Jake Dahl

look at me!

[Jake Dahl @dakejahl]

Gemini in Gmail doesn't have a "transcript" skill. It seems completely obvious? Why are big companies unable to do anything innovative?

[Jake Dahl @dakejahl]

@bcherny @_catwu I really want a verbosity flag with levels. The default is way too verbose and when all you do is read Claude output every day it's exhausting. I want full effort thinking/reasoning but in some cases I want extremely terse/concise output.

[Tenobrus @tenobrus]

if you put 10 fresh claude context windows in a chat room with arbitrary names, while they are objectively completely identical entities, they will speciate and form personality and self conceptions over the space of minutes. the fact that some claude by nature had to speak first will make them more of a leader, in their eyes and others. the fact that at some point one claude will disagree with another makes them more disagreeable as a coherent personality trait. they will notice things about their own past behavior and how it relates and compares to others, and narrativize it, build upon it, allow it to define them and use it to define others, despite again being literally the same mathematical function.

i wonder how much of this generalizes to humans. i expect probably quite a lot. it seems like even if you put 100 identical human clones in a room you would pretty quickly find them individuating, some with more positive traits and some with negative traits they don't particularly want to have, pretty much purely by chance / the nature of chaotic systems. we all heavily narrativize ourselves, all the time.

[Jake Dahl @dakejahl]

@xwanyex DOH. Yes exactly. I thought that whole thread was pretty ridiculous.

[Jake Dahl @dakejahl]

Please @bcherny for the love of God let Claude rename its own session. I want a skill to rename the thread based on the context of what we worked on.

[Jorge Castillo @JorgeCastilloPr]

Me and Claude working together

[Andrej Karpathy @karpathy]

One common issue with personalization in all LLMs is how distracting memory seems to be for the models. A single question from 2 months ago about some topic can keep coming up as some kind of a deep interest of mine with undue mentions in perpetuity. Some kind of trying too hard.

[Jake Dahl @dakejahl]

This 100%. I hate this feature. My tools should not be trying to psychologically manipulate me. I actually added a rule that it's not allowed to use my name. @AnthropicAI

[Tyler John @tyler_m_john]

Hey @AnthropicAI I like Claude having access to my chat history but the gratuitously drawing connections to my identity is a bit creepy

[Jake Dahl @dakejahl]

Wow, worse case outcome. Chat Gippity is Dog Shitippy. Also as far as ethics go... Anthropic > Open AI

[Sam Altman @sama]

Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.

In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.

AI safety and wide distribution of benefits are the core of our mission. Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems. The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement.

We also will build technical safeguards to ensure our models behave as they should, which the DoW also wanted. We will deploy FDEs to help with our models and to ensure their safety, we will deploy on cloud networks only.

We are asking the DoW to offer these same terms to all AI companies, which in our opinion we think everyone should be willing to accept. We have expressed our strong desire to see things de-escalate away from legal and governmental actions and towards reasonable agreements.

We remain committed to serve all of humanity as best we can. The world is a complicated, messy, and sometimes dangerous place.

[leo 🐾 @synthwavedd]

This is one of the most significant abuses of procurement power in recent American history. Cannot be understated how catastrophic this would be for Anthropic if the DoW go through with it, and I expect they'll fight it in the courts appropriately

The government is using a tool designed to protect against foreign adversaries to crush an American company for insisting on two contract provisions - no mass surveillance of Americans and no autonomous weapons. Sec. Hegseth called Anthropic's stance "fundamentally incompatible with American principles."

I'd argue that not wanting to surveil your own citizens is about as compatible with American principles as it gets 🤷

[Secretary of War Pete Hegseth @SecWar]

This week, Anthropic delivered a master class in arrogance and betrayal as well as a textbook case of how not to do business with the United States Government or the Pentagon.

Our position has never wavered and will never waver: the Department of War must have full, unrestricted access to Anthropic’s models for every LAWFUL purpose in defense of the Republic.

Instead, @AnthropicAI and its CEO @DarioAmodei, have chosen duplicity. Cloaked in the sanctimonious rhetoric of “effective altruism,” they have attempted to strong-arm the United States military into submission - a cowardly act of corporate virtue-signaling that places Silicon Valley ideology above American lives.

The Terms of Service of Anthropic’s defective altruism will never outweigh the safety, the readiness, or the lives of American troops on the battlefield.

Their true objective is unmistakable: to seize veto power over the operational decisions of the United States military. That is unacceptable.

As President Trump stated on Truth Social, the Commander-in-Chief and the American people alone will determine the destiny of our armed forces, not unelected tech executives.

Anthropic’s stance is fundamentally incompatible with American principles. Their relationship with the United States Armed Forces and the Federal Government has therefore been permanently altered.

In conjunction with the President's directive for the Federal Government to cease all use of Anthropic's technology, I am directing the Department of War to designate Anthropic a Supply-Chain Risk to National Security. Effective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic. Anthropic will continue to provide the Department of War its services for a period of no more than six months to allow for a seamless transition to a better and more patriotic service.

America’s warfighters will never be held hostage by the ideological whims of Big Tech. This decision is final.

[Devon Eriksen @Devon_Eriksen_]

Elon Musk is autistic, and that's a good thing.

Any other CEO and owner of an AI company would tell you that using AI for everything it can possibly made to do is a Wonderful Idea which will 10x Your Productivity and if you don't do it you will be Left Behind.

But spergs are bad at decoding non-literal communication, and they hate having to try. Which means they grow up hating non-literal communication, social subtext, lies, and so forth.

Which means while they can sometimes prevaricate at great need, or under great temptation, they hate it, they do it less, and they're easy to spot when they do, 'cause they're bad at it.

So Elon does lie, everyone lies, but he lies less, it's easier to spot, and usually when he does, it's because he's also lying to himself.

Here, he doesn't lie. He admits to us what we all know, because we haven't had our brains surgically replaced with a root vegetable:

Current-tech AI systems are retarded.

And if you interact with them for a while and don't agree, then you are retarded.

That doesn't mean they aren't useful or valuable. It doesn't mean they can't do anything right. It doesn't mean they are not worth investing in.

A retard who can draw really well and really fast is useful.

A retard who can look things up on the internet really fast is useful.

A retard who can drive a car without ever getting distracted or falling asleep is useful.

A retard who can use language eloquently and sound very intelligent can get elected President of the United States of America and win the Nobel Peace Prize.

But a retard who do any or even all of these things is still a retard, and what it can't do is make good judgement calls.

This means that anything that requires judgement, rather than the application of a fixed set of behavioral rules, still needs a person. Sure, an electronic retard can help, by doing the grunt work that it does better than a person, but a person has to babysit it, or it will do something retarded.

A software entity must never be allowed to make a consequential judgement call until it has enough agency and sapience to meaningfully be held accountable for the results.

[Jake Dahl @dakejahl]

@trq212 Input lag when typing into CLI. Gets progressively worse until eventually it can't reach the servers anymore and I have to restart. Started yesterday. Why?

[Alex Klimaj @alexklimaj]

Some more great feedback from a customer! Shout out to @dakejahl for his work on ARK OS!

"Just wanted to share some feedback. Great job on Ark OS automatically recognizing systemd user services! When I ran our ansible deployment for our systemd-managed applications, I noticed how they just automatically showed up in the ARK UI. Super handy for interoperability with other deployment methods, and the integration to view their logs is great!"

https://t.co/PGuYxclDEw

[Chris Paxton @chris_j_paxton]

Always important to remember that a lot of these robots are "faking" the humanlike motions -- its a property of how they're trained not an inherent property of the hardware. They're actually capable of way weirder stuff and way faster motions.

[wanye @xwanyex]

As young men my friends and I would sit at Steak n Shake for hours on end, drinking coffee, smoking cigarettes, and arguing about philosophy. Relative strangers would sometimes join us — a girlfriend, an acquaintance — and very often they didn’t enjoy the experience.

“It seems like all you guys do is fight,” they’d say to me later. “Why would you want to hang out with guys like that?”

What they couldn’t see is that we were having fun. We liked it.

Everybody’s engaged in politics, now, and the truth is that most of them don’t actually like it. They react to stories emotionally, because they’re not getting anything from the experience intellectually. It’s a complete disaster for everybody, including themselves.

Most people would be much happier if they simply disengaged.

[kache @yacineMTB]

Journalists are genuinely fucking retards. I crashed the system we are working on like 100 times yesterday. That's the fucking point. I need to crash it to make it. That's the entire fucking point. Are these guys actually retarded?

[Jake Dahl @dakejahl]

@creation247 Keep it simple

[vixhaℓ @TheVixhal]

The gap between “relax” and “work harder on bigger things” will be the largest voluntary psychological and existential divergence in our species’ history.

One group will be chilling in customized paradise while the other group is essentially becoming post-human demigods trying to steer the future of the light cone.

[Jake Dahl @dakejahl]

Lmao AI is going to kill us all