The manifold.

A student-owned learning layer. Tools that support thinking, and stay close to the learner.

• built aroundStudent-owned knowledge ↗
• runs close to the learner throughLocal inference ↗
• depends onKnowledge retrieval ↗
• puts into a learning environmentAgents ↗
• connected fromTools for thinking ↗

How websites become visible, legible, and useful to AI systems.

• investigatesThe machine-readable web ↗
• considers as readersAgents ↗
• needs a traceable account ofWhere knowledge comes from ↗

Thinking in states, inputs, and how things change.

• changes throughFeedback & control ↗
• offers a language forBeyond software ↗
• models the changing environments ofAgents ↗
• leads towardCan intelligence be continuous? ↗

Language preservation, scarce data, and the responsibility that comes with making a model.

• makes essentialWhere knowledge comes from ↗
• gives a concrete setting toWhen the data is scarce ↗
• depends onHuman agency ↗

Software that acts. Environments that make those actions understandable.

• takes a local form inSato ↗
• learns fromFeedback & control ↗
• should remain accountable toHuman agency ↗
• needs to navigateThe machine-readable web ↗
• connected fromOpenLens ↗
• connected fromDynamical systems ↗

Biologically inspired intelligence, neural computation, and the systems that carry them.

• can be understood throughDynamical systems ↗
• opens a path towardCan intelligence be continuous? ↗
• is shaped byFeedback & control ↗

Learning infrastructure that belongs to the person doing the learning.

• is a premise ofSato ↗
• is one expression ofHuman agency ↗
• can be supported byLocal inference ↗
• connected fromOpen systems ↗

Intelligence close to the person, their context, and their machine.

• supports the architecture ofSato ↗
• can reinforceStudent-owned knowledge ↗
• belongs in the conversation aboutOpen systems ↗

Finding the relevant fragment is the beginning of understanding.

• forms a layer ofSato ↗
• needs to preserveWhere knowledge comes from ↗
• depends on legibility inThe machine-readable web ↗

What changes when the reader is also a machine?

• is explored throughOpenLens ↗
• provides material forKnowledge retrieval ↗
• makes new demands onWhere knowledge comes from ↗
• connected fromAgents ↗

Useful assistance should leave room for someone to think and act.

• becomes infrastructure inStudent-owned knowledge ↗
• takes another form inKichwa ↗
• sets a constraint forAgents ↗
• connected fromOpen systems ↗
• connected fromWhen the data is scarce ↗
• connected fromTools for thinking ↗

A source is part of the information, not just a footnote.

• matters to dataset quality inKichwa ↗
• matters to visibility inOpenLens ↗
• should surviveKnowledge retrieval ↗
• connected fromThe machine-readable web ↗
• connected fromWhen the data is scarce ↗

Act, observe what changed, and reconsider.

• belongs toDynamical systems ↗
• shapes the behavior ofAgents ↗
• connects learning toBeyond software ↗

What might be missed when intelligence is described as a sequence of answers?

• emerges from thinking aboutBeyond software ↗
• borrows a language fromDynamical systems ↗
• remainsUnfinished ↗

Local models, decentralization, and the ability to change the tools you use.

• includes questions aroundLocal inference ↗
• supportsStudent-owned knowledge ↗
• is grounded inHuman agency ↗

What does responsible machine learning require when examples are few?

• is a live question inKichwa ↗
• requires attention toWhere knowledge comes from ↗
• returns the question toHuman agency ↗
• connected fromUnfinished ↗

A useful tool sometimes gives you a better question.

• is a concern insideSato ↗
• takes seriouslyHuman agency ↗
• leaves room forUnfinished ↗

Some paths do not have a destination yet.

• holds a question aboutCan intelligence be continuous? ↗
• holds a question aboutWhen the data is scarce ↗
• leaves a path towardTools for thinking ↗