Marcus

Hefeweisen Dunkel. First beer in Germany. 🇩🇪🍻

I'm not excited by what AI can do. I'm excited by what I can do.

I admit, I chuckled...

Good summary of the transition I'm still working on

Such a cool art piece!! From www.etsy.com/shop/Painted... #cosmere #stormlightarchive #sanderson

Writing software and maintaining software are two very different things. It's a credit to @edzitron.com's research that he picks up on this so clearly as a non-engineer. Also a great example of how coding is only a small part of software engineering.

So are the pricing deals ending? Is this the beginning of actually paying what these things cost?

Could developers make this mistake? Sure Have developers made this mistake? Many times However in production systems there are deterministic tools specifically setup to prevent this. Also if a human accidentally destroys an environment a new one is deployed in minutes, maybe an hour. Not 13 hours.

These are also the kind of details of love to hear more about. Do analysts or other journalists actually check into this stuff??

"up to" does a lot of heavy lifting. Many people seem to forget $0.00 is included in "up to." If a deal doesn't have a minimum commitment, it doesn't matter.

"The test of our progress is not whether we add more to the abundance of those who have much; it is whether we provide enough for those who have too little." —FDR, 1937

AI is a bubble, but too many people confuse a bad investment strategy with the tech itself. AI will not go away after the bubble bursts any more than websites went away after the dot-com crash. The tech is too useful (though not the panacea the hype mongers would have us believe). 1/3

@garymarcus.bsky.social: LLMs still hallucinate and continue to make boneheaded errors, and this Caltech / Stanford paper is yet another suggesting that this is inherent. Even LLMs marketed as offering "reasoning" systems have major problems with reasoning. garymarcus.substack.com/p/breaking-l...

Hehe, quoting @garymarcus.bsky.social "if you care about the security of your device or the privacy of your data, don’t use OpenClaw. Period." (Bonus : if your friend has OpenClaw installed, don’t use their machine. A. Don’t catch a CTD — chatbot transmitted disease)

I love this art. It's from a Brotherwise Games announcement about the Cosmere RPG.

You know what also happens if Oracle can’t spend this much on capex every quarter? NVIDIA stops selling as many GPUs, by which I mean Oracle stops ordering billions of servers through Hon Hai/Foxconn, Quanta, Wistron and Wiwynn, AKA the Taiwanese server manufacturers that really buy the chips.

This needs to be shared!! 🤣🤣

There's a ton of stuff in here that Ed put the stats in better than I could, so if you got the time it's well worth the read. But this line killed me! "Every part of the AI bubble — this fucking charade — is unprofitable, save for NVIDIA and the construction firms erecting future laser tag arenas"

Ok so using N8N and an LLM to build a standup report that: 1 - lists PR's merged yesterday. 2 - lists open PR's as blockers 3 - lists In-progress Jira's for what I'm doing today Is it just me, or is all of this very possible without an LLM??

Google and AI search are two different things. The ad traffic problem isn't something I considered before hearing about Cloudflare trying to tackle it. The next couple years for genAI will be interesting. The era of free training data may be closing.

This sounds way more accurate than most of what I hear put out there.

"... basic critical thinking goes out the window because you feel pressured to follow the crowd."

Chatbots as a luxury good. on.ft.com/4p4LRQS

Great discussion. I really like Cory's proposal about the international community piercing vendor lock-in.

Number go up, number go up, no think about number number always good number always up number up number up number up

AI is cooked

Two trials will be repeated endlessly despite always producing the same results: 1) Does UBI work? (yes) 2) Does AI improve productivity? (no)

Not a company I've heard of, but I got a kick out of this exchange. I think the interviewer didn't know how to deal with someone so rational. He really seems more used to leaning into the hype. This founder is way more grounded!

Some very refreshing content. Another arrow in my quiver of competent people speaking more pragmatically about LLMs and their capabilities.

Anybody remember AI a month ago? • Everybody in the Bay Area seemed to think that AGI was imminent • Expectations for GPT-5 were through the roof • Zuckerberg was spending bajillions of dollars on Alexandr Wang, staff and GPUs, and seemed to have a coherent plan

I haven't followed AI closely for some time. Deep dived this last week due to some stuff at work. I can see some helpful use cases, but the vast majority of what I'm seeing is Hype. I'm looking forward to the bubble bursting so we can work with it rationally.

America cannot long remain free, nor first among nations, if it becomes the kind of place where universities are dismantled because they don't align politically with the current head of the government.

In light of everything happening, I wrote about how the KKK in the 1920s felt unstoppable, about the people that fought against them anyway, and about how fascism always fails. dansinker.com/posts/202…

Interesting piece on how prevalent algorithms are. It's reinforcing what I think I'm liking about BlueSky.