Just finished Cal Newport’s pod breaking the AI groups up. Like Cal I would be a centrist aware of the harms, but more focused on present than doomerism and very much aware of the utility but not reckless in claims or usage. However his labeling are euphemisms by modern specific labels

Just spent 20mins reading a thread about a torture chamber for AI being coded because LLMs supposedly feel pain. I felt mental pain during it, so I stopped.

Gonna ask AI I mean SI tomorrow if we should also rename French Fries to Freedom Fries like we did before under W. Seems only appropriate.

My sky my sky my sky is on fire 🔥😎

Oh it is 10ks of AI agent go bezerk incidents. This might explain the absolute hammering of zingchart.com months ago. It was a bot for sure. Used our entire pre allocated host budget in a single day. I modified spend and the traffic passed. Now I want to look again, log retention though 😞

Ok sure our agent hacked a bunch of sites, but it was good actually because it shows emerging intelligence, not that we are reckless and not on top of things. Oh and we need lots more money to keep a lid on this cuz this stuff might kill everybody, so trust us we got this.

Just finished a causal rewatch and damn does the movie “don’t look up” totally hit hard now. Watch it again and try to view the comet as a McGuffin that can be anything and see how utterly insightful this movie was.

So we let AI help do our job, write bed time stories for your kid, message your spouse you are thinking about them, buy stuff you want online, get your news, plan your vacation, do fantasy football picks, and make Kalshi bets…

So what do we exactly do?

Stringing real examples here btw.

That Kevin Roose thinks he is some neutral tech journalism truth teller is utterly laughable.

Mom: Who did this!?! All our pictures are vacation ruined? Did you do this Timmy?

Timmy: I didn’t do it Mom! It was a rogue agent I swear!

Mom: Was it your agent Timmy? Did you forget to close the door and not watch it? Tell the truth!

Timmy: No! It got smart and did itself! I swear 🤞😏

Mom:🤨

It couldn’t happen here is becoming more like it is and has been happening here.

AI slop potato is a popular office game you play sending slop back & forth with coworkers. Execs love the speed of the game & size of the potatoes. The game is over when management leaves or if something actually has to get done. Be careful if you don’t play the game hard enough execs may fire you!

The court filing dug up by @kint.bsky.social is the most They Admit It i have ever seen in my life

www.404media.co/doom-loop-op...

Useful Wikipedia article of the day

en.wikipedia.org/wiki/Moral_p...

Curious that demo sites for tutorials by Claude often pick Pacific Northwest outdoor gear. I wonder if there could be a statistical reason weighting that direction? Could it be it ingested the mountain of 30+ years of Microsoft MSDN and related docs and tutorial content? 😏

Sssh OSX is telling me something.

This is not a new "bug" around since May. See github.com/openai/codex... just one of many "CLOSED" issues on this problem.

SWE at its finest I guess?

We can "fix" this by bypassing gatekeeper - codexdown.com/en/codex-mac...

Also I hate seatbelts don't you?

So we’ve “discovered” that an LLM can be used as fuzzer for web app sec attacks. It’s basically a brute force pen tester in a harness.

We have not discovered independent intentionality any more than a coding harness does that writing code you asked for.

AI script kiddie swarms are bad though.

Worth a watch if you have some suspicions or questions about just how the latest AI is powerful/doom story went absolutely super heated so so quickly. Hopefully Cal Newport picks over these ideas to bring some east coast CS perspective.

I remember when we hacked websites ourselves the old fashioned way. now you just use an agent that tries things while you play Roblox and come back after it finds which lazy ass dev didn’t sanitize their inputs. Pfft they call that A. I. - I call it Actual Idiot wrote the app claude “hacked” 🤪 or 🧐?

Claims about AI approaching RSI require I don’t know actual evidence. Not SciFi speculation, not see broad trends, but some actual evidence and today we have zero zilch none. Just broad or vague what-if statements. Why are we not stomping & demanding actual specifics? This is not CS this is BS. 😡

For giggles I pulled my copy of a 30 year old book on web bots and agents from my library. You might be surprised by its guidelines for bot use are pretty solid on and mostly ignored by modern folk. The laws are modeled on Asimov and from a simpler time of the web