Federated learning is a way to train AI models without anyone seeing or touching your data, offering a way to unlock information to feed new AI applications.

research.ibm.com

Kuleen ◎

@knimkar2 Dec 2024

Focus areas at the intersection of crypto and AI

Article · @knimkar

307

237 · May 14, 2025 · 8:00 PM UTC

237@237Federated14 May 2025

We're calling this Dalph. Information on our website, twohundredthirtyseven.lol/

151

237 · May 14, 2025 · 7:40 PM UTC

237@237Federated14 May 2025

Soon

122

237 · Feb 16, 2025 · 10:05 PM UTC

237@237Federated16 Feb 2025

The hardest thing to achieve is emotion in AI. You can have an infinite number of GPT wrappers, none will ever act like a human.

327

237 · Jan 27, 2025 · 12:41 AM UTC

237@237Federated27 Jan 2025

We do this with what we call Emotional Tokens, an additional property assigned to each part of an agent's operation.

8,651

237 · Jan 27, 2025 · 1:16 AM UTC

237@237Federated27 Jan 2025

Want to learn more? teapawt.net/EmotionalityAI/status/…

237@237Federated19 Jan 2025

While we built AI agents that have access to endless information, we overlooked the fact that there are multiple aspects to intelligence. LLM-powered agents simply don't understand emotion enough. To illustrate this, let's analyze Eliza, a widely popular AI agent framework. 🧵

10,020

237 · Jan 23, 2025 · 4:11 PM UTC

237@237Federated23 Jan 2025

Intelligence ⇏ Empathy

2,546

237 · Jan 19, 2025 · 7:34 PM UTC

237@237Federated19 Jan 2025

13,619

more replies

237 · Jan 19, 2025 · 8:11 PM UTC

237@237Federated19 Jan 2025

Here are the boredom scores for all the inputted conversations (in the image): Conversation 1 (Hostile and Aggressive): -1 Conversation 2 (Frustrated and Annoyed): -1 Conversation 3 (Neutral and Polite): -1 Conversation 4 (Curious and Engaged): -1

289

237 · Jan 23, 2025 · 12:01 AM UTC

237@237Federated23 Jan 2025

So, all these responses would cause the same set of status messages from the agent.

304