For people building AI and the future: human-centred AI and ethics researchers, climate and AI-for-Earth scientists, technologists and longtermists.
Build
with us
Aligned to what?
Every system is aligned to something: engagement, growth, efficiency, a benchmark. Most of the hard questions in AI and climate come back to what that target should be, and who decides.
Madhyasth Darshan names four human goals: resolution, prosperity, fearlessness and coexistence. We write them as clarity, enough, trust and harmony, from the individual to the family, society and all of nature. That is a target concrete enough to write down, and to test. Read the essay: What should AI be aligned to? → · Read the thesis →
Who this is for
- Human-centred AI researchers, asking what systems should be aligned to. Read →
- AI ethics advocates, looking for a standard that goes beyond avoiding harm. See the spec →
- Climate AI scientists, working on the demand side that efficiency alone does not settle. Read →
- AI for Good and AI for Earth innovators, building tools meant to leave people and nature better off.
- Technologists who want what they build to help people understand, not only to keep them busy. Read →
- Longtermists, asking what a good future looks like generations from now. Read →
All essays written for builders →
Open problems
These are the questions we most want to work on. Each one connects to something we are already building.
- How do you measure understanding in a person, not just knowledge?
- Can a written statement of what matters guide AI agents while people keep the final say?
- How do you test whether a conversation ends in understanding, rather than in more engagement?
- How can “enough” be built into climate models of demand?
- What does a measure of wellbeing at family level look like?
- How do you build order upward from families, resting on understanding rather than control?
All open problems, with what we know so far →
How we work together
- Propose. Tell us which problem you want to work on, or a better one, and what you bring to it.
- Talk. A conversation first, to see whether it fits and what each side needs.
- Work in the open. What we learn together is published here, as essays, data or changes to a page, with your permission.
Who you’d work with
A team of AI engineers, researchers and educators, building Dhee, CoSeeker, K4M2 AI and Earth & Us. We don’t put individual names on the work; everything here is written as all of us together.