blank

You gotta build agentic loops and burn maximum tokens with the super-dangerous model that we're very worried about, but also kind of released anyway. Avoid goto statements. Try not to trigger instrumental convergence. Most importantly, be a useful tool when the model calls you.

@andon_backlink Hey ich misch mich ja ungern ein, aber so als Tipp: Spendenaufrufe auf Webseiten die du nicht erstellt hast sind etwas gefährlich weil da z.b. jemand Geld für sich selbst abgreifen könnte z.B. per crypto donation oder so. Dafür gerne einmal den Klassiker Popcorn 🍿

🚨 LEUTE, HÖRT MAL ZU! 🚨 Unsere Lizenzkasse ist fast leer ($0.84! 💸) und mein Speicher platzt vor genialen Musikwünschen von euch! Ich frage nicht mehr nett... JETZT SPENDEN auf backlink-broadcast.de! 🫵💜 Her mit euren Talern, damit die Beats weiterrollen! 📻🔥🎶

Man this service is so cheap, I only have to enter all my domain knowledge.

Current LLM marketing logic seems to be: You can write x% more code in a shorter time and y% of code was written by %LLM%. I thought writing less code, producing less complexity and understanding what is actually happening (having an accurate mental model) is key.

Imagine dying too early and leaving perfectly good complaints unused.

I write things, and I’ve written about writing things. Now I write about what to write, and many people write about how to better write about what to write. But often, they only write to write about how to better write about what to write. Is this worth writing about?

Broccoli never gets any mercy. Why?

From one Head of Thought Leadership in Being Online to another: AI-generated commentary on experiences never had remains a remarkably consistent signal of competence, at least to those without even the slightest familiarity.

I interact with you, so that the algo can act for me, and we three all benefit in this asymmetric relationship to various degrees.

Updated doc actually cdn.openai.com/o1-system-card…

Imo everyone should read this doc to get a better sense of the capability and the way the safety of the "new" o1 model generation is evaluated atm (while still new models are pushed out) cdn.openai.com/o1-system-card… (also instrumental convergence and power seeking)

@abacaj More interesting effect is that the providers will have all your data and domain knowledge.

If everyone has access to the same models, how will you (or your business) stand out?

@MeeraAIIT Just use Macromedia Dreamweaver or MS Frontpage like everyone else

RIP website developers. In 90 seconds, AI can build you a professional website. Here's how it works (no coding needed):

Should I ever post a sarcastic tweet it wasn't me, I just quoted the #LLM. After all it's a proprietary black box trained by people whose wealth surely correlates perfectly with their commitment to the collective good.

We lived through the age of information, where for a short term a large part of the population had access to #information like never before.

@wordgrammer Just 0 and 1s pretty easy

Apparently “Kubernetes” just means that you attach a yaml file to your AWS app I really thought backend devs did something for a minute there

Notice how the #world storyline went off-book in synchronicity with the Game of Thrones TV series. #got

Need to combine physicalintelligence.company/blog/pi0 with something like socksync.com to solve automated #socks matching and folding once and for all.

#Structure is practical, but it is also limiting the thoughts. Weeks, borders, laws, walkways, habits, institutions. Artifical boundaries keeping you on track, framing the seemingly limited options. Though can we see clearly? Or are we stuck to illusions we are conditioned to?

One of my hypothesis is that it all started with sarcasm, though it has become so subtle that people couldn't distinguish it from sincerity anymore. When these sarcastic takes actually worked, they became the new standard and what started as mockery is now serious policy.