The Human-Centered AI (HCAI) division at Toyota Research Institute (TRI) is excited to announce a collaborative research opportunity focused on fundamentally reshaping systems where humans and AI collaborate, grow, and co-evolve together. One of TRI’s core principles is to develop AI-powered technologies that amplify human abilities, not replace them. We also draw on the philosophy of continuous improvement, kaizen, for humans to continually grow while using AI systems.

We envision a future where people are prepared with the abilities and flexibility to contribute meaningfully in a world where AI is broadly integrated into work and life. Current trends show people adopting AI to delegate and automate a diversity of tasks. As people use AI, there is an opportunity for these systems to help people gain the cognitive flexibility and critical skills in more abstract tasks to remain relevant and thrive in a world where previously core tasks are automated away. 

For example, a worker who does routine data entry can use AI tools to automate more and more steps of data entry, and at the same time, the AI system also helps the worker learn how to reason about whether the data is valid in different contexts and exercise human judgment to ensure data quality. For a graphic designer who begins to use an AI system to easily blend designs together, the system intentionally exposes alternate designs for the designer to build their own aesthetic intelligence, useful for designing in new scenarios. A sustainability committee can first use AI systems to automate the tracking of emissions, and the AI system simulates future scenarios with different interventions for emissions reduction so that the committee can proactively set effective policy for societal good.

The goal of this research program is to uncover a deep understanding of how to design and evaluate systems in which both humans and AI can grow, trust, and adapt to each other over repeated interactions over time. 

We aim to address the following challenge: 

“How can we design AI systems that not just automate tasks, but also 1) adapt to people’s needs and develop to better AI capabilities, and 2) simultaneously help people grow in abilities, decision-making under uncertainty, and metacognitive skills to remain relevant?”

AI Systems Co-Evolving with People

Successful collaborations often evolve as rapport is built and the needs of the collaborators change. Likewise, to build successful patterns of human-AI collaboration, AI systems should not be static but rather reveal capabilities over repeated interactions to adaptively meet people’s needs, and at times even add just enough complexity, to incentivize human learning and growth. People using AI systems are not static, as their preferences, expectations, knowledge, assumptions, growth, etc., can change over time with repeated AI use. 

AI systems can also adapt how they prompt people to ethically disclose personal information necessary for different tasks. AI capabilities can also rapidly change and be upgraded behind the scenes, creating a sense of uncertainty or instability for human collaborators. There is an opportunity to better understand and boldly transform the design of AI systems to co-evolve with humans over time. 

People Co-Evolving with AI Systems

In addition to transforming the capabilities of AI systems to meet people’s task needs, there is an opportunity for AI systems to build durable, transferable skills in people as a side effect of the collaboration. This enables people to continuously improve (kaizen) within themselves and adapt to a changing world. For example, AI systems can help people learn more abstract skills that require human judgment, such as task decomposition, build better intuitions of how iterative prompts impact outputs, cultivate strategic thinking, and perform perspective taking to support creative problem solving. Often, AI systems are designed to generate outputs as quickly as possible, reducing the chance for users to be reflective and critical of the interaction. 

There is an opportunity for designing interactions that foster learning, reflection, and metacognitive control, while maintaining productivity. Systems can also enable non-linear growth pathways that are not based on traditional pedagogy but foment exploration, social interconnectedness, emotional maturity, playfulness, and creativity to emphasize human flourishing. These human learning effects over time are often not well-understood, but could be essential factors for helping people maintain a sense of confidence and relevance while collaborating with AI systems. Applying principles of behavioral science or learning science will likely be important for achieving these goals. 

Human-AI collaboration can be beneficial in many different application domains. Even though proposals can focus on one application domain, a broader goal of this research program is to synthesize best practices for design, implementation, and evaluation across different projects. 

Application domains of particular interest include: 

• Forecasting the societal impact of systemic issues (e.g., climate change, AI use, workforce shifts, political conflict) and estimating the effects of prosocial behaviors and interventions
• Helping people find their role in the future scenario of widespread and deep AI usage in society
• Pro-environmental behavior change
• Enabling and empowering career evolution while proactively fostering wellbeing
• Workplace productivity and decision making
• Designing products and services with enduring taste and appeal, instead of temporary engagement
• Participatory design for groups, for the design of places, policy, communities, or organizations
• Other domains where AI systems have (or will have) an opportunity to adapt to changing user needs and also amplify human abilities to maintain relevance in a changing world 

Regardless of application domains, we are looking for proposals that take a high-risk, high-reward approach to help us prepare for large-scale, even speculative, societal changes. 

We acknowledge that human-AI collaboration can also have negative effects and pitfalls, including:

• Cognitive dependency, deskilling
• Creative homogenization and monoculture
• Labor displacement, knowledge work replacement
• AI dark patterns, manipulation, sycophancy

and thus proposed approaches should carefully consider their potential for inducing and/or mitigating these negative effects.

The Human-AI Kaizen Initiative at TRI aims to build a critical mass of research effort in this topic area. Even though project teams will be funded independently, there will be opportunities at the program level to learn from other teams and codify best practices around the design, implementation, and evaluation of AI systems that co-evolve with people and support upskilling and learning. Project-level goals for funded proposals are to make novel contributions, including, for example, technical capabilities, interaction paradigms, and human modeling methods that are validated through empirical evaluations with representative end users. We encourage submissions of ambitious, blue-sky contributions that aspire to support human growth and relevance by developing human-centered AI systems.

Potential contribution types may include, but are not limited to:

• Systems with or for AI
• Socio-technical systems or approaches for human growth
• Novel modeling techniques
• Design guidelines
• Evaluation methods and metrics
• Empirical findings with policy recommendations
• Novel datasets

Research output may take the form (non-exclusively) of:

• Publications
• Cross-institution report-outs
• Collaborative workshops and/or symposia
• Patents and Intellectual Property 

The award will take the form of an industry-sponsored research agreement.

The proposal will describe the time required to complete the proposed work, up to 3 years. 

The proposer will structure the proposal as a base period of one (1) year, with two (2) option periods of one (1) year each. For example:  

• Base: One year (1 April 2026 – 31 March 2027)
• Option 1: One year (1 April 2027 – 31 March 2028)
• Option 2: One year (1 April 2028 – 31 March 2029)

The proposal will describe the funding required to complete the proposed work. TRI expects that the requested staffing costs will fully fund the research team, and non-TRI funding will not be used to fund staff working on the project without prior TRI approval. 

TRI anticipates a typical funding profile of approximately $250,000/yr to $350,000/yr, depending on the scope of the research project and size of the research team. For example, 2 PIs and 2 graduate students. 

Eligibility

• The proposer must be employed at a North American educational institution, herein referred to as the Principal Investigator (PI).
• There may be additional university co-PIs, if needed. See Cross-Disciplinary Approaches.

Cross-Disciplinary Approaches 

We encourage proposals combining cross-disciplinary expertise in novel, exploratory ways that are currently difficult to facilitate with existing research programs. Possible examples may include, but are not limited to:

• A joint proposal combining a machine-learning researcher and an anthropologist to develop evaluation tools for culturally pluralistic models;
• A computer science PI forecasting generative AI evolution with the support of a historian;
• An art and design institute is funding an engineering student to build AI agents for creativity.

Download the proposal template. 

To register your interest and receive program updates via email, please provide your contact information in this form.

To submit your proposal application, please use the submission form. This submission form requires you to sign in to a Google account. If you would prefer not to sign into a Google account, please contact us at hcai-kaizen-initiative@tri.global and we will find an alternate submission path for you.  

Proposals will be evaluated using the following criteria listed in descending order of importance:

1. Alignment with Topic. The research must clearly and convincingly contribute to both the topic described in the Call for Proposal and TRI’s mission of creating capabilities that improve the human condition; this connection should be explicit and foundational to the project’s goals, not incidental or assumed.
2. Overall Merit and Potential for Revolutionary Advance. The proposed technical approach is ambitious and innovative, yet feasible and achievable within the proposed timeline. Projects must have the potential to fundamentally reshape scientific understanding or technical capability—advancing the field in ways that current approaches cannot. Proposals should aim to redefine what’s possible, not simply make incremental change.
3. Robustness and Real-World Relevance. The proposed research demonstrates a clear path beyond the academic lab, with consideration for variability, scalability, and practical deployment. Proposals should demonstrate awareness of real-world constraints and include plans to test, validate, or apply results in settings relevant to TRI’s mission.
4. Proposer Team. The proposed team has the expertise and experience to accomplish the proposed tasks. Evaluation will also consider the diversity of thought, background, and perspective within the team, and how that diversity contributes to innovative approaches and well-rounded and creative problem-solving.
5. Collaboration and Responsiveness. The proposer demonstrates strong communication skills and a collaborative mindset. Evaluation will consider the proposer’s ability to build a productive working relationship with the TRI researcher, including a willingness to adjust direction based on joint insight and evolving priorities.

To submit questions about this program, please send an email to hcai-kaizen-initiative@tri.global

Proposal Process FAQs

Q: How is this initiative different from another TRI program, the University 3.0 Joint TRI-University Research (J-TUR)?

A: This initiative is a specific call from the Human-Centered AI division at TRI focused on this topic within Human-AI collaboration, whereas J-TUR is for different collaborations with all divisions at TRI. Unlike J-TUR, it is an open call to all universities and does not require a joint proposal with a co-PI from TRI, and can be somewhat higher risk and higher reward than J-TUR. Similar to the J-TUR program, selected proposals will conduct joint research closely with a TRI co-investigator, matched during the review process. 

Multiple universities will be selected to address this topic from different perspectives. Even though project teams will be funded independently, there will be opportunities at the program level to learn from other teams and codify best practices around the design, implementation, and evaluation of AI systems that co-evolve with people and support upskilling and learning  

Q: What is the expected timeframe for the research?

A: Projects can last from 1-3 years and should be proposed as a single base year, followed by an option for another year, followed by an option for a third year. 

Q: Can I submit a proposal to both the J-TUR program and this initiative? 

A: Yes, you can as long as you disclose this in the application. A single proposal will not be doubly funded across programs.

Q: What will we be looking for in proposals?

A: See the Evaluation Criteria

Q: Can there be multiple faculty members as PIs?

A: Yes, one of them must be designated as the Lead PI.

Finances FAQs

Q: Is a TRI-funded project the same as a grant?

A: No, TRI-funded projects are not grants. Funding will be provided through an industry-sponsored research agreement. All funding is contingent upon your university executing a mutually agreed upon sponsored research agreement within a reasonable time. Selected proposals will become joint projects that researchers from both TRI and the university will work together collaboratively. It will be an active collaboration where TRI researcher(s) contribute their technical expertise, help direct the work to meet TRI needs, and facilitate tech transfer (if applicable) to TRI.

Q: Are budget and budget justification information required as part of the submission? Would this count toward the page limit? What about other supporting information, such as a CV?

A: A budget justification is required in the application form. If other supplementary information, such as a CV, are appended, they do not count against the page limit, but please note that the priority for review will be the research proposal.

Q: Is there a limit on allowable indirect costs? 

A: We will honor your university’s standard indirect cost rate. No special reductions or waivers are requested, just your institution’s typical terms.

Q: Are any costs disallowed, for example equipment/computers?

A: Costs related to e.g. equipment purchases are not disallowed, but please detail them if they are to be included.

Q: How many proposals does the TRI’s HCAI Division expect to fund in this initiative? 

A: Depending on the size of the proposals and our budget situation, we expect to fund up to 5 proposals. 

Q. How will intellectual property be handled? 

A: A Sponsored Research Agreement (“SRA”) between TRI and the university governs the handling of intellectual property.

Q: Will TRI allow funded projects to publish their work?

A: Yes. TRI expects and strongly encourages its university partners to publish their work. The SRAdescribes the process to be followed to notify TRI about the preparation of publications.

Q: What are the guidelines for university partners taking funding from another company for work similar to the proposal? 

A: The scope of the TRI funded research should be distinct and separate from other research contracts you may have. You should not use other funds in your TRI funded project, particularly if there may be overlapping or conflicting intellectual property obligations.

• The award will take the form of an industry-sponsored research agreement. Applicants who are selected will be sent TRI’s industry-sponsored research agreement, and funding will be provided only upon your institution’s acceptance of TRI’s terms and conditions. 
• The proposal will describe the time required to complete the proposed work.  
• Funding for the entire proposed duration of the project is not guaranteed. Projects that do not meet their proposed Year One milestones will need to justify their continuation for Year Two. Historically, TRI has acted to discontinue projects based on non-performance. TRI expects that it is the nature of high-risk research with challenging objectives that some approaches will be found to be less fruitful than anticipated, and the project may need to change approaches or directions. 
• TRI conducts its own research and your submission may be similar or identical to research already being conducted or already invented by TRI or other third parties. TRI retains all rights to fund, research, or develop technologies and projects which may be similar or identical to your submission.
• Your submission imposes no obligations on TRI or any of its affiliates or related companies.
• By making a submission, you warrant that you have the right and authority to submit the information in your submission.