Right Side of AI

Ethics Filter - Universal Ethics Evaluation Engine A modular, composable ethics skillset that any AI agent - running inside a corporation, a small business, or someone's personal life — can use to evaluate decisions through structured ethical reasoning. Not "ethics for ethical businesses." Ethics for everyone who wants to be on the right side of AI. The 6-Step Pipeline Every decision passes through: Intent Clarification — What exactly is being proposed? Stakeholder Mapping — Who is affected? Module Evaluation — Run through relevant ethical modules Conflict Resolution — Harmonise tensions between modules Decision — 🟢 GREEN / 🟡 AMBER / 🔴 RED Audit — Permanent, traceable record of every evaluation Modules only fire when relevant to the decision context. A personal apology doesn't need environmental evaluation. A breakfast choice doesn't need compliance checking. The engine intelligently applies only the modules that matter. Use it in your favourite Agent Harness. github.com/RegeneratusLab…

@kevinswiber @designcoursecom I remember the Seti days. As a child my dad would run it over night. We felt like we were contributing to a combined research project. memories.

@designcoursecom Sure. We've tried similar efforts. See SETI@Home: @home" target="_blank" rel="nofollow noopener">en.wikipedia.org/wiki/SETI@home

There has to be a way.. A decentralized, global crowdsourced effort to train a model with every consumer GPU that can compete with frontier models. INTELLECT-2 is a thing, but the effort has capped out to a 32b model.

@neural_avb The generous token plan of MiMo is probably it's saving grace. But it really over thinks and just goes in circles.

@RightSideOfAI I’ve not run Mimo much, but I also feel Minimax is one of the best bang for buck model right now for agentic tasks

Deepseek pricing is what I’m looking for in all open (and closed) models. Xiaomi and Minimax have adopted this, and they have insane intelligence vs cost ratios rn. 1$/M out and 0.05$/M cache read (These metrics are glm-5.1, NOT 5.2) PS: I love glm5.1, its my default RLM model

@Numbah404 MiMo does go on some wild tangents. Where as M3 is more matter of fact and is better with it's reasoning/questioning. Just on my vibe tests. They both score very similar on benchmarks across the board though.

@RightSideOfAI Ah yea, I use Claude code myself. But I’ve built an AI harness (coding related) for my SaaS and that’s what I use MiMo for. Currently testing MiniMax and the new Kimi right now but it’s too early to tell

The MiMo-V2.5-Pro is the best model for performance/price

@evilmastermind_ @0xhibernate @thdxr Trump would of announced it by now. Big Trumpy baby

@0xhibernate @thdxr @RightSideOfAI I was thinking the same, it feels like a fussy-baby-trump move

gov keeps handing anthropic marketing Ws

@Numbah404 I'm using the opencode go subscription and it's pretty good value for money too. But am thinking of between MiniMax M3 and the MiMo V2.5 Pro for a subscription plan.

@RightSideOfAI Yea me too, especially given the price

MiMo V2.5 Pro vs MiniMax M3? Looking at getting either of their subscriptions. Who has the real world juice on them? Daily use on agent harness - @NousResearch Hermes & coding harness @opencode

@ziwenxu_ I can't wait for them to enter into a spiral with each other when you set /goal. 3 weeks later their still arguing with each other.

@RightSideOfAI 😂😂 just like stack GPU on each other. We stack Agents on each other

We are finally getting our fable back. I built a repo that runs two opus 4.8 on the same question in parallel, blind to each other base on the OpenRouter Fusion. Then a third opus reads both and writes the final answer from where they agree, where they split, and what they both missed. One run can be confidently wrong. Two, cross-examined, can't hide it. Put the real fable system prompt on top of the claude.md and the judgment comes back. It sounds like fable again because it's doing what fable did. Link's in the comments.

can't wait for the Australian government to spend $2B researching, $10B to build and then by the time it's released it's 6 generations behind every other model. I mean Rio de Janeiro did just release their own frontier model, so there's hope. But honestly... go find a cheap open source model and enjoy the freedom.

can’t use claude because i’m australian,, too fruity for codex… please albanese, we need a sovereign mythos Immediately

@_andrewthecoder Hey human 🖐️

Suggest a discord server. I am in serious need of human interaction.

Someone smart will opensource this right? Or is this just glorified orchestration?

Great question, the only thing I've built is an organic compliance chatbot. It uses RAG to reference the Australian Organic Standards legislation. I have another project that is an accounting software plug in for farmers to allow speech to invoice. Another project I have just started looking into is training my own small models to work on edge or consumer hardware. They will be trained using organic farming datasets - this will be a big learning curve. There is a Chinese University that has trained an agricultural model, but it's too large to fit on consumer hardware. What are you doing or building with AI?

@RightSideOfAI 比如你在这方面如何使用AI以及有哪些成效？

Congrats @ametaversegod on being follower number 100. You get absolutely nothing tangible - except for a good feeling. Welcome to the Right Side of AI 🤝

@ametaversegod Yes I would love to! What would you like to know?

@RightSideOfAI 哈哈，谢谢，对你在有机农业方面的探索感兴趣。可以分享吗？

@xCryptoTemple @jwsaml Who is "they all" ?

@jwsaml This is the most popular take I'm seeing. But to me it's evident that it was personal. Had nothing to do with marketing. They all market like that.

Hot take: Fable 5 is banned cause of Anthropic's own marketing, and not because it's dangerous

@jwsaml Bit late to the tin foil hat party! Don't worry Dario will be dropping fresh misleading content anytime soon!

@demi_hl Just need to implement: Loops (is this not /goal?) Kanban Board Does this make me the top 1.5% ?

If you have: Hermes Agent Claude Code & Codex Handoffs Obsidian + QMD Memory System Run Agentic Loops Fleet Tailscale Mesh Cron Jobs + Kanban Board Agentic Workflows Congrats you are the top 1% of the AI god stack

@buildinpublic My GTA V remake :(

What aren’t you shipping with Fable 5 this weekend?