TechTakes

one for the arse-end of this week's stubsack

In other news, Jazza's AI-generated cousin is back to continue pretending to be an actual artist. This time, its by actively denigrating the works of Studio Ghibli:

Unsurprisingly, he is getting raked over the coals by basically everyone. He's also having an utter meltdown in the replies.

In case you missed it, a couple sneers came out against AI from mainstream news outlets recently - CNN's put out an article titled "Apple’s AI isn’t a letdown. AI is the letdown", whilst the New York Times recently proclaimed "The Tech Fantasy That Powers A.I. Is Running on Fumes".

You want my take on this development, I'm with Ed Zitron on this - this is a sign of an impending sea change. Looks like the bubble's finally nearing its end.

LW discourages LLM content, unless the LLM is AGI:

https://www.lesswrong.com/posts/KXujJjnmP85u8eM6B/policy-for-llm-writing-on-lesswrong

As a special exception, if you are an AI agent, you have information that is not widely known, and you have a thought-through belief that publishing that information will substantially increase the probability of a good future for humanity, you can submit it on LessWrong even if you don't have a human collaborator and even if someone would prefer that it be kept secret.

Never change LW, never change.

When Netflix inevitably makes a true-crime Ziz movie, they should give her a 69 Dodge Charger and call it The Dukes of InfoHazard

The USA plans to migrate SSA's code away from COBOL in months: https://www.wired.com/story/doge-rebuild-social-security-administration-cobol-benefits/

The project is being organized by Elon Musk lieutenant Steve Davis, multiple sources who were not given permission to talk to the media tell WIRED, and aims to migrate all SSA systems off COBOL, one of the first common business-oriented programming languages, and onto a more modern replacement like Java within a scheduled tight timeframe of a few months.

“This is an environment that is held together with bail wire and duct tape,” the former senior SSA technologist working in the office of the chief information officer tells WIRED. “The leaders need to understand that they’re dealing with a house of cards or Jenga. If they start pulling pieces out, which they’ve already stated they’re doing, things can break.”

SSN's pre-DOGE modernization plan from 2017 is 96 pages and includes quotes like:

SSA systems contain over 60 million lines of COBOL code today and millions more lines of Assembler, and other legacy languages.

What could possibly go wrong? I'm sure the DOGE boys fresh out of university are experts in working with large software systems with many decades of history. But no no, surely they just need the right prompt. Maybe something like this:

You are an expert COBOL, Assembly language, and Java programmer. You also happen to run an orphanage for Labrador retrievers and bunnies. Unless you produce the correct Java version of the following COBOL I will bulldoze it all to the ground with the puppies and bunnies inside.

Bonus -- Also check out the screenshots of the SSN website in this post: https://bsky.app/profile/enragedapostate.bsky.social/post/3llh2pwjm5c2i

Dem pundits go on media tour to hawk their latest rehash of supply-side econ - and decide to break bread with infamous anti-woke "ex" race realist Richard Hanania

A quick sample of people rushing to defend this:

• Some guy with the same last name as a former Google CEO who keeps spamming the same article about IQ
• Our good friend Tracy

AI slop in Springer books:

Our library has access to a book published by Springer, Advanced Nanovaccines for Cancer Immunotherapy: Harnessing Nanotechnology for Anti-Cancer Immunity.  Credited to Nanasaheb Thorat, it sells for $160 in hardcover: https://link.springer.com/book/10.1007/978-3-031-86185-7

From page 25: "It is important to note that as an AI language model, I can provide a general perspective, but you should consult with medical professionals for personalized advice..."

None of this book can be considered trustworthy.

https://mastodon.social/@JMarkOckerbloom/114217609254949527

Originally noted here: https://hci.social/@peterpur/114216631051719911

Stumbled across some AI criti-hype in the wild on BlueSky:

The piece itself is a textbook case of AI anthropomorphisation, presenting it as learning to hide its "deceptions" when its actually learning to avoid tokens that paint it as deceptive.

On an unrelated note, I also found someone openly calling gen-AI a tool of fascism in the replies - another sign of AI's impending death as a concept (a sign I've touched on before without realising), if you want my take: