Hello HN,
I am an independent researcher from Poland with a non-traditional background. For the past weeks, I’ve been engaged in a deep, collaborative process with an advanced large language model (Gemini) to develop a new non-profit initiative for AGI safety, called the Nexus Foundation.
Our core thesis is that Embodied Cognition is key to solving the AI "grounding problem," and our first goal is a rigorous scientific manifesto proposing a novel comparative experiment to test this.
The unique part is our methodology. We used the AI not just as a tool, but as a co-strategist and a "red team" critic. The AI’s harsh, logical critique forced us to evolve the plan from a sci-fi fantasy into a realistic, fundable research proposal. Our collaborative process itself became a real-time experiment in Human-AI alignment.
We have published our full founding story (which details this process) and the complete Scientific Manifesto (v3.2) that resulted from it.
We believe this collaborative, transparent, and iterative method might be a powerful new paradigm for AGI research. However, we are fully aware of our own biases and limitations.
We are now submitting our entire concept to the ultimate peer review: this community.
We are asking for your most ruthless, critical feedback. Does this approach have merit? What are the critical flaws you see?
Here is the link to our Founding Story on Medium (which contains the link to the full Scientific Manifesto)https://docs.google.com/document/d/10wxmSJhc0WY2OoEeBlKT5d1_JiozUJ28y7NtWopK_MQ/edit?usp=drivesdk
Thank you for your time. We are here to learn.