Expand "Humanity Tomorrow" website with a multilingual AI Risks & Mitigations Map connecting risks, causal pathways, mitigation strategies, remaining gaps, and project ideas, while maintaining the existing AI Risks FAQ.
Expand "Humanity Tomorrow" website with a multilingual AI Risks & Mitigations Map connecting risks, causal pathways, mitigation strategies, remaining gaps, and project ideas, while maintaining the existing AI Risks FAQ.
About:
Humanity Tomorrow (https://humanitytomorrow.site/) is a multilingual educational project focused on the existential and catastrophic risks of artificial intelligence. The website also features content on other topics (such as life extension), but funding is being sought specifically for content on AI risks.
The initial funding request was to create a comprehensive FAQ on AI risks that is accessible to non-experts. Since submitting the application, I have completed the first version of this FAQ on my own: it now contains 25 key questions and counterarguments, an explanation of the fundamentals of AI risk, and a separate section on what individuals can do.
The FAQ still requires refinement, review, and minor corrections, but the first version is already complete and published.
The project will be primarily in English, with Ukrainian and Russian versions of the main materials. To disseminate the results, I will also use my existing YouTube channel (https://www.youtube.com/@markelkori) on AI safety, which has over 3,700 subscribers. The most popular video, based on the AI-2027 article, has garnered over 75,000 views. So there is already an audience that will be interested in this website.
AI Risks & Mitigations Map:
The next phase of the project is the creation of the AI Risks & Mitigations Map: a structured and interactive map that will be regularly updated, linking various sources of risk from advanced AI, specific mechanisms by which disasters occur, existing and proposed mitigation measures, and remaining security gaps.
The basic logic behind the map is as follows:
source of risk → mechanism of occurrence → potential harm → mitigation measures → remaining gaps, ideas for projects.
For example, the section on biological risks might link:
the growth of AI’s biological capabilities → facilitating the development of dangerous agents → relatively easy physical access to synthesis → potential dissemination.
with corresponding measures such as screening DNA synthesis.
A similar structure will be applied to other areas.
In this way, the user will be able to follow this path:
What risk exists? → How could it materialize? → What would need to happen for that to occur? → What is already being done to prevent it? → Which parts of the chain are still virtually unprotected?
On the other hand, while the AI FAQ already exists, it is not a document that is set in stone.
The landscape of AI safety is changing quite rapidly, so some examples, assessments of model capabilities, and references quickly become outdated. During the grant period, I will:
- correct factual errors;
- update outdated information;
- improve the reasoning;
- add important counterarguments if I find any gaps;
Why I Am Capable of Doing This:
The project already exists, and the first major portion of the originally proposed work was completed before funding was secured.
On my own, I have: created a website, written a comprehensive AI Risks FAQ, researched and organized numerous arguments regarding AI risk, maintained versions in several languages, regularly create educational materials about AI, have attracted an audience of several thousand people to AI-risk content.
This funding will allow me to devote significantly more time specifically to researching, structuring, and verifying the material, rather than working on the project only in my spare time outside of my main job.
There is a vast amount of information available on existential and catastrophic AI risks, but it is extremely fragmented.
Research on alignment, model evaluations, cybersecurity, biosecurity, governance, autonomous weapons, compute governance, and other areas often exists almost entirely in isolation from one another. It is difficult for a newcomer to even grasp the full scope of the problems, let alone understand what solutions exist for each of them.
Humanity Tomorrow seeks to reduce this information barrier in two ways:
1. Making arguments about AI risk understandable
The FAQ provides an accessible entry point for people who are not yet convinced that catastrophic AI risk deserves serious attention.
Instead of assuming that the reader already accepts the core ideas of AI safety, the resource systematically addresses specific objections.
2. Making the landscape of risks and solutions comprehensible
The AI Risks & Mitigations Map addresses another problem: even if someone already considers AI risk important, it is difficult for them to see all at once the full range of threats, the causal relationships between them, and the existing strategies for addressing these problems.
Team Member
$3,500–$5,000 — Project development: my time spent researching AI risks and mitigation strategies, creating a taxonomy and map, writing explanations, verifying sources, and regularly updating and refining the existing FAQ.
$500–$800 — Website and technical implementation: hosting, web services, development tools (AI subscriptions).
$400–$700 — Editing, external expert review, and usability and readability testing.
$300–$500 — Distribution and promotion of the resource.
No comments yet. Be the first to share your thoughts.
The map is designed to reduce the time required to gain this overview.
Why Multilingualism Can Offer Added Value
Most high-quality content on AI safety is available in English. Although this FAQ is primarily intended for an English-speaking audience, I’m translating it into my native languages: Ukrainian and Russian. Eastern Europe is home to a large number of technical specialists and engineers, yet there is a particularly scarce supply of Ukrainian- and Russian-language materials on catastrophic AI risk.
I already have an audience of several thousand Ukrainian speakers and can share these ideas with them without having to build an audience from scratch.
