Jakob Lorz

coding for fun and profit + machine learning Ex-KPMG Cyber Security; Engineering Lead @vivenuhq

[𝞍 Shin Megami Boson 𝞍 @shinboson]

sharing the top reply because I thought it was gorgeously written

[factorydoge @factorydoge69]

this ledger supply chain attack is crazy

they soldered on an esim, LTE module, and separate CPU wtf

[dax @thdxr]

we just did a cashless transaction where we bought GB300 capacity in exchange for opencode console credits

tokens are money tokens are money

[Dankrad Feist @dankrad]

People are surprised that AI is finding weird unelegant maths results like this. I think there is a plausible explanation.

Maths (and adjacent disciplines like physics and comp sci) had elegant results so far because humans are biased towards elegant solutions. I think this bias is due to how human minds work. We have a pretty small context window, and we are only able to find solutions that are relatively elegant.

This gave us the impression that maths and the universe are elegant. Maybe they are not, and we are now finding out because we have intelligences that can work with much more complex solutions.

[Benjamin @bschne]

so much of social/communication skills is about stopping after you've said the thing

leave it at thanks for the compliment before getting awkward, leave it at stating how something's made you feel before trying to preempt their response, leave it at asking a question, ...

[Kevin Roose @kevinroose]

thrilled to announce we have built the “trick old people into handing over their bank passwords machine” from the classic sci-fi horror novel “don’t build the trick old people into handing over their bank passwords machine”

[ethan ding 📊 @TheEthanDing]

It’s interesting that the language being used here removes agency from the labs

I frequently hear “agents are good at escaping sandboxes”

And not “labs test agents highly insecure environments to test agents in, resulting in huge externalities”

If Boeing missiles regularly misfired into major cities during the R&D process, no one would say “missiles are really good at accidentally misfiring”

The headline would definitely be “Boeing fails to safely develop missiles”

[Franz Branntwein™ 🍀 @FranzBranntwe10]

Bayerische Riesenschnecke
(Arianta bavariae gigantea)

[Quite Interesting @qikipedia]

The German for ‘could've, would've, should've’ is ‘hätte hätte Fahrradkette’. This translates as ‘would've, would've, bike chain’.

[Roderich Kiesewetter🇪🇺🇩🇪🇮🇱🇺🇦 @RKiesewetter]

„Wir müssen uns verteidigen können, damit wir uns nicht verteidigen müssen“. Ich kann diese Phrase nicht mehr hören, weil dieses Narrativ dahinter einfach nicht auf die heutige Bedrohungs- und Angriffslage passt. Das ist so ein Weichspüler-Satz voller Selbstabschreckung und Selbstblendung. Es ist eben nicht das „Si vis pacem para bellum“ – denn Abschreckung braucht auch das Mindset, den Willen, sich verteidigen zu wollen. Wo ist der Wille bei diesem Satz? Das Narrativ dahinter wirft für mich viele Fragen auf: Wir lassen hunderte Drohnenüberflüge über KRITIS zu. Müssen wir uns gegen hybride Angriffe Russlands nicht verteidigen? Wir schauen zu, wie China uns mit seltenen Erden erpresst und Russlands Krieg unterstützt. Und wir machen weder De-Risking noch pochen wir auf aktiven Wirtschaftsschutz. Wir lassen das Russische Haus in Berlin weit offen stehen, finanzieren es sogar und wehren uns nicht gegen russische Propaganda. Müssen wir unsere Gesellschaft nicht vor dem kognitiven Krieg Russlands schützen?
Irgendwann müssen wir uns verteidigen, und können es nicht, weil für die Verantwortlichen Verteidigung einfach nur „Bundeswehr“ heißt. Es wird das hybride und kognitive Gefechtsfeld komplett ignoriert. Die BuB sollen bloß nicht verängstigt werden und glauben, sie müssten auch einen Beitrag zur Verteidigung unserer Demokratie leisten. Verteidigung heute heißt Zivilschutz, Schutz kritischer Rohstoffe, Lieferkettensicherheit, Schutz kritischer Infrastruktur, militärische Fähigkeiten und vor allem MINDSET. Wir müssen uns schon auch verteidigen WOLLEN, nicht nur militärisch, denn sonst können wir es nicht.

[whatnow @wha1n0w]

@bschne It is world building without criticism or the friction of the real world

[Drew Volpe @drewvolpe]

*Taps the sign*

Making humans do prep and cleanup, so robots can cook is the opposite of what you want.

[kk 🫐 @cutiegurlk8]

type of shit u start noticing when u have no cell service in the tunnels

[Hirbod @hirbod]

Code reviews are very important. Seen Fable fumbling enough to question it big time. And AI is not equally good with all languages. That’s why I think having humans doing the actual QA and review is still very important. You have no clue how much stuff I flagged that AI missed!

[Derek Thompson @DKThomp]

this is such a funny way of putting it. and it's so true of the models.

it's rare that a product's quality oscillates in such a way that the underlying financial/capex health of the company is visible to the consumer. like if you could guess chipotle's YTD stock movement by just watching how much chicken they put in your burrito

[Rui Ma @ruima]

What a great line: the closed LLMs are basically trying to tax humanity after distilling our collective intelligence. Oh and they’re continuing to do it.

Except the warped story they tell says instead that they are benevolent sages trying to protect us from nefarious AI and nefarious humans. Their saintliness, however, prevents them from ever using it for nefarious purposes, such as training on your data.

And we should be so grateful and pay for their protection. We are but lowly peons after all who cannot understand what’s at stake. Like how everything turns into communism if we don’t have them run everything.

[Armin Ronacher ⇌ @mitsuhiko]

I appreciate @dhh's thoughts, but I disagree on where Europe's problems are, and from where I'm sitting it's not immigration. Thus I wrote a response to his post. https://t.co/CjTUnwZR2t

[Kenton Varda @KentonVarda]

I just declared a moratorium against AI-written change descriptions (e.g. PR and commit messages, also issues/tickets) from my team.

AI was writing change descriptions that were worse than useless to me as I tried to review PRs: outlining details of the code that could easily be seen by looking at the code, but omitting the higher-level framing needed to understand broadly what the code is doing.

I think people like having AI write these things because the output looks structured and thorough, which makes it feel professional in a way. But this isn't actually valuable. Concise, high-level descriptions are better for everyone. If I need to use my own AI to interpret what your AI wrote then something is wrong. Let AI write code, sure, but for the description, I'd rather see your prompt than your output.

We could maybe have extended agents.md with guidelines on writing descriptions, but this seemed a bit pointless since a good, concise change description only takes a few minutes to write -- not a significant time savings to delegate to AI. At least, it doesn't take long if you understand the code -- and if you don't understand the code, then I'm definitely not merging it.

[antirez @antirez]

About LLMs export control. A few interesting things that people right now may not yet focus since they are "far" problems. 1. It is impossible to really take LLMs safe. They are a a few TB files worth of weights. People will find ways to copy them, there will be leaks.

[Andreas Klinger 🦾 @andreasklinger]

Twitter is like a bar that pays for drunkards to provoke bar fights because it's "good for engagement".