HAIant (智然体): A human-centered AI framework for secure, personalized intelligence augmentation in biological research

IF 32.7 1区生物学 Q1 BIOCHEMISTRY & MOLECULAR BIOLOGY

Molecular Plant Pub Date : 2026-09-09 DOI:10.1016/j.molp.2026.08.011

Yunlong Zhang,Yunming Wang,Zhilong Zheng,Hao Wu,Xiangfeng Zeng,Qishuai Huang,Wendi Wei,Youkun Fan,Jiachen Du,Jinpeng Liu,Tao Zhou,Guanzhen Zhu,Rui Han,Ruozhuo Zheng,Hongwei Zhang,Chuang Zhang,Yi Wang,Xiangguo Liu,Weifu Li,Zaiwen Feng,Lin Li

{"title":"HAIant (智然体): A human-centered AI framework for secure, personalized intelligence augmentation in biological research","authors":"Yunlong Zhang,Yunming Wang,Zhilong Zheng,Hao Wu,Xiangfeng Zeng,Qishuai Huang,Wendi Wei,Youkun Fan,Jiachen Du,Jinpeng Liu,Tao Zhou,Guanzhen Zhu,Rui Han,Ruozhuo Zheng,Hongwei Zhang,Chuang Zhang,Yi Wang,Xiangguo Liu,Weifu Li,Zaiwen Feng,Lin Li","doi":"10.1016/j.molp.2026.08.011","DOIUrl":null,"url":null,"abstract":"Artificial intelligence (AI) has achieved human-level capabilities across multiple domains; however, concerns regarding data security, privacy, and domain-specific capability enhancement remain. Here, we present human + AI + ant (HAIant), a human-centered framework for secure and personalized intelligence augmentation based on locally deployed AI. HAIant integrates locally deployed large language models with personal data, individual reasoning processes, and domain-specific tools, enabling user-controlled AI systems while preserving data privacy. Within this framework, HAIant establishes a unified AI-assisted biological research workflow through personal biological knowledge base construction, multi-user communication, and bioinformatics tool invocation. To support domain-specific applications, a bioinformatics skill-execution module named BioSkills, consisting of 84 analytical tools and two domain knowledge bases, is integrated into HAIant. As a proof of concept, we show that the Biological HAIant can support experimental record management, bioinformatics analysis, and automated manuscript generation from experimental data. Collectively, out work demonstrates that HAIant and BioSkills provide a scalable architecture for secure, human-oriented AI systems that advance intelligent scientific workflows.","PeriodicalId":19012,"journal":{"name":"Molecular Plant","volume":"10 1","pages":""},"PeriodicalIF":32.7000,"publicationDate":"2026-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Molecular Plant","FirstCategoryId":"99","ListUrlMain":"https://doi.org/10.1016/j.molp.2026.08.011","RegionNum":1,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"BIOCHEMISTRY & MOLECULAR BIOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Artificial intelligence (AI) has achieved human-level capabilities across multiple domains; however, concerns regarding data security, privacy, and domain-specific capability enhancement remain. Here, we present human + AI + ant (HAIant), a human-centered framework for secure and personalized intelligence augmentation based on locally deployed AI. HAIant integrates locally deployed large language models with personal data, individual reasoning processes, and domain-specific tools, enabling user-controlled AI systems while preserving data privacy. Within this framework, HAIant establishes a unified AI-assisted biological research workflow through personal biological knowledge base construction, multi-user communication, and bioinformatics tool invocation. To support domain-specific applications, a bioinformatics skill-execution module named BioSkills, consisting of 84 analytical tools and two domain knowledge bases, is integrated into HAIant. As a proof of concept, we show that the Biological HAIant can support experimental record management, bioinformatics analysis, and automated manuscript generation from experimental data. Collectively, out work demonstrates that HAIant and BioSkills provide a scalable architecture for secure, human-oriented AI systems that advance intelligent scientific workflows.

查看原文本刊更多论文

海蚁（HAIant）：一个以人为中心的AI框架，用于生物研究中安全、个性化的智能增强

人工智能（AI）已经在多个领域实现了人类水平的能力；然而，关于数据安全、隐私和特定于领域的功能增强的关注仍然存在。在这里，我们提出了人类+人工智能+蚂蚁（HAIant），这是一个基于本地部署的人工智能，以人为中心的安全和个性化智能增强框架。HAIant将本地部署的大型语言模型与个人数据、个人推理过程和领域特定工具集成在一起，在保护数据隐私的同时实现用户控制的人工智能系统。在此框架下，海蚂蚁通过构建个人生物知识库、多用户交流、生物信息学工具调用等方式，建立了统一的人工智能辅助生物研究工作流程。为了支持特定领域的应用，HAIant集成了一个名为BioSkills的生物信息学技能执行模块，该模块由84个分析工具和两个领域知识库组成。作为概念验证，我们展示了生物海特可以支持实验记录管理，生物信息学分析和从实验数据自动生成手稿。总的来说，我们的工作表明，haant和BioSkills为安全的、以人为本的人工智能系统提供了一个可扩展的架构，可以推进智能科学工作流程。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Molecular Plant 植物科学-生化与分子生物学

CiteScore

37.60

自引率

2.20%

发文量

1784

审稿时长

1 months

期刊介绍： Molecular Plant is dedicated to serving the plant science community by publishing novel and exciting findings with high significance in plant biology. The journal focuses broadly on cellular biology, physiology, biochemistry, molecular biology, genetics, development, plant-microbe interaction, genomics, bioinformatics, and molecular evolution. Molecular Plant publishes original research articles, reviews, Correspondence, and Spotlights on the most important developments in plant biology.