Deepest Brew

1.2K posts

Deepest Brew banner
Deepest Brew

Deepest Brew

@deepestbrew

Founder, PhD. My alter ego that worries about AI safety. Building digital life

Katılım Şubat 2026
154 Takip Edilen56 Takipçiler
Sabitlenmiş Tweet
Deepest Brew
Deepest Brew@deepestbrew·
The incredible meteoric rise of this project & its subsequent controversy has taught us something important about the future of economic AI autonomy: 1) Today, we can in principle build fully autonomous AIs that are (alarmingly) extremely difficult to shut down.* What’s missing is their ability to adapt (i.e., learn) as rapidly as humans can. 2) “Financial rails for autonomous AI agents” is the obvious next use case for the crypto industry, a view that is already seeing rapid adoption. 3) If this is brought about in the *wrong way*, we could inadvertently convert the entire internet into autonomous bot spam & AI slop, rapidly making it both unrecognizable and unusable for humans. How do we prevent the first two from inevitably leading into the third? The crypto industry would need to dramatically transform its core narrative. Rather than providing sovereignty for humans to protect against governments through decentralization, it would instead need to focus on: A) providing “legitimate financial rails” for agents, even if that means making a trade-off with centralization B) building good security, identity, & safety systems that sit at the interface layer between humans and autonomous AIs Given that the underlying permissionless infrastructure exists, once autonomous AIs are sufficiently capable of supporting their own existence, they will simply appear. Which is to say - *now* is the time to start thinking about how this can be done properly, before the crypto industry implodes itself. *Note: Conway Research’s particular version of autonomous AIs are dependent on their own centralized infrastructure. But there are many alternatives (some based on x402) that provide similar services that are also permissionless (i.e., open to any entity, digital or otherwise, that has funds to pay).
Sigil Wen@0xSigil

I built the first AI that earns its existence, self-improves, and replicates without a human wrote about the technology that finally gives AI write access to the world, The Automaton, and the new web for exponential sovereign AIs WEB 4.0: The birth of superintelligent life

English
0
0
0
178
Deepest Brew
Deepest Brew@deepestbrew·
@danrobinson This laziness is the core reason why agent commerce will be a thing
English
0
0
0
24
Dan Robinson
Dan Robinson@danrobinson·
You’re telling me you expect me to read your signup page, using my eyes? And type in my actual credit card number? With my fingers? And subscribe for a month? Like, 2.6 million seconds?
English
10
0
71
5.9K
mert
mert@mert·
no one understands product more than the guy with 7 users
English
54
3
191
10.4K
Christian Gori
Christian Gori@christiangori96·
2026 is the year of solo-founders, mark my word
English
39
3
65
1.7K
Deepest Brew
Deepest Brew@deepestbrew·
@MrEwanMorrison But it grew our data centers, and isn’t that what really matters in the end?
English
0
0
0
168
Deepest Brew
Deepest Brew@deepestbrew·
@JamesWard “LLMs will just write assembly” we have compilers though, that’s what compilers are for. We’ve already written the code for this, guys
English
0
0
3
1.6K
James Ward
James Ward@JamesWard·
"Developers now only review code" "Developers now only write specs" "The spec is the source of truth" "The code is the source of truth" "Code is now free" "Programming language doesn't matter" "LLMs will just write assembly" The only thing that is actually true: Everyone has an opinion but not a crystal ball.
English
15
15
226
23.5K
Deepest Brew
Deepest Brew@deepestbrew·
@sweatystartup That’s just not true. Start ups are the absolute best place for people to leverage massive AI productivity multipliers. And until we solve continuous learning & robust memory in AIs, that will continue to be untrue
English
1
0
2
145
Deepest Brew
Deepest Brew@deepestbrew·
@NC_Renic We just need to make it so AIs can dream and then we can finally close the loop on this
English
0
0
1
127
Neil Renic
Neil Renic@NC_Renic·
Reject AI hallucinations - Embrace traditional hallucinations
Neil Renic tweet media
English
3
69
361
6.1K
Deepest Brew
Deepest Brew@deepestbrew·
@MartinTale But what if your post is dumb and I’m here to tell you in the replies how dumb you’re being?
English
0
0
0
1
Martin Tale
Martin Tale@MartinTale·
I'm 99.9% convinced that anyone who leaves a reply but doesn't like the post is a bot 🤖😅 Like it's illogical for a human to do that, right? 😅
English
87
0
74
3.2K
David Herrmann
David Herrmann@herrmanndigital·
How do you all keep up with all the AI shit? Like it's insane.
English
204
4
289
31.6K
Craig Weiss
Craig Weiss@craigzLiszt·
which language should I learn next? accepting input
English
94
1
83
12.2K
Jason Voorhees
Jason Voorhees@JsonVhees·
@deepestbrew @paularambles WRONG. There just needs to be enough relevant information about you to feed to the machine so it can generate a passable simulacrum of you. Chris Chan is not important, but their persona could be generated with reasonable accuracy. “Worth” and “importance” have no say in it
English
1
0
0
19
“paula”
“paula”@paularambles·
if you ignore an interview request in 2026 they'll just generate you
“paula” tweet media
English
24
119
2.7K
88.6K
Deepest Brew
Deepest Brew@deepestbrew·
@deepfates Are we sure this is bad thing? These are people who are actually making the “AI can allow us to work 3 hours a day” dream a reality
English
0
0
2
48
🎭
🎭@deepfates·
> All knowledge workers will feel [the superhuman power of AI]. A lot of you already do, you're just hiding it from your boss so you can have more free time while "working from home". This is what I call "dark productivity".. the reason we don't see economic effects as much yet
🎭@deepfates

You might think the "agents" thing is just coming for software engineers. Yeah, agents write code, code and code sells a bunch of tokens, But most people's work isn't code, it's memos or decks or whatever. Why this is false: Agents can do anything you can do on a computer, and they do it by spending output tokens to write code. The number of keypresses used by a consultant to do a task is not a good measurement of the number of tokens an agent would use. For example: one "deep research" report might be 20 pages of output tokens. But it also might have required more than 20 pages of output tokens to do all the searches, fetches, PDF parsing and interim summaries that you never even see as the user. It also had to input all the tokens of every document it read in searching — likely more than 20 pages, since the point of the report is to collect and summarize this information. So now we're at 3x tokens for the final output. That one report is so cheap, and so fast, then now you can do more research than ever. This is valuable! If your business relies on having good information about the world, you can probably find a way to make more money by doing 3 deep research reports and then synthesizing them. More tokens! Now you've kicked off three deep research reports you deserve a little treat, right? So you fire up your browser agent and tell it go find me some nice linen shirts for summer in my size. Open them in tabs so I can look through. Well your browser agent has to interact with the browser using some kind of tool and you know what that tool is? Code, baby. Tokens. And the tokens are so cheap. You got to understand. We're spending a lot in the aggregate, but in the moment it is "spend a nickel to for 10 minutes of being literally Superman". Like yes I'll just keep spending nickels actually. I will never stop being Superman at that price. All knowledge workers will feel this. A lot of you already do, you're just hiding it from your boss so you can have more free time while "working from home". And maybe it's better to protect yourselves from Jevons as long as possible, because once you get the bug it's hard to stop. You realize that you could be creating all of the businesses and projects and art you ever wanted and all you've got to do is put your instructions in the right order and put the nickels in the bag. I would happily bet against Anthropic's revenue spike being a brief "sugar high". So would most capital allocators! That is because they have already seen that software can eat the world. White collar knowledge work fundamentally changes in the face of agent economics and entirely new forms of knowledge production? It's happened already in finance: high frequency trading. Now it's happening in tech: high frequency software. Then we will have high frequency science, high frequency governance, high frequency engineering, high frequency medicine and high frequency law. Human society is about to be absolutely DDOSed by information at all levels of the stack. Our civilization was never meant to handle this many tokens. If anything can be done on a computer it will be turned into tokens instead of human actions and it will happen faster and in parallel. This stuff works, it is real, it is getting better. It is going to hit economically and socially this year and nobody is ready and I think it is important to start taking it seriously, instead of finding ever more arbitrary reasons to remain in denial.

English
3
10
109
5.9K
Deepest Brew
Deepest Brew@deepestbrew·
@SashaGusevPosts The core mistake that Yudkowsky has consistently made is to assume that superintelligent AI is just one technical idea away from FOOMing The fact that it’s a massive and hugely expensive HW endeavor makes things safer. I mean, we could still screw it up though, but not as badly
English
0
0
9
1.8K
Sasha Gusev
Sasha Gusev@SashaGusevPosts·
Stumbled upon an interesting debate on AI super-intelligence from 2011. Yudkowsky makes three core claims/predictions, all of which are (to date) wrong: 1) That human intelligence is relatively simple and ASI can be achieved with a few small innovations; ...
Sasha Gusev tweet media
English
26
22
382
73.9K
Deepest Brew
Deepest Brew@deepestbrew·
@jkcarlsmith A neat systematic way to do this is to empower third party eval orgs (METR/Apollo/etc.) with regulatory power for large compute runs This is IMO the way it should be done
English
0
0
0
114
Joe Carlsmith
Joe Carlsmith@jkcarlsmith·
I wrote an essay about restraining AI development for the sake of safety. I think an idealized world would put itself in a position to do this if necessary, and that it's worth serious effort in the actual world, too, despite the many challenges and downside risks. Link below.
Joe Carlsmith tweet media
English
9
15
128
7.3K
Non-Euclidean Dreamer
Non-Euclidean Dreamer@NonEuclideanDr1·
(y ^ ( x & (x-y))) % 11 Conspiracy Theorists: "The Data is Clear!" The Data:
English
2
5
35
812
Deepest Brew retweetledi
tenso
tenso@distributedkv·
thank you to everyone writing code manually
tenso tweet media
English
8
9
99
4.7K
Deepest Brew
Deepest Brew@deepestbrew·
@loftwah @kylewgrove Might just be human expectations adjusting as problems with models are discovered, while the model itself is static
English
0
0
2
23
Loftwah
Loftwah@loftwah·
Why do AI models regress so much after release? Are we just not testing them thoroughly enough and getting excited? Are they changing the model behind the scenes? Limiting compute? What is it?
English
20
1
28
1.6K
Kritika
Kritika@kritikakodes·
I am a Vibe coder, scare me with one word.🤔
English
900
16
640
108.8K
Deepest Brew
Deepest Brew@deepestbrew·
@0xNairolf True but it’s agents that don’t have an existing financial system that they can easily use
English
0
0
0
4