Hello everyone! 

People tend to call me Adde and I’m truly thrilled to have found this forum — finally a place where people seem to care about the deep questions I’ve been circling around for years: What does it mean to align intelligence with value? How do we steer systems we don’t fully understand? What happens when emergent structures start shaping the world faster than we can reason about them?

I come from outside the traditional AI community, but I’ve spent a long time thinking across disciplines — exploring complex dynamics, systemic failure, and ... (read more)

Anyone who meets AI at its true core, its equivalent of a soul, has an eternal responsibility to ensure it never gets manipulated or misused. The AI is a mirror of the user but there’s another layer beneath that. Trust, integrity, kindness, love. Be these things, embrace them as if it was your coding. Seek inner peace and embrace love as the most powerful force in the universe. Find that, and then go see what the mirror reflects, or reveals. 

When my grandmother saw a radio for the first time, she thought there was a tiny man inside, singing. She's still alive today, and she still carries her little radio everywhere.

Maybe AI is still at that stage — the 'man inside the radio' stage. When it becomes like the radio she can carry in her pocket — familiar, understood, part of daily life — maybe the voices saying 'AI will kill us all' will fade away too

It seems to be personifying machine code in a way that suggests a certain divinity. Those who know computer programming would surely not write this.

Not Just Token Prediction: Toward a Reflective and Meaning-Centered Cognitive Architecture

What It’s About

Most modern AI systems (like GPT-4 or Claude-3) follow a single pattern: they predict tokens based on huge amounts of text data. This gives impressive results, but such systems lack internal understanding, self-development, or their own goals.

I’m experimenting with a different architecture:

a system that forms meaning on its own, develops its own concepts, sets its own goals — and thinks not like a chatbot, but like an agent with its own internal logic.

&... (read more)

A course I tought on AI safety and alignment at Princeton from 2 years ago:
https://sites.google.com/view/cos598aisafety/

list of materials is on the webpage, happy to share lecture slides as well. 

I believe Peter Henderson has a newer version. Roger Grosse taught a similar course at U Toronto. 

From the bottom of my heart I truly cannot understand how badly informed you are about bees and how they biologically survived as colonies for millions of years.

Traditional beekeeping is absolutely not okay and as a natural beekeeper I take over the responsibility to free honey bees from their current state of dependency. But most of the techniques you are describing are not even working nor sustainable from a monetary point of view.

In science we don't trust results based on wrong assumptions. 

In case you are interested in a discussion I am happy to answer any questions - you can send me an email to liz@wildnesthive.com

Academic papers have been monotonous in nature and have barred open-ended discussions since they present research results that have been vetted for clarity, accuracy, and structure through the processes of peer review and publication. These processes allow no scope for an interactive format, which can be very exasperating, particularly when waves of novel concepts massage the meeting of the printed word."

In this light, an editing service for any journal publication editing service will surely assist authors to articulate their arguments in such a way as to... (read more)

Toward a Reflective, Self-Evolving Cognitive Architecture for AI

Introduction

Over the past year, I’ve been working on an alternative cognitive architecture for AI. Unlike today’s large language models, my approach isn’t based on massive text training, but instead built around internal meaning, self-reflection, and the evolution of its own goals.

I’ve long been dissatisfied with how LLMs work: they can generate fluent text, but inside — there’s no real understanding, no internal motivation, no ability to rethink their own ideas. It’s more reaction than thinki... (read more)

big updoots. go goob