:10.18653/v1/2025.findings-naacl.448

    work page doi:10.18653/v1/2025.findings-naacl.448 2025

  72. [72]

    Which Agent Causes Task Failures and When? On Automated Failure Attribution of

    Shaokun Zhang and Ming Yin and Jieyu Zhang and Jiale Liu and Zhiguang Han and Jingyang Zhang and Beibin Li and Chi Wang and Huazheng Wang and Yiran Chen and Qingyun Wu , booktitle=. Which Agent Causes Task Failures and When? On Automated Failure Attribution of. 2025 , url=

    work page 2025

  73. [73]

    Artificial General Intelligence Is Already Here , howpublished =

    Ag. Artificial General Intelligence Is Already Here , howpublished =. 2023 , month = oct, note =

    work page 2023

  74. [74]

    2024 , howpublished =

    Scale is Not All You Need! , author =. 2024 , howpublished =

    work page 2024

  75. [75]

    Proceedings of the 41st International Conference on Machine Learning , articleno =

    Dohmatob, Elvis and Feng, Yunzhen and Yang, Pu and Charton, Francois and Kempe, Julia , title =. Proceedings of the 41st International Conference on Machine Learning , articleno =. 2024 , publisher =

    work page 2024

  76. [76]

    2018 , URL =

    Relational inductive biases, deep learning, and graph networks , author =. 2018 , URL =

    work page 2018

  77. [77]

    Edward J Hu and Yelong Shen and Phillip Wallis and Zeyuan Allen-Zhu and Yuanzhi Li and Shean Wang and Lu Wang and Weizhu Chen , booktitle=. Lo. 2022 , url=

    work page 2022

  78. [78]

    Cours d'analyse de l'

    Cauchy, Augustin-Louis , year=. Cours d'analyse de l'

    work page

  79. [79]

    , booktitle=

    Gudwin, R.R. , booktitle=. Evaluating intelligence: a computational semiotics perspective , year=

    work page

  80. [80]

    Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop , year =

    A Native Intelligence Metric for Artificial Systems , author =. Proceedings of the Performance Metrics for Intelligent Systems (PerMIS) Workshop , year =

    work page

Showing first 80 references.