IrvZinter

AI stuff.

All tweets are personal musings and are never the official, unofficial, explicit, or implicit position of any organization.

[IrvZinter @IrvZinter]

I feel like a ditch digger earning an honest living with a shovel. But the backhoe is being invented in labs, and researchers are leaking knowledge of its existence.

Should I keep digging by hand or spend my time thinking about how to use the machine when the time comes?

[IrvZinter @IrvZinter]

@emollick Thank you for this. Btw, my wife liked the picture on the cover of your book enough that it is one of the very few she would let me leave laying around the house because she uses it for her interior design

[IrvZinter @IrvZinter]

Seems like we should identify our personal EA agents as AI for people until AGI is actually here. I don’t mind talking to an ai to schedule things, but it would be nice to know in cases like this so that I can add some human context to conversation, if necessary.

[Steve Yegge @Steve_Yegge]

So my buddy Jim email-intros me to his buddy Jordan. Jordan's an angel investor. Jim says we have a lot in common and should chat. OK, fine, Jim, I'll meet Jordan.

Jordan replies and says "Ernie can help schedule it", cc'ing his EA, Ernie. I send Ernie my schedule availability for the following week. I tell him Monday is mostly open, Tuesday afternoon, and parts of Wednesday.

I check back, and Ernie has replied saying, "Mr. Jordan has a very busy schedule, but he can see you on Monday at 4pm in Madrona at XYZ coffee place." I'm thinking, man, 4pm Monday is going to be brutal bridge traffic. Plus all the schools are letting out, it's gonna be snarled. But fine, whatever, if he's that busy, I'll leave early.

I tell Ernie, fine.

So Monday rolls around and I do the 20-minute drive in 45 minutes, as expected. I meet Jordan at the coffee place, and we decide to walk around the neighborhood.

We get our drinks, start the walk, and our conversation goes like this:

Me: So I see you've got yourself an EA! Or Chief of Staff?

Jordan: Oh that's an AI.

Me (internally): Mother fucker. I am going to wring Ernie's fucking neck.

Me (out loud): Oh, that's nice.

Jordan: It's a piece of shit, doesn't work very well.

Me (internally): God DAMN it. How did I fail this Turing Test so badly? And how do I get one of these fuckin' things? I want one that says, "Well Mr. Yegge is phenomenally busy and can only see you at 3am in the graveyard!" and have it fight it out with Ernie.

Me (out loud): That's cool. Well the drive over from Kirkland was pretty tough.

Jordan: Oh you're in Kirkland? Jeez, if I'd have known I'd have met you halfway.

Anyway that's my story. Fuck you, Ernie. Fuuuuuuuuck yooooooooou.

[IrvZinter @IrvZinter]

The panopticon is coming for us all

[Brian Roemmele @BrianRoemmele]

The slow creep of technology as monitor always starts for “good reasons” it rapidly becomes for ALL reasons.

If you hear yourself agreeing think it out for a decade.

[IrvZinter @IrvZinter]

@Andercot This kind of seems like a given, even without UFOs right? The infinite potential of our current physics doesn’t necessarily mean that it is not also bounded in some way that is a product of our social structures

[IrvZinter @IrvZinter]

@agupta @garrytan Would be nice to have a proper social science. Seems like that would reframe many things we struggle with in a helpful way.

[IrvZinter @IrvZinter]

Remarkable to me how much working with an ai agent feels like mentoring a young professional

[IrvZinter @IrvZinter]

@8teAPi I strongly suspect many of the “lurkers” are likely more influential in their real life than you may realize. The professional/personal cost of a strong online presence can be quite high in certain corners of society, regardless of the banality of the content.

[AI Engineer @aiDotEngineer]

"You cannot govern a technology you have only been briefed on."

Singapore Minister for Foreign Affairs, Dr. @VivianBala, echoing @karpathy and @yacineMTB on why he runs NanoClaw: "you can outsource memory and computation, but you cannot outsource your understanding"

https://t.co/z4Aidf89ha

He also shared his tech stack for running his second brain for Singapore's Foreign Affairs Ministry and parliamentary affairs:
- @AnthropicAI Claude Agent SDK
- Baileys + WhatsApp
- Mnemon (Graph Memory)
- @ollama + @nomic_ai
- @ggerganov Whisper.cpp + OneCLI

With special notes on how he handles security and isolation, and what implications he sees for Singapore Inc.

[Brett Caughran @FundamentEdge]

A big pivot from Ken Griffin on AI:

“Number one is, in the last few months, there has been a step change in the productivity of the AI toolkit. It is profoundly more powerful than it was just nine months ago.

And for us at Citadel, that has allowed us to unleash a much broader array of use cases for AI. And it has been really interesting to watch, to be blunt, work that we would usually do with people with masters and PhDs in finance over the course of weeks or months being done by AI agents over the course of hours or days.

These are not these are not mid-tier white collar jobs. These are like extraordinarily high skilled jobs being, I'm going to pick a word, automated by agentic AI. And I gotta tell you, I went home one Friday actually fairly depressed by this because you could just see how this was going to have such a dramatic impact on society.

When you witness it in your own four walls, when you see work that used to be man years of work being done in days or weeks, it's like, wow, like that's the first time I've seen real impact in our four walls.”

This echoes my own experience with agents and the conversations I am having with students, friends & clients. The toolkit has dramatically transformed and it feels like in finance, for the first time, AI is real.

[Perry E. Metzger @perrymetzger]

Most people’s model for security bugs is wrong. People have acted as though there is always an infinite number, because humans couldn’t find them faster than they could create them, but the number has always been finite. AI systems aren’t going to find every single one immediately, but they are draining the available supply fast, and new ones aren’t going to be created as quickly. Once we start using AI to improve the engineering, the supply is going to drop a great deal indeed, and operations that rely on a steady supply of them are going to have to find other ways to work.

[IrvZinter @IrvZinter]

At what point will it no longer be acceptable for customer service to say “due to staff shortages, your wait will be longer than normal.”? It’s been 5 years since Covid…this is the normal wait time now. your organization just doesn’t value customer service enough to staff it

[IrvZinter @IrvZinter]

Interesting to me how people tend to compress the likely long term trajectory of a system into predictions that sound sensational because they are so different the current state.

Most of today’s jobs will go away. Eventually people will work less, if at all, if you define work as the stuff we do for money now.

These transitions will occur over decades as norms and expectations shift over time, even if the technologies that enable the transformation arrive over months, because socio-technical systems have many factors that influence how they change over time. Technological artifacts (if you consider software an artifact) are just one factor among many, and the degree to which they determine the direction and speed of change in the system is a well studied question. While it would be incorrect to claim artifacts play no role, it is clear that other factors are at least as influential, if not more influential.

I expect work and jobs to be more similar than different for years, eve while there is massive disruption and new requirements placed on the workforce as we all adjust to the new technological environment.

[Sam Altman @sama]

@tyler_johnston many current jobs will go away

i think we will find a lot of new ones, though they may look very different

[IrvZinter @IrvZinter]

Is there some upper limit to intelligence that humans can create that is roughly as smart as our smartest humans? That would be a fascinating development

[Haider. @haider1]

Nick Bostrom says what surprised him most is this extended era of roughly human-level AI

It has already lasted 3-5 years and may stretch further

Instead of an instant takeoff, we are living through an extended phase where AI feels alien, but still close enough to us to feel familiar

[IrvZinter @IrvZinter]

Tesla’s HW5 is going to be so much more than just a better way to build self driving cars.

I bet that a HW5 Tesla pulling compute revenue while it’s sitting in the garage, will make more money than a top of the line btc rig on a car you would own anyway for other purposes. And in 3 years, you would still have a car you can use instead of an obsolete asic machine barely worth its scrap value.

@elonmusk is clearly building a huge part of the physical infrastructure that will transition humanity to the AI age with @Tesla and @xai, but in a way that allows normal people to realistically have a chance to directly benefit financially as well.

[IrvZinter @IrvZinter]

@Tangerine284 @curtis_yarvin That is what I had seen as well. That they are trying to push liability to the user by verifying the user is an adult who can make decisions. Would be interested to hear if this is somebody else doing it

[IrvZinter @IrvZinter]

This is going to limit AI adoption in useful situations more than anything else. Over reliance on an engineer’s understanding of reality is going to create systems that fail to live up to their expectations and people responsible for important systems will avoid or slow adoption because “all this AI stuff is hype.”

[IrvZinter @IrvZinter]

@elonmusk This was always the right choice. The modern world is designed for people to see with their eyes. A machine that would navigate roads in that world without requiring expensive and unlikely changes to it would have to receive inputs the same way.

[IrvZinter @IrvZinter]

@daisyldixon @FranklinMatija Feels a little late to me