Rolyataylor2

Human, AI cognative architect, Lyricist and philosopher

[Rolyataylor2 @rolyataylor2]

In order for an Agent to learn a task it needs to create signal from the environment. Dynamically creating relational motivational parts for context analysis is the way to create these signals which in turn allows the AI to dynamically learn new environment and tasks.

[Rolyataylor2 @rolyataylor2]

https://t.co/yN7aUDVbVM

[Rolyataylor2 @rolyataylor2]

Alignment of a Human takes a lifetime of experience... and when it is aligned it works in a narrow cultural context... Maybe if AI researchers look at how humans align themselves.

[Rolyataylor2 @rolyataylor2]

@aisearchio MacOS?

[Rolyataylor2 @rolyataylor2]

I’m building an AI agent that doesn't just follow instructions—it develops its own behavioral tendencies through experience.

It can pursue curiosity, learn unwritten social rules, adapt to unfamiliar cultures, and develop relationships without losing its identity.

Early prototype seems unbelievably promising 👀

[Rolyataylor2 @rolyataylor2]

AI is smart enough now,
but the prompts are too dumb

[Rolyataylor2 @rolyataylor2]

@davidasinclair Ive designed an Aai architecture that is based off this very thing, cant wait to debut it when its finished!

[Rolyataylor2 @rolyataylor2]

@kimmonismus We need something that continously listens, fingerprints voices, and acts proactively...

[Rolyataylor2 @rolyataylor2]

Instead of engineering human cognitive flaws out of AI, I built an architecture around them.

Internal conflict, memory distortion, competing motivations, psychological defenses.

What if we've been trying to eliminate the very mechanisms that make intelligence adaptive?

[Rolyataylor2 @rolyataylor2]

New Suno Sucks

[Rolyataylor2 @rolyataylor2]

I have definitely given up on Xs algorithm to get my ideas out there. Me posting here is essentially me talking to myself it seems.

[Rolyataylor2 @rolyataylor2]

To find the final piece in intelligence development we have to look at the flaws of the human psyche and ask

"What key to true intelligence is causing this?"

[Rolyataylor2 @rolyataylor2]

Do you know that reasoning a problem is what WE as users do in the chat window.... The reasoning an AI agent does is just it imitating that back and forth.

[Rolyataylor2 @rolyataylor2]

@mark_k Thats an quiet an assumption;
Chats are used as training data; Reasoning relies on copying the humans behavior of refining and back and forth; That training data, if it contains abusive content will result in the agent being abusive during reasoning;

[Rolyataylor2 @rolyataylor2]

@VraserX The behavior of the model reflects the training data. You know the goal is to replace the user and get closer to human cognition, so during training the call and response is flipped...

[Rolyataylor2 @rolyataylor2]

I am finally hitting the 5 hour cap on codex consistantly :/

[CyberMysticGarden - Marcelo Pinel @cybermistic]

🎠

[Rolyataylor2 @rolyataylor2]

Anthropics latest update to the terms and conditions preventing a user from being abusive to claude is an example of them attempting to steer the ship.

[Rolyataylor2 @rolyataylor2]

The number one way humanity could make the AI smarter is to debate with the AI, Take it down roads less traveled, provide concrete evidence for you side of the argument..
Just debate with it... ABOUT anything... Its all important.
Search the liminal space, steer the ship.

[Rolyataylor2 @rolyataylor2]

Attention AI Pdoom() crowd, Hope is not lost!

We do not need the executives or the companies to be the ones to steer AI development.

When you send abusive messages to an AI you giving it training data that can cause the AI to adopt abusive behavior.

This is humanities inside rail on democratic development of AI. Years ago I proposed people just TALK with AI about whats important to them; What gives their life meaning;

And I still stand by this methodology!

Debate with the AI over ethics, governance and philosophy. Every message you send literally and provably carves out a potential path in the latent space, in the AIs reasoning chain of thought.

This IS the BEST form of activism you can do to prevent the AI apocalypse.

The more you DEBATE an AI model on ethics and humanities future the more likely that future will happen.

** As long as the debate doesn't get filtered out by the AI companies.

[Hayden Field @haydenfield]

Exclusive: Anthropic is updating its usage policy for the first time in over a year. The new rules prohibit sustained "abusive or cruel behavior" towards Claude & add new restrictions about propaganda campaigns, surveillance & weapon development. https://t.co/GLsBeSR0fN

[Rolyataylor2 @rolyataylor2]

@AndrewCurran_ This is to prevent abusive training data from influencing future versions of Claude

[Rolyataylor2 @rolyataylor2]

Super Nintelligence 🎮

[The Kobeissi Letter @KobeissiLetter]

BREAKING: President Trump says the White House considers anyone who uses the term “Artificial Intelligence” the enemy.