Thinking about

How do we measure, study and reason about the safety and security of many agents interacting, rather than one agent in isolation?

This workshop from the Institute for Decentralized AI presented the latest academic research into this area. The future is going to be populated with imperfect agents that will need to be able to coordinate with untrusted others. As activity starts to depend on long chains of agentic action, we need to ensure that each edge in the network is secure enough that the probability of failure across the entire chain is within acceptable bounds. This is critical if agentic economies of scale are to emerge.

It was fascinating to see how ideas from cryptography are being applied to this problem domain to reason about the limits of possibility. This is currently a very nascent field, where the rules and assumptions underpinning the discipline are still being discovered, defined and challenged. The bottleneck holding back the space is figuring out which of those assumptions to challenge and which to base our systems around.

Other topics included:

• How do we resolve the tension between stochastic AI agents and deterministic smart contracts, and get the best of both worlds — the open-ended intelligence of the one, the bounded guarantees of the other?
• What can we safely compartmentalise, decompose, and delegate down the chain?

Interesting papers I will add to my to read pile:

• Scalable AI Safety via Doubly Efficient Debate
• Multi-Agent Risks from Advanced AI

The cultivated track is trace-making – how an organization produces the patterns of coordination that a new medium makes possible, the equivalent of the desire lines worn across a park, the kits and practices that no central planner could have specified in advance. The governed track is trace-selection – how an organization reads those emerging patterns and decides which to reinforce into shared infrastructure and which to let fade.

A governed track with nothing underneath it standardizes too early. It freezes a layout before new traces have formed, the way laggard manufacturers kept the line shaft. A cultivated track with no governance generates patterns endlessly but never consolidates them, accumulating risk it cannot see. It is frontier territory with no ecosystem to sustain and learn from it. The maturity ladder is really a measure of how well an organization coordinates these two tracks as the ground keeps shifting.

Very eloquently put!

We are indeed rearchitecting the digital environment within which knowledge work takes place.

A good reminder of the importance and the interplay between bottom up exploration and top down structure. The desire lines get worn by people moving through the space; the work is reading which paths to pave and which to let the grass grow back over.

Definitely worth checking out this thought provoking guide from Rafa and Sachin: https://ai.protocolized.dev/#cover.

They may be portals: structures that increase the number of routes through reality and thereby expand the space of possible becomings.

This is wonderful. Puts words on something I have been exploring for a while now.

My favourite kind of stigmergic environment is the public surfaces that reveal the hidden layers of the urban environment.

The flyposting for the funky event space. The sticker bombing that reveals those who were once there. The graffiti and the billboarding spreading messages of hope, joy and resistance.

I am especially interested in portals located within the physical environment into digital spaces. I think there is a huge possibility space to explore here around layered cryptographic capabilities. The QR code barely scratches the surface.

We should all have the capability to leave stigmergic breadcrumbs throughout our built environment for our future selves and those who follow in our footsteps.

As multiple people work on the same latent space of problems, we pull forward the same sticky ideas along the same gradients in the latent space, ending with roughly the same ideas.

Nice synthesis!

It made me think of LLM’s as an instrument and the music we make with them as another analogy with similar meanings.

We are all playing the same instrument (more or less), we bring our context to tuning the ideas it produces, but the instrument only has so many notes. Not surprising we play similar tunes.

I like this because music demonstrates the rich generative potential of combinatorial composition, especially when in harmony with collaborators.

It also tells us it would be great if we could design some alternative instruments, attuned to and trained on different frequencies and bodies of knowledge.

During one the same week I produced a submission for the Long Now Labs Book of Time, I also listened to Bayo Akomalafe’s talk The Untimely and attended Derya Karli’s workshop on Quantum AI.

Both the Untimely and Quantum AI were touching on the same ideas. The inadequacy of our binary, classical, simplified understandings of reality to capture its beauty, humanity and complexity.

The space between the tick and the tock of a clock.

The humanity between the lash of the whip or the call of the bell.

The richness between the 0’s and the 1’s of a digital dataset.

We are realising that the world is not all that it would appear to be.

It is more than the binary order and structure we impose apon it.

How can we attune ourselves to our quantum reality?

How can we inhabit this reality?

What becomes possible when we do?