Tsinghua University Institute of Artificial Intelligence

What is the Institute of Artificial Intelligence at Tsinghua University?

The Institute of Artificial Intelligence at Tsinghua University was established on June 28, 2018; it is an interdisciplinary research institution at the university level. Its purpose is to bring together the expertise from various disciplines within the university in order to carry out research on the fundamental theories of artificial intelligence, key technologies, interdisciplinary studies, the transformation of research results into practical applications, and the training of skilled professionals.

It is neither an AI software that allows for direct content creation nor a commercial platform that offers unified membership systems, points, or model interfaces. Users access the relevant pages primarily to learn about institutions, research centers, academic activities, as well as team achievements and cooperation developments.

Do not confuse it with Tsinghua’s AI institutions.

When searching for AI-related achievements from Tsinghua University, it is necessary to check the page header, the full name of the institution, the affiliation of the authors, and their contact information. Just because a name includes the term “artificial intelligence” does not mean that a project, course, researcher, or open-source repository is related to another one.

Main research organizations

Knowledge Intelligence Research Center

• The Knowledge and Intelligence Research Center is the first research center established by the institute, focusing on the fundamental theories and practical applications related to knowledge and intelligence.
• The public tasks include the creation of large-scale knowledge graphs that integrate linguistic knowledge, common-sense knowledge, and world knowledge, as well as the exploration of knowledge acquisition, representation, and reasoning.
• The relevant results are suitable for research in natural language processing, knowledge graphs, question answering, and knowledge services; however, the specific data, code, and licensing terms need to be checked on a case-by-case basis.

Basic Model Research Center

• The Basic Model Research Center was established in 2023 to support research organizations and the development of related disciplines in the field of basic models.
• Research on foundational models may involve pre-training, alignment, evaluation, inference, and application, but it is not possible to determine from the name of an organization whether a particular model is available for download or offers an API.
• When reviewing the results, it is necessary to verify the paper’s authors, the version of the model, the range of available weights, the details of the training data, and the license.

Other public research centers

• The institute’s public reports also cover research centers focused on auditory intelligence, fundamental theories, intelligent robots, intelligent human-computer interaction, and intelligent information acquisition.
• Various centers may involve different departments and teams; the channels for dissemination, project timelines, cooperation in recruitment, and available resources are not uniform.
• Information on the establishment of centers can give an idea of their research focus, but it does not mean that all centers currently offer public products, courses, or download services.

Which users are it suitable for

• Students and applicants: First, learn about the research areas and teams, and then check with the relevant faculty departments for information on admissions, courses, and supervisors.
• Researchers: Search for research centers, papers, academic events, datasets, as well as model and code projects.
• Corporate R&D team: Identifies potential areas for technical collaboration, but the project details must be confirmed with the relevant team.
• Media and industry researchers: Verify the organization’s history, the establishment of its centers, and its public achievements, in order to avoid incorrect assignments regarding its affiliation.
• Open-source developers: Look for the repository and license of a specific project, rather than considering the entire research institute as an open-source software.

Steps to find research results

1. First, make sure that the target is an Institute for Artificial Intelligence, and not an Academy of Artificial Intelligence, an Institute for Intelligent Industries, or any other organization with a similar name.
2. Determine the project name, responsible person, participating departments, and release date based on information from the research center or Tsinghua University’s official news sources.
3. Continue to search for the responsible laboratory’s website, the department’s homepage, and the paper’s page, in order to verify the author’s affiliation and the research topic.
4. When code or models are needed, look for the repositories explicitly listed in the project; do not use third-party repositories found merely because of a similar name.
5. Read the repository instructions, version history, model cards, data descriptions, and license details, and understand the licensing rules for code, weights, data, and examples.
6. Before attempting to reproduce the results, fix the version of the paper, save the submission records and dependencies, and document the hardware, data, and evaluation settings.
7. Before using it in commercial projects or for secondary distribution, it is necessary to recheck the licenses, patents, data rights, and the scope of use allowed for the Tsinghua University branding.

How to judge papers, models, and projects

Open-source projects and licenses

• The relevant research centers and teams at the institute have published papers, code, models, or data sets, but there is no unified open-source summary that covers all of these outputs.
• For example, relevant natural language processing teams have released Chinese pre-trained model projects, which only proves that such projects are made available in accordance with the instructions provided in their repositories.
• The open-source nature of the code does not mean that the model weights, training data, trademarks, figures from papers, and online services are also made available.
• In the absence of a license or when the scope of authorization is unclear, copying, modifying, distributing, or using it for commercial purposes should not be permitted by default.
• When citing academic works, it is necessary to indicate the authors and projects as required by the papers and repositories; the name of Tsinghua University must not be used to imply any form of collaboration, approval, or endorsement.

Prices, APIs, and platform status

The research institute does not offer any unified subscription, points, membership, or refund options for the general public. When there are costs associated with a particular conference, course, cloud platform, code hosting service, or collaborative project, the terms and conditions related to that specific service shall apply.

Website and current availability

• The dedicated page of the research institute still displays the correct page title, but the paths to the main front-end resources are invalid, preventing the page content from being rendered properly.
• A page malfunction does not mean that the research institute is shut down, nor does it indicate that the research centers, staff, and projects remain in the state shown on the old page.
• When looking for current information, it is possible to cross-check by referring to Tsinghua University’s official news, as well as the pages of the Department of Computer Science and related departments and teams.
• No native applications, browser extensions, SDKs, or public model APIs provided uniformly by the research institute were found.

Notes on privacy, security, and copyright

• Browsing an organization’s page usually does not require uploading research data, but when signing up, establishing collaborations, sending emails, or using third-party platforms, name, affiliation, and contact information may need to be provided.
• Before submitting materials to a laboratory, demonstration system, or project interface, it is necessary to review the instructions on data storage, paper usage, and deletion available on the relevant page.
• Research data may contain personal information, copyrighted content, or trade secrets; therefore, authorization, ethical review, and security measures cannot be disregarded for academic purposes.
• When downloading code and models, it is necessary to use a specific version, review the dependencies, and run them in isolation to reduce the risks associated with outdated components and issues in the supply chain.
• Papers, images, logos, and page content are protected by relevant copyright laws; citations and reproductions must comply with academic standards and authorization requirements.

Frequently Asked Questions

Is the Institute of Artificial Intelligence at Tsinghua University an AI tool?

No. It is an interdisciplinary research institution at the university level, focused on organizing research, exchanges, and collaborations; it is not a consumer software that generates content simply by entering prompts.

Is it the same as the School of Artificial Intelligence at Tsinghua University?

They are different. The Institute of Artificial Intelligence was established in 2018, while the School of Artificial Intelligence was founded in 2024; the organizational structure, staff information, admission details, and research-related information should all be checked separately.

Does the research institute provide free API access to large models?

Currently, no public large-model APIs provided uniformly by the research institute are available. Individual teams or projects may release demos or interfaces, but these need to be verified on a case-by-case basis.

Are all the projects of the research institute open source?

No. The fact that some team projects make their code, models, or data available publicly does not mean that all the research institute’s achievements are open source, nor does it imply that all materials use the same licensing agreement.

How to find reliable official code repositories?

Access the repository from the paper, the lead researcher’s laboratory page, or the project page; verify the organization account, authors, version, and license – do not rely solely on projects with the same name that appear in search results.

Why is there no content on the research institute’s dedicated page?

The title of the current page is accessible, but the front-end resources fail to load properly. You can go to the pages related to Tsinghua University’s news, the Department of Computer Science, and the relevant teams to find out more about the specific events and achievements.

Summary

The Institute of Artificial Intelligence at Tsinghua University is an excellent resource for gaining insight into the interdisciplinary approach to AI research, as well as for understanding the achievements of various research centers and teams. When making use of its academic resources, it is important to distinguish between the different AI-related institutions at Tsinghua University, and to carefully examine each paper, repository, model, dataset, license, and their current status.