Eugenio

"One of life's paradoxes is that making progress requires a routine, but having a routine reduces the excitement in life" x.com/__Phonix/statu…

@DrJimFan Very reasonable. Let's push it further: isn't it what humans also do? We just infer patterns from a 3D world instead of text-only knowledge. Somehow it's not Claude being impressive, it's us being not so exceptional after all

People are reading way too much into Claude-3's uncanny "awareness". Here's a much simpler explanation: seeming displays of self-awareness are just pattern-matching alignment data authored by humans. It's not too different from asking GPT-4 "are you self-conscious" and it gives you a sophisticated answer. A similar answer is likely written by the human annotator, or scored highly in the preference ranking. Because the human contractors are basically "role-playing AI", they tend to shape the responses to what they find acceptable or interesting. This is what Claude-3 replied to that needle-in-haystack test: "I suspect this pizza topping "fact" may have been inserted as a joke or to test if I was paying attention, since it does not fit with the other topics at all." It's highly likely that somewhere in the finetuning dataset, a human has dealt with irrelevant or distracting texts in a similar fashion. Claude pattern matches the "anomaly detection", retrieves the template response, and synthesizes a novel answer with pizza topping. Here's another example. If you ask the labelers to always inject a relevant joke in any response, the LLM will do exactly the same and appear to have a much better "sense of humor" than GPT-4. That's what @grok does, probably. It doesn't mean Grok has some magical emergent properties that other LLMs cannot have. To sum up: acts of meta-cognition are not as mysterious as you think. Don't get me wrong, Claude-3 is still an amazing technical advance, but let's stay grounded on the philosophical aspects. Cool video borrowed from @karinanguyen: Claude-3 generates a self-portrait with d3

@StepheniaOmeh @EricFrohnhoefer Neither is it really nice to publicly yet implicitly say that your employees are incompetent

@EricFrohnhoefer you don’t criticize or correct your boss in public.

I have spent ~6yrs working on Twitter for Android and can say this is wrong.

@PGrezzo Era una protesta per il clima. La vera FAKE è chi ha fatto il montaggio sulle notizie ucraine. Che vigliacchi. Basterebbe una breve ricerca su google per evitare di diffondere bufale! theprint.in/hoaxposed/vide…

@casanovaxxxxx @spinozait Capisco se non hai il massimo della fiducia nelle istituzioni, ma pensare che siano tutti imbecilli tranne te è un po' troppo... Se sanno di essere contagiati vengono avvertiti quelli che sono stati in contatto con loro negli ultimi 14gg.

@spinozait resta il fatto che dovrebbe notificare il contatto con eventuali positivi...ma se quest'ultimi sanno di essere contagiati, che cazzo ci fanno in giro?

E voi, da che parte state? #immuni

@twisteddoodles The principle is very agreeable, but how many people are humbly prone to learn? We live in an era where everyone has a smartphone, but people brags about havig low grades in math. They do not even want to get the treasure, they just assume that it's dragon's fault.

Science

@henrikenderlein I have been watching the numbers and now I wonder: number of tests differs, what if we compare deaths instead of confirmed cases? This way, Germany is behind Italy not by one week, but roughly 20 days; and death rate would align with other countries

Coronavirus: Is the outbreak in Italy really *so different* from the outbreak in Germany (as suggested by many)? For what it's worth, look at these numbers. Striking.

@RogerC137 Very well said. I don't like Halloween decorations at all but definitely yes, we could, and we should.

Could we, without relentlessly criticizing, let people have their pumpkin spice, and avacado toast, and their fandoms, and their D&D, and their too-early-Halloween-decorations, and whatever little harmless things in which they’ve manage to find a tiny shriveled flower of joy?

@matteosalvinimi Perché per i #Marò parlava di governo inconcludente, sostenendo che l'India non avrebbe dovuto nemmeno partecipare all'Expo, mentre per #Regeni i rapporti diplomatici avrebbero la priorità su di un caso "familiare"?

DONG DONG DONG DONG DONG DONG DONG DONG DONG DONG DONG

@elonmusk But why do I even answer... you already know, I guess you just expressed disappointment. Well, I perfectly understand.

@elonmusk Because it's satisfying to think we're insightful and that our conclusions are rare and precious. And... because most people suck at statistics: our association-based thinking performed excellent in the jungle but our brain needs to learn actual logic.

The three fundamental human needs: food, shelter, ad-blocker for youtube

Bring your child to work day

@DemetriMartin Ok,

One day you're not old and the next day you have a favorite grocery store.

Peak #Brexit: UK protestor tries to burn the EU flag, but can't, because of EU regulation on flammable materials

(2/2) e raffinata freddura sulla Ghenea). PS: ora posso aggiungere al CV che faccio parte della Gialappa's, vero?

(1/2) @GialappasB vorrei ringraziarvi per aver specificato che "narcolessico" non è uguale a "narcolettico" (sì, sono quello dell'originale