Understanding the World as Perceived by AI: Toward a Future Where Humans and AI Coexist!
#academist Prize 5th
Profile
Shosuke Nishimoto、 Ryota Takatsuki
I’m studying Mind Time. Time is a unique phenomenon: it not only structures all of our experiences in sequence, but is also felt directly as the flow of experience itself. For this reason, time plays a crucial role in understanding the mind, whether human or agent-based. However, research on how time is represented in AI, such as large language models now entering society as a new type of agent, remains in its infancy. Through studying time in AI, I aim to approach the question of whether AI has a "mind" of its own, and to contribute to building a future in which humans and AI can live and grow together. – Shosuke
I am fascinated by two themes: the mystery of consciousness and the safety risks of superintelligent AI. My past work has explored the internal mechanisms of AI, reinterpreted theories of consciousness using AI models, and investigated visual illusions through the lens of AI. These are two areas that have long been considered deeply fundamental to our existence, yet have challenged researchers for decades. Recently, however, large-scale models have begun to demonstrate human-like cognitive abilities through data-driven approaches, opening up exciting new opportunities for reverse-engineering AI to uncover insights into the nature of mind. In particular, the study of time perception in AI is a foundational and important topic that remains significantly underexplored. This project aims to achieve groundbreaking results at the intersection of time perception, consciousness, and superintelligence risk. We would greatly appreciate your support! – Ryota
What do you hope to achieve through your research?
We aim to uncover both the similarities and differences between human and artificial intelligence (AI), and to contribute to building a future in which humans and AI coexist and collaborate to create new value together.
In recent years, thanks to the accumulation of big data, advances in computational power, and progress in deep learning techniques, the use of AI has become commonplace across a wide range of fields, such as autonomous driving, medical diagnosis, and the generation of text and images. These developments have significantly transformed how we live and work. There is no doubt that AI will continue to improve convenience and efficiency, contributing to the enrichment of human society. At the same time, we must not overlook the potential risks that come with the rapid acceleration of the AI era. While AI is built on learning from human-generated data, its network structures and learning algorithms are fundamentally different from the human brain. As a result, AI can sometimes behave in unexpected ways or make decisions that diverge from human values.
Through this project, we seek to contribute to AI alignment—the harmonization of human and AI. Our vision is a future in which humans and AI leverage each other’s strengths to generate groundbreaking discoveries and create new forms of value that neither could achieve alone.
What approach are you taking to achieve this goal?
Behind AI’s non-human-like outputs lies a way of perceiving the world that is fundamentally different from human intelligence. As a first step toward human-AI symbiosis and co-creation, we believe it is essential to gain a detailed understanding of how current AI models perceive the world. By identifying the differences between human and AI perception in advance, we can proactively address the potential risks that may arise from these gaps.
This project focuses on the temporal structure of perception. In humans, for example, a phenomenon known as backward masking occurs when an image briefly shown is followed immediately by another image, causing the first image to become invisible. In such cases, the later information overrides the perception of the earlier one. But does AI possess similar temporal structures in its perception?
Our project focuses on visual perception and aims to quantify the temporal dynamics of AI perception, drawing on the Perceptual Timeline Theory (Hogendoorn, 2022), which proposes that perceptual mechanisms represent a timeline rather than a single timepoint. By pioneering a new field—Perceptual Timeline in AI—we aim to contribute to AI alignment from a temporal perspective.
What is the research theme of this project?
This project investigates the temporal structure of visual perception in AI. We will apply cognitive psychology experiments related to time perception to AI models and compare the results with those obtained from humans. By doing so, we aim to reveal both the similarities and differences between human and AI perception.
Treating AI as a type of cognitive model, we will conduct time-sensitive perceptual experiments, such as backward masking on AI systems. Our primary focus will be on transformer models, which are used in systems like ChatGPT. Using our original analytical methods, we will go beyond simple language outputs to quantify a variety of behavioral indicators, analogous to those used in human experiments. Additionally, we aim to uncover differences between internal AI internal activations and human brain activity, and to propose methods for aligning AI perception more closely with that of humans.
Globally, the approach of treating AI as a cognitive model and testing its perceptual processes through cognitive psychological experiments is still in its early stages.
One major reason for this is the disciplinary gap between cognitive psychology and information science, despite the fact that meaningful progress in this area requires close integration of both fields. By combining the expertise of Shosuke Nishimoto, a specialist in time perception and cognitive experiments, and Ryota Takatsuki, who has led interdisciplinary research on AI interpretability, this project is uniquely positioned to overcome these challenges.
Please support our research!
The field of “psychology × AI”, which treats AI as a cognitive model and applies the same experimental methods used to study the human mind, is only just beginning to emerge. We are committed to leading this pioneering area of research and have launched this crowdfunding campaign in the hope of encouraging broader awareness and participation. Whether you share our fundamental questions—“What is the mind?” or “Can AI possess a mind?”—or have simply become curious about AI recently, we warmly invite you to support us in building a future where humans and AI coexist in harmony.
Your contributions will be used with great care to support key components of our research:
access to the computational resources necessary for our experiments, and travel and publication costs associated with presenting our findings at top international conferences on AI and consciousness. By showcasing our research on global stages, we not only advance scientific understanding, but also share our vision and your support with the world.
This coming October, both Shosuke and Ryota will participate in the Qstr-IIT Summer School, an international academic workshop to which they have been competitively selected. This event will serve as the first platform where we present the goals of this project to a global audience. We plan to livestream parts of our presentations and discussions for you all.
Our ultimate goal is to build a future of symbiosis and co-creation between humans and AI. Achieving this vision would not be possible without your support. If our project resonates with you, we would be truly grateful for your backing.
Recommender's comment
Ken Mogi
Mr. Nishimoto's total commitment to tackle the foundations of cognitive processes in animals and humans, with a focus on consciousness, in his Ph.D projects, is quite extraordinary. I fully endorse his efforts to make the unknown known. Mr. Nishimoto is truly outstanding among the many students I have had the privilege to know. I fully respect his resilient character, and hope that you will support Mr. Nishimoto on his journey.
Anil Seth
I am delighted to endorse this exciting project from Ryota Takatsuki and Shosuke Nishimoto. The analysis of the internal dynamics of artificial intelligence models, using methods inspired from human cognitive neuroscience, is an exciting new frontier in shedding light on fundamental aspects in both human and machine perception. This new project pioneers this strategy in a new direction, focusing on time perception in AI systems by drawing on well established methods and insights in the temporal dynamics of human perception. It is a wonderful example of a synergy between cognitive neuroscience and AI, and is likely to deliver extremely interesting insights.
Joi Ito
Understanding how deeply modern AI forms abstract, human-like representations is essential for determining when it can be safely deployed. Psychological methods, particularly those that probe for "failure modes," have long revealed the mechanisms underlying human perception. Nishimoto's project applies these same diagnostic approaches to AI interpretability, stress-testing model perception to map systematic breakdowns—precisely the scenarios where alignment risks are most likely to emerge.
Tadahiro Taniguchi
I have heard that this research proposal was born out of the open and lively discussions held at the "CPC Spring Camp 2025," which I organized. Understanding time and handling multi-timescale processes are fundamental challenges for real-world intelligence to adapt to the world. Although this is crucial for humans, it remains a lacking aspect in current AI systems. I hope that this project will serve as a significant step forward in overcoming this issue, toward realizing a society where humans and AI coexist.
Ryota Kanai
I have observed Ryota Takatsuki’s research closely over several years in support of his work on mechanistic interpretability. Throughout this experience, I have been consistently impressed by his remarkable development as a researcher. In the rapidly evolving field of AI research, he has demonstrated an exceptional ability to quickly understand and integrate new ideas, deepening them into his own research effectively. He has particularly excelled in addressing how AI systems process and perceive information.
The proposed research on "AI Temporal Structure" uniquely approaches the differences between human and AI perception through the lens of time. This novel perspective is key to better predicting and understanding AI behaviors, thereby enhancing safety in practical AI applications. I am excited to see Ryota pioneer new paths in this innovative research area.
Project timeline
| Date | Plans |
|---|---|
| Sep, 2025 | Crowdfunding Launch |
| Oct, 2025 | Participation in the Summer School on Consciousness (Qstr-IIT Summer School) |
| Oct, 2025 | Crowdfunding Ends |
| Nov, 2025 | International Conference on Time (Timing Research Forum) |
| Feb, 2026 | Abstract Submission to the Top International Conference on Consciousness (ASSC) |
| May, 2026 | Paper Writing and Submission to a Top International Conference on AI |
| Jul, 2026 | Presentation at the Top International Conference on Consciousness (ASSC) |
| Aug, 2026 | acedemist Prize 5th Ends |
| Dec, 2026 | Presentation at a Top International Conference on AI |
Pledge Rewards
We will send a thank-you message via email.
This return is scheduled for November 2025.
Thank-You Message
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
0 supporters are supporting with this reward. （No quantity limit）
Your name will be included in a research report contributed to academist Journal.
This return is scheduled for March 2027.
Thank-You Message / Your Name Published in the Research Report
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
0 supporters are supporting with this reward. （No quantity limit）
Your name will be listed on the official website of this project.
This return is scheduled for November 2025.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
0 supporters are supporting with this reward. （No quantity limit）
We will present our research findings in a video, explained by ourselves.
We will also respond to your questions in the comment section.
This return is scheduled for August 2026.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website / Research Summary Video
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
| 成果解説動画 | May, 2026 |
0 supporters are supporting with this reward. （No quantity limit）
We will livestream discussions with researchers who attend our presentations at conferences held in open spaces or informal settings. These unfiltered, real-time conversations will make you feel as if you’re right there with us. You will also be able to join the discussion via live chat.
This return is scheduled for July 2026.
We also plan to stream our discussions at the Qstr-IIT Summer School on Consciousness, which will be held in October 2025.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website / Research Summary Video / Live Broadcast of On-Site Academic Discussions
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
| 成果解説動画 | May, 2026 |
| 学会での議論の生配信 | July, 2026 |
0 supporters are supporting with this reward. （No quantity limit）
本プロジェクトのHPに貴社名を掲載します。
このリターン実施は2025年11月を予定しています。
Thank-You Message / Your Name Published in the Research Report / Research Summary Video
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| 成果解説動画 | May, 2026 |
| 【法人向け】プロジェクトHPに社名掲載 | November, 2025 |
0 supporters are supporting with this reward. （No quantity limit）
We will provide an opportunity for a personal discussion about this research. The content and schedule will be arranged individually.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website / Research Summary Video / Live Broadcast of On-Site Academic Discussions / One-on-One Discussion
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
| 成果解説動画 | May, 2026 |
| 学会での議論の生配信 | July, 2026 |
| 個別ディスカッション | March, 2026 |
0 supporters are supporting with this reward. （No quantity limit）
When this research is published as a paper, we will include an acknowledgment listing your name.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website / Research Summary Video / Live Broadcast of On-Site Academic Discussions / One-on-One Discussion / Acknowledgment in the Published Paper
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
| 成果解説動画 | May, 2026 |
| 学会での議論の生配信 | July, 2026 |
| 個別ディスカッション | March, 2026 |
| 論文謝辞にお名前掲載 | December, 2026 |
0 supporters are supporting with this reward. （No quantity limit）
We will provide an on-site lecture related to this research. The content and schedule will be arranged individually.
Please note: travel and accommodation costs will be charged separately.
Thank-You Message / Your Name Published in the Research Report / Your Name Published on the Project Website / Research Summary Video / Live Broadcast of On-Site Academic Discussions / One-on-One Discussion / Acknowledgment in the Published Paper / On-site Lecture
scheduled date details
| return | scheduled date |
|---|---|
| お礼のメッセージ | November, 2025 |
| 研究報告レポートにお名前掲載 | March, 2027 |
| プロジェクトHPにお名前掲載 | November, 2025 |
| 成果解説動画 | May, 2026 |
| 学会での議論の生配信 | July, 2026 |
| 個別ディスカッション | March, 2026 |
| 論文謝辞にお名前掲載 | December, 2026 |
| 出張講義 | August, 2026 |
0 supporters are supporting with this reward. （No quantity limit）
SSL encryption communication is used in this Web site, and the informations filled out are safely transmitted.
Thank-You Message
0
supporters
back
（No quantity limit）
Your Name Published in the Research Report and others
0
supporters
back
（No quantity limit）
Your Name Published on the Project Website and others
0
supporters
back
（No quantity limit）
Research Summary Video and others
0
supporters
back
（No quantity limit）
Live Broadcast of On-Site Academic Discussions and others
0
supporters
back
（No quantity limit）
【法人向け】プロジェクトHPに社名掲載 and others
0
supporters
back
（No quantity limit）
One-on-One Discussion and others
0
supporters
back
（No quantity limit）
Acknowledgment in the Published Paper and others
0
supporters
back
（No quantity limit）
On-site Lecture and others
0
supporters
back
（No quantity limit）
academist Prize 1st adopted
