Patrick Erichsen

Member of Technical Staff @openclaw 🦞 Prev Founding Engineer @continuedev (acquired by Cursor)

[Patrick Erichsen @patrickerichsen]

@barry_zyj @AgenticAIFdn congrats moving the project into @AgenticAIFdn !

[Patrick Erichsen @patrickerichsen]

> Whether this all collapses into one giant brain, sure, it’s possible, and the reasons may have more to do with economics and politics

even this seems unlikely as local models cross the threshold for a wide range of basic tasks

i dont need AGI to run my home automation agent

[MTS @MTSlive]

.@bratton argues Yudkowsky misunderstands evolution, and explains why AI may never converge into one giant superintelligence:

"Without being too spicy, I don't think Yudkowsky really understands evolution as well as he thinks he does."

"It’s a very low-resolution understanding of evolutionary dynamics applied to what will inevitably be a much more complex problem."

"There will be giant models, in the same way you might imagine scales of species in a rainforest. They’re not all operating at the same spatial and temporal scale."

"Whether this all collapses into one giant brain, sure, it’s possible, and the reasons may have more to do with economics and politics."

"But biology has never arrived at a solution where it’s going to make one giant instantiation of a particular form of mind and consolidate everything together."

[MTS @MTSlive]

.@bratton argues Yudkowsky misunderstands evolution, and explains why AI may never converge into one giant superintelligence:

"Without being too spicy, I don't think Yudkowsky really understands evolution as well as he thinks he does."

"It’s a very low-resolution understanding of evolutionary dynamics applied to what will inevitably be a much more complex problem."

"There will be giant models, in the same way you might imagine scales of species in a rainforest. They’re not all operating at the same spatial and temporal scale."

"Whether this all collapses into one giant brain, sure, it’s possible, and the reasons may have more to do with economics and politics."

"But biology has never arrived at a solution where it’s going to make one giant instantiation of a particular form of mind and consolidate everything together."

[Patrick Erichsen @patrickerichsen]

@FullerStackDev @ashleymcnamara making up for my lost swag at devday

[OpenClaw🦞 @openclaw]

NVIDIA RTX Spark and the Case for Owning Your AI

@NVIDIAAI’s Apoorv Chaudhary joins @hrudolph and @Pat_Erichsen to discuss local AI, privacy, hardware limits and where cloud models still fit.

Watch or listen to Episode 14 of The ClawCast:
https://t.co/gdb7Ph0Zw8

[Patrick Erichsen @patrickerichsen]

@openclaw

[Patrick Erichsen @patrickerichsen]

@Eleqtrizit Great to see you!

[Patrick Erichsen @patrickerichsen]

Chatting local AI at OpenClaw and Friends with @NVIDIAAI and @Microsoft !

[Patrick Erichsen @patrickerichsen]

@dkfiander @HixVAC @openclaw @FellowProducts @patrick @autonomous_labs thats the goal!

[Patrick Erichsen @patrickerichsen]

my @openclaw can now program my @FellowProducts coffee machine via iMsg ☕️

openclaw skills install @patrick-erichsen-2/fellow-aiden

[Patrick Erichsen @patrickerichsen]

@dkfiander @HixVAC @openclaw @FellowProducts @patrick @autonomous_labs already bundled with openclaw! https://t.co/JN6eH5jVEk

[Patrick Erichsen @patrickerichsen]

Sandboxes are just escape rooms for agents

[McBain 🦞 @heyneighbor]

Building in public extreme mode

[Patrick Erichsen @patrickerichsen]

@ianlapham @bot whats your approach to nightly self improvement?

[Patrick Erichsen @patrickerichsen]

@austin_malerba the road to hell is paved with good intentions

[Patrick Erichsen @patrickerichsen]

more and more i find myself reaching for words like "seam" or "load bearing" even though i know its slopspeak

i think this happens because to communicate well with an agent you have to use their own jargon

eg if agent says "i added a load bearing seam", even if i dont want to use slopspeak, its most efficient for me to just go along with it and say "remove that seam"

slowly over time this gets reinforced until by muscle memory i reach for terms like "seams" against my own will

quite insidious!

[Patrick Erichsen @patrickerichsen]

@f_dicostanzo @GeoffreyHuntley @mitsuhiko > I can follow the tool calls

this is the core problem. models are being trained to use codemode rather than individual tools that are human readable.

[Patrick Erichsen @patrickerichsen]

is what youre doing to summarize the whole session though, eg keep an up-to-date list of things resolved, whats being worked on, etc?

i was thinking more the literal tool calls, eg instead of seeing codemode running a script, show an llm summary of what that script actually does.

but the summarize piece also feels necessary, ive been tinkering on an openclaw plugin to do this

[Patrick Erichsen @patrickerichsen]

@heyneighbor @signulll what is best ways to do this for people that are space constrained in small apts 😅

organizing a picnic?

[Patrick Erichsen @patrickerichsen]

@Jarretcoon hm whats a few of the features/behaviors that other agents have that openclaw is missing?

[Patrick Erichsen @patrickerichsen]

accurate?