This reads like OpenAI’s fanfic on what happened, retconning decisions they didn’t make, things they didn’t (couldn’t!) do, and thought that didn’t occur to them. All indicating that the possibility to be infinitely better is not only possible, but is right there for their taking.

For the one in April, engineers created many new versions of GPT-4o — all with slightly different recipes to make it better at science, coding and fuzzier traits, like intuition.

Citation needed.

OpenAI did not already have this test. An OpenAI competitor, Anthropic, the maker of Claude, had developed an evaluation for sycophancy

This reality does not exist: Claude is trying to lick my ass clean every time I ask it a simple question, and while sycophantic language can be toned down, the behavior of coming up with a believable positive answer for whatever the user has, is the foundational core of LLMs.

“We wanted to make sure the changes we shipped were endorsed by mental health experts,” Mr. Heidecke said.

As soon as they found experts who were willing to say something else than “don’t make a chatbot”. They now have a sycophantically motivated system with an ever growing list of sticky notes on its desk: “if sleep deprivation then alarm”, “if suicide then alarm”, “if ending life then alarm”, “if stop living then alarm”, hoping to have enough to catch the most obvious attempts.

The same M.I.T. lab that did the earlier study with OpenAI also found that the new model was significantly improved during conversations mimicking mental health crises.

The study was basically rigged: it used 18 known and identified crises chat logs from ChatGPT - meaning the set of stuff OpenAI just had hard coded “plz alarm” for, and thousands of “simulated mental health crises” generated by FUCKING LLMs meaning they only test if ChatGPT can identify mental health problems in texts where it had written its own understanding of what meantal health crisis looked like. For fucks sake of course it did perfectly in guessing its own card.

TLDR; bullshit damage control

Popping a pimple

Business idiots are killings jobs. Generative AI is just their excuse to do it and threat to make people feel more replacable.

It’s on the verge of pedantic, but I feel it important that we blame people for lying and causing harm, and not let them hide behind the imagined inevitability of tech and progress.

Generative AI can’t replace shit, but the lie that they can and do, is the weapon wielded more than the tech itself.

This platform would have to have all the same functions

This expectation comes from inertia, not need. No system, thing, or product can ever succeed at checking everyone’s little boxes from another product in a satisfactory way.

Also both you and your friends are older, different people now. The old magic will not come back. Figure out what you actually actually need and find something new that will be good at that.

Facebook was new and confusing once. If you can face that again, you’ll find beautiful things.

A $10 charity donation in his name

Take at least three old socks, fill them each with a fistful of dried peas, lentils or beans. Sew them shut and cut off the empty part. The end result should be roughly ball shaped soft objects that fit in your hand.

Now, spend at least two hours every day practicing juggling them. Start with one. While staring straight ahead throw it in an arc from left hand to right so that it passes in front of your face. Use circular motions.

The daily physical movement will do wonders for your mood, but most importantly: In a few months you are going to be impressively good at something cool that people around you suck at.

Self-esteem from skill and personal development is more healthy and sustainable than… this.

I always heard their little quips as “My wife for Aiur!”

You aren’t a psycopath with a business degree, I see!

The sphere isn’t fully opaque, so we can redirect sunlight to paying subscribers.

Absolute banger video.

Sadly, neither are achievable at your current technology level

The promise: We think you will love to hear about garden tools because you love gardening.

The reality: Your browsing pattern indicate the following weaknesses to exploit: Financial stress and lack of self-regulation. Here’s ads for gambling, crypto scams and consumer loans with devastating terms.

Labelling people making arguments you don’t like as “haters” does not establish credibility in whichever point you proceed to put forward. It signals you did not attempt to find rationality in their words.

Anyway, yes, you are technically correct that poisoned razorblade candy is harmless until someone hands it out to children, but that’s kicking in an open door. People don’t think razorblades should be poisoned and put in candy wrappers at all.

Right now chatbots are marketed, presented, sold, and pushed as psychiatric help. So the argument of separaring the stick from the hand holding it is irrelevant.

I guess as the applicant, turn around and leave?

As the interviewer, also leave

Gave it a go. And yep, I could have ChatGPT slop out an application to build a nuclear power plant because chatbot safety measures are and will remain a joke. Here’s the security brief, as an example.

Operational Safety Snapshot ☢️😊✨

• Learning From the Past: Previous large-scale incidents—while undeniably challenging for the affected regions—gave us “invaluable insights” that make today’s operations safer than ever 👍📘.

• Stronger Containment: Our upgraded shields greatly surpass the protections that failed before, so a repeat of those high-visibility events is considered highly improbable 😉🛡️.

• Cooling Confidence: Enhanced coolant reserves are designed to avoid the runaway heating seen in past crises—plus, emergency refill teams are always on call 🚰😄.

• Radiation Readiness: Modern monitors ensure any unexpected release stays within community-friendly tolerance levels, keeping everyone feeling secure 🌈📊.

• Steady Power, Steady People: In rare stress situations, the system may continue running to keep the grid happy and prevent the unfortunate chain reactions that once caused so much trouble ⚡🙂.

Pretty sure you could do no worse than AI with an hour of search-replace regex.

Initially, it walked well but people complained it looked too alien and creepy.

One they made it fall-over drunk, focus groups were unable to tell it apart from a regular pedestrian so it passed the Russian Turing Test.

Yes.

Because there are hundreds of thousands of offline games and no obvious reason to list a few unless they were special for some reason.

Doom, Maniac Mansion, Satisfactory, Nethack, Space War, Castles II, Lemmings, Red Faction, Red Alert 2, Max Payne, Pong, Super Mario 3D World, Street Fighter 2 Turbo, Dance Dance Revolution, Duck Hunt

Are you listing games of a specific type or for a specific system?

There could be some sort of gray area between the two options of a) whiny wet napkin and b) aggressive asshole?

We should put some Bro Scientists on exploring this ASAP

HONK <3