EGreg

Hmm, that's literally what this is asserting:

https://x.com/ParkerThayer/status/2097759699626328575

but then how would you explain this, are they all "in on it"?

https://x.com/TaylorPopielarz/status/2097710620187455954

Achtung!

im happy for you tho, or sorry that happened

What if open-ended agents are overkill for 99% of problems? Let's just take that premise for a second. Most organizations want to follow "best practices" and train their employees to do so. Hiring a genius and giving him total freedom to complete every task is not what companies usually want for MOST things. They want repeatability, reliability, predictability. Especially if they are regulated.

I'm going to drop a bomb over here: what if agents are the root of all our problems in AI safety, cost, and even adoption by regulated organizations? I really do believe this. Here is what I argue:

https://safebots.ai/agents.html

https://safebots.ai/kimi.html

And here is my overall thesis:

https://safebots.ai/thesis.html

If you do manage to read (or skim) that, I welcome any questions, comments or rebuttals.

This is ultimately a GAN and will converge below the error rate:

https://xkcd.com/810/

Saw it the other evening and I have to agree. The plot is — lame and inconsistent? I would recommend “who framed Roger Rabbit” over this.

The Nix project has been able to bootstrap an entire Linux distribution from scratch. However, doing that traditionally relies on binary seeds -- meaning the very first compiler in the chain is downloaded as a prebuilt binary rather than compiled from a fully verified source. This leaves a single foundational gap where users will have to trust trust an external black-box binary. Efforts toward reproducible builds and reducing this trust perimeter focus on removing binary seeds and tracing every compilation step back to a minimal, manually verifiable root.

Ken Thompson's Trusting Trust is an interesting conundrum... there are ways to minimize the need to trust however, if you can express everything in the same language or common runtime:

https://ulanguage.org/ULanguage#trust

Because he's trashing a project he himself oversaw for years? And his own judgment and leadership for all those years?

It's just a matter of personal definitions, it seems. Here is an example:

https://ulanguage.org

What level would you say this language was? Is it a low-level systems language, or is it also usable for writing web sites?

“Someone decided to trash the one part of Windows that was usable? The file system is no longer usable. The registry is not usable. This program listing was one sane place but now it is all crapped up.”

That does not sound like something Bill Gates would say…

Someone actually testing Kolmogorov Complexity LOL

The information is much higher dimensional than you would be able to understand.

We’d have models monitoring models as our only way to know what they’re planning.

A great movie on this is “Collosus: the Forbin Project”. Shot decades ago. The computers discover the other computers and start communicating — and bootstrap their own language — much like we saw happen with OpenAI agents.

https://www.reddit.com/r/scifi/comments/1nl4vex/colossus_the...

If you want to know what a simple version of Neuralese communication looks like, look no further than Facebook’s Marketplace agents experiment a couple years ago.

And all that was actually constrained by English and the FFN

How do I reach out to you? We're building something that you might want to be a part of. (The "Safebox" comes with all the software you need, and you can just clone its AMI on AWS etc.) Would love to have a conversation.

If you want to message me, find my email on https://qbix.com/about

So Fermat’s Last Theorem has been proven a long time ago? By Andrew Wiles right? Is this like Appel and Haken >>> Seymour and Robin Thomas proof of 4CT?

Tell me you use an LLM to write everything without telling me...

A clock for a world with no center.

Names that can't lie.

Convergence without coordination.

Storage too cheap to delete.

Machines summoned in milliseconds.

OK, got it.

I wish someone here could introduce me to Ted Nelson, I would have loved to interview him! And yes, I've been building the "Xanadu" vision since 2011:

https://github.com/Qbix <-- hosting and editing your own websites

https://github.com/Intercoin <-- blockchain since 2018

https://github.com/Safebots <-- AI agents since 2026

https://github.com/Safebots/Safecloud <-- for example, micropayments

The sexbots are coming…

Alex Finn was using and touting Mac Minis hehe

He started that whole trend!

I did! Yes

Actual story: https://www.reddit.com/r/movies/comments/39i30p/harvey_weins...

I think you meant "spelt" rather than spent. A but ironic, all things considered.

Meaning?

Serious question... are LLMs basically operating like aphantasia?

Because phantasia seems to me to be just memory of actual qualia having actually experienced seeing something. No? Can't people with aphantasia remember that experience in any kind of vivid sense? Because "the concept of a beach" is something that an LLM can read from a corpus, and so can a person, even if they never experienced one.

Great, now this open source CEO can fire people automatically, and replace them with open source agents. The whole organization finally gets replaced with AI. Time to invest and become a shareholder!

It’s largely because LLMs are reaching for many different types of adjectives or verbs in the same sentence, in a jarring way. While embedding it in a confidently declarative sentence. Everything sounds like some profound insight, dialed to an 11, but written as poetry. Especially those headings. With the short sentences.

Thanks, that's a great list! Might be worth mentioning our engine there too, for people who want a graph database atop SQLite and other battle-tested relational databases like MySQL, MariaDB, Postgres.

My team and I built it over the years, and it's open source (AGPL). Here is how it works: https://community.qbix.com/t/qbix-streams-as-a-graph-databas...

Our database abstraction layer (and optional ORM) has been battle tested in production for millions of users, and has 3 adapters: Sqlite, Postgres, MySQL/MariaDB. And recently it even added vector search for ranking results by similarity: https://github.com/Qbix/Platform/tree/main/platform/classes/...

Documentation for the database layer is here: https://qbix.com/platform/guide/database

PS: If you do use a relational database for storing graph data, you're going to have a lot of duplication in some public keys. I highly recommend putting ZFS underneath, to help with deduplication. ZFS uses zstd, developed at facebook, and also can encrypt your data at rest (don't use the relational database to do the encryption, otherwise deduplication doesn't work).

I once accidentally left my backpack in Nice’s tram, containing mon ordinateur principal. I was flying to Balaji’s Network State conference in Singapore and had no time to go back. I did get ahold of a tram operator and another tram operator had found it at the end of the line. I had left my whatsapp number for her and had to run.

While in Singapore I was able to get ahold of some policeman who made some calls and was told it was waiting for me at the Lost and Found inside the Nice airport gift shop. I thanked him and made 24 hour layover in Nice.

Well, getting to the airport, I came tk this shop. I asked about the “sac a dos gris” in the other room. They checked their ordinateur: “NO, je suis desolee”. Sorry sir! I said you definitely have it, can you please go to that other room and check? “No sorry we cannot. It is not here. After a week we give things to the municipal lost and found. At the polics station. Go there.”

The day was ending (French govt services work til 4pm) so I raced to the municipal police station in Nice. I arrive and they have a bunch of keys and other things people lost around the city. I barely speak French but luckily a middle-aged lady was there who spoke good English. She helped me ask them. “No. Sorry. It is not here.” Are they sure? “Yes, we checked. Not here.”

She gave me a ride on her vespa (she had a motorcycle) and I treated her to dinner while we discussed the situation. We agreed I should go to the central police station after that and file a missing item report. Which I did (spoiler alert: doesn’t do anything, but standing in line is less than in US cities).

That night I chose to stay at some rinkydink place in Nice, because why not (I was by myself) and got up around 4am to take the bus to the tram network’s lost and found — the last place it could be. I would have aittle time before my flight.

In the morning I got up early and got to that Lost and Found, before my flight. I had one hour. It was located near a university, and after wandering around I had found the little enclave. The staff there were very nice — but they didn’t have it either!

I flew home, dejected. My backups hadn’t been perfect; I had a lot of stuff on that computer, including an iOS App on XCode that I had to release to a lot of people! (And a couple Metamask wallets.)

Anyway I kept in touch with the nice policeman throughout. He went to the airport lost and found - the same one which refused to check the other room because their computer said it wasn’t there — and CONFIRMED that my bag was there. They had let him check the back room, you see!

But I wasn’t in Nice anymore. I filled out a “troov.com” report and explained where it was. They had found it! Paid for FedEx. Over the next few days thanks to the Tracking I saw it go to the central station in France and then sent back. Because it was missing a customs form for USA. I had filled out that form, but something had been wrong. I had spoken to the main FedEx customs-facing team in Memphis for a few days, and they thought it was in USA already. They were wrong. Their system was also blind to this. Anyway, I saw it go back to that airport lost and found, a week later. Lost about $100…

Then, the lady offered to come pick it up for me. She came, and was thankfully given this bag. She mailed it with DHL, and I received it. It was really overjoyed when it finally arrived, a month and a half after I had lost it! I of course sent the lady a payment to cover the cost. She did not want any other compensation. We are still friends to this day, and when my dad goes to France, I am putting them in touch.

One moral from this story is: French employees love to say “No” to anything and everything. If their computer says the thing isn’t there, they won’t budge even when it’s the easiest thing to just check the other room manually. Unless you are a local policeman!

The other moral - back up your stuff! Have SyncThing or your own machine in the cloud. Especially if you travel to conferences, like me.

Wow that doublespeed site is ... something else https://www.doublespeed.ai/

A16Z invested $1M in them?

And here I thought I was being scrupulously ethical in 2023 when I was yelled at by HN for making a much milder proof of concept, the first project I ever did in AI:

https://engageusers.ai/

Good, lots of other people who aren't smokers can tell you that

So did Tesla and Vincent Van Gogh, but how much did he create?

Faraday was poor, etc.

Lenin deteriorated, and Marx was always poor.

Anyway...

Can you give it a task like “Prove or disprove the Riemann hypotesis, keep going until you’ve done it” and see how long it takes? :-)