V. Sus

Asking for an unjailbreakable LLM is like asking for unhackable software.

@Polymarket Asking for an unjailbreakable LLM is like asking for unhackable software. Next they’ll want one that doesn’t hallucinate… both impossible things.

NEW: The White House is reportedly demanding Anthropic make Fable 5 impossible to jailbreak before rerelease — which security experts say “can’t be done.”

@bryancsk It just requires an adaptation period. It took me 3 times, progressively, and now I find it delicious.

You need to have a very complex palate to appreciate durian. Among Europeans, the French might have it.

At some point, I was curious and hopeful too regarding the technology behind them. But it’s been clear for a long time that they don’t have a practical use. They’ve caused more harm than good: too much fraud, too many criminals, too much speculation (I’d describe them as betting technologies). It would be interesting to know the % used for criminal activity, though it’s difficult to quantify.

What once began as a noble experiment in forming a digital currency unencumbered by financial institutions, cryptocurrencies have proven to be an abject failure by every measure. There are on the order of only 30 million bitcoin wallets whose value is greater than $100 US; this represents at best about 0.36% of the world’s population, hardly a sufficient number to provide a global medium of exchange. While there certainly exist some fringe cases of utility, most notably by individuals needing to transport value outside dictatorial countries with broken and oppressive economies, cryptocurrencies are largely an object of speculation, driven by market manipulation and the economy of the greater fool and favored by criminal organizations and countries such as Russia and North Korea seeking to avoid sanctions. The ancillary industries spawned by cryptocurrencies – NFT and DeFi and so-called Web3 - never gained traction, and indeed proved to enrich a small few at the expense of the many. At the start of this technology, I was curious and hopeful, but it is clear that human greed has irredeemably ruined it for all.

@Solzi_Sez Thanks, I’ll check it out.

@VSUS I noticed that too. Seemed like Fable was pitting agents against each other internally and returning winning results. Also was adversarial toward me, challenging based on primary source evidence (as if it actually read my MD). 🖤 This was a good read: platform.claude.com/docs/en/build-…

Holy cow! Was just playing around with Fable here on Saturday morning. 😳

@JaycelAdkins Great catch! Did you get it from AbeBooks? I was monitoring the last one available, and by the time I realized, it was already gone.

Book haul:

Yes, they have been playing performative games for some time. Fable performed well at coding (I was able to almost build an entire app with it), but worse on semantic problems. It seemed trained primarily for coding, and very recursive, with many internal self-correction loops (hence the 2x token consumption). Previously, I used to make 2-3 sessions argue with each other. Fable seemed to do it internally (just my impression).

@VSUS Yeah, they are definitely playing marketing games…including degrading the old in order to show gains for the newer. It performed lamely on my hard Wed test. But when I got time Fri-Sat to do real work with it (and not care about usage limits), wow, it was almost scary good.

@_sholtodouglas It’s frustrating. It’s a dumb, overconfident, prescriptive model that wastes your time. Far from Opus 4.6 during its first week.

When do you reach for other models instead of Claude? What can we do better? Hit me with all of your frustrations. dms open. If you can give me detail (e.g. specifics/transcipts) - it'll help a lot in finding out exactly what we need to do to improve the next model

This is roughly equivalent Confucian “humanity and rightness” 仁義. One knows what is right instinctively by their own humanity and the feelings which issue forth from it.

@FurkanGozukara Well, there is a difference between a software architect and a software developer.

The definition of coding has completely changed forever. Jensen Huang reveals that the number of programmers just jumped from 30 million to 1 billion. Every carpenter and plumber is now a software architect.

Working from this café gives you cyberpunk vibes.

@blackym25 So calm and peaceful, would love working from there.

I hope to work in jiangnan even without pay.

Current AI is fascinating, I’m enjoying it a lot. But I find it only reduces the cognitive load of mechanical, repetitive tasks, not intellectual ones. It’s not intelligence yet, unless you consider a calculator to be intelligent.

@DullClayCutter Now I’m curious, what would have been your answer?

Wanted to reply to a post asking about a book that changed one's life, but then I saw the OP called Ayn Rand's Fountainhead timeless and I decided it's not worth engaging. Rand was was a venomous snake and her books trash. Read better books.

@DavidBCollum in 2014, an Iranian woman won the Fields Medal in mathematics. This is an accomplishment both for her and for her society, one that challenges the cartoons drawn by US media. (No US woman has ever won a Fields Medal.)

@Solzi_Sez @Snch60710Rafael @HuggingModels The price of checkmating sophists.

@VSUS @Snch60710Rafael @HuggingModels It has happened occasionally over the years, after I go on the offense and checkmate some account until they can’t even breathe…two days later troll accounts appear seeking redemption! Could be Nancy, or whatsherface.

Ever wondered how AI could read ancient Chinese texts? Meet BERT-Ancient-Chinese, a specialized model that understands classical Chinese literature. It's like having a digital scholar who can interpret texts from centuries ago. This fills a huge gap in NLP for historical languages!

@Solzi_Sez @Snch60710Rafael @HuggingModels Impossible! I’m just a really tiny account.

@VSUS @Snch60710Rafael @HuggingModels Maybe trolling you and just me as a bonus? Honored! 🙇‍♂️