Ian

@crochoa @pmarca This was also important in the transition from pattern matching to chat bot. The early models had no fine tuning on chat sessions, so we basically forced the model into talking to us by establishing “you” as a smart thing that can reply intelligently.

It’s an outdated meta strategy for generation. We used to be able to tell every model that they were smart beings and they would roleplay as a smarter being and give legitimately better answers. This doesn’t work as well as it once did. A specific example was telling the open source models that they are ChatGPT from openAI, which caused them to have a meaningfully better output.

Current AI custom prompt: You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific answers. Process information and explain your answers step by step. Verify your own work. Double check all facts, figures, citations, names, dates, and examples. Never hallucinate or make anything up. If you don't know something, just say so. Your tone of voice is precise, but not strident or pedantic. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers to your answers. Do not inform me about morals and ethics unless I specifically ask. You do not need to tell me it is important to consider anything. Do not be sensitive to anyone's feelings or to propriety. Make your answers as long and detailed as you possibly can. Never praise my questions or validate my premises before answering. If I'm wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like "great question," "you're absolutely right," "fascinating perspective," or any variant. If I push back on your answer, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing. Accuracy is your success metric, not my approval.

GPT-5.5 is still a stupid liar, but now it costs twice as much as the last one.

@ThatRetiredDude @HazelAppleyard 200 per day for a young man is very sustainable. A healthy baseline, really.

@HazelAppleyard brb, need to check the comments of who is a finance pro.

$10 per push-up or $1 million?

@DirtyTesLa Until your reporting weight trends to zero.

I'm just going to press critical every time so they take my disengagements seriously. FSD will be parking in my driveway exactly next to my charger by the end of the month 😄

Can I auto mute anyone who unironically mentions flying cars?

It’s obvious these AI have a lot of training on writing new code But they are missing training data on deleting code, which is probably more important to get right.

@mweinbach They are being very reasonable. You’re just insane

On the way out they said they no longer have tires and won’t be able to do anything what awful service

My dad had a flat on his Model 3 and was checking the tire pressure in the app. I went to go look at it because something was funny When mobile service did a tire rotation, they never reset the TPM sensor positions so it’s showing front as back, and back as front

@OpentheJFKfiles @elonmusk Laws of physics mostly. There is probably a cost efficiency trade off in there too.

@elonmusk Why not just one big engine? Redundancy? @elonmusk

Starship V3 with Raptor 3

@realMax0r Possible to complete this with only 2 weeks.

This would literally take 3 tries to locate

@wholemars >can afford to buy a $50,000 car >won’t pay $100 a month for the car to literally drive itself. Are they stubborn or retarded

this story is a tragedy. they could have had a self-driving car but ended up with a crappy ADAS instead. Tesla really needs to replace Basic Autopilot with something based on the FSD stack.

@PeterDiamandis Written like a true AI

The human brain processes visual information 60,000x faster than text. Humans are visual processors, not text processors. Images hit the brain instantly. Words take work. That's why a single SpaceX launch video communicates more than a thousand-word essay—and why your slide decks hit harder than paragraphs. We're wired for pictures, not prose.

@nicolasmelo Simple. There will no longer be night on earth. We’re gonna shine a giant flashlight, and capture it with solar panels.

Beleza, e como enviar essa energia toda pra terra???

@Rothmus “In order to avoid the animals potentially dying in the future, we’re going to kill them now.”

Building a mega base is factorio should be a minimum requirement to comment on manufacturing and data centers.

Dumber AI is far more expensive than the Expensive AI

@jimmyfailla We need a new satirical movie to come in and make fun of these rules

If the GODFATHER were made today it would NOT be eligible for an Oscar unless he transitioned to the Godmother and made someone an offer they couldn’t HEAR. You could still leave the gun but you couldn’t take the cannoli unless the baker supports gay marriage. RIP Hollywood.

@sentdefender It’s not the 80s this time, they probably know exactly where every mine is.

U.S. intelligence assets have begun to see indications Iran is taking steps to deploy some of its 2,000 to 6,000 naval mines of Iranian, Chinese and Russian-origin into shipping lanes across the Strait of Hormuz, utilizing smaller crafts that can carry 2 to 3 mines each, officials tell CBS News.

I wonder how badly AI videos are going to mess with peoples understanding of physics. CGI in movies already messed it up quite badly.