You don't really know what level GPT-5 is going to be.

Regardless, you're right - we're not going to leapfrog right over the scientific method with AI. Experimentation and verification will be necessary.

But ask yourself how much things would accelerate if there was an essentially limitless army of postdocs capable of working tirelessly and drawing from a superhuman breadth of interdisciplinary research...

Yeah, I agree. We're actually communicating in natural English with artificial intelligences that can reason and create. It's literally the future I had been waiting for but that never seemed to arrive.

And yet... things are still early enough for goalposts to be moved. There's still enough gray area to think that this might not be it, that maybe it's just hype and that maybe life will continue with no more disruption than was caused by, say, the Internet.

The next phase shift happens when artificial systems start doing science and research more or less autonomously. That's the goal. And when that happens, what we're currently experiencing will seem like a lazy Sunday morning.

If there is anything that this horrible tragedy can teach us, it's that a male model's life is a precious, precious commodity.

Wow. Technical writing was an area that I can close to seriously going into as a profession. Crazy to see it made near redundant whole cloth.

AI made will be faster, cheaper and higher quality.

Well... let's give it to him!

Artificial intelligence is no more meaningful than artificial ice or artificial fire.

I'm also thinking that we're getting close to the Young Lady's Illustrated Primer from The Diamond Age.

The Fermi Paradox explained.

That was a really fun story.

I think everyone is going to fall in love with large language model simulacra and stop dating, having kids and interacting with each other.

Knowing that the world may change dramatically or even end in the very near term means that everything you do needs to be for its own sake. Don't despair and don't drop out but equally, don't waste any time on things you dislike and are doing based on some expected future utility.

Personally, I just find myself really feeling lucky to have my family and I'm determined to appreciate every day I have with them - which is what I should be doing anyway!

That's... really profound.

I had never considered the possibility that our version of intelligence might be the flawed, impure one.

Of course... that's why I would never claim that a parrot had mastered language. It may know words but it can't use language in creative, communicative, problem solving. LLMs can.

Language mastery is a function of communication and problem solving ability in that language. Understand should be judged based on results not some mysterious inferred grammar understanding.

I would call it a sign of meta-cognition which is something that I don't think LLMs have at the moment.

Couldn't multimodal models capable of incorporating realtime non-textual (visual, auditory, kinesthetic, etc) data be a solution?

The current generation have pretty much mastered language anyway so more text seems kinda redundant anyway.

The inability to experience doubt.

Great article. It articulated what I've been thinking but couldn't quite put into words.

Sorry, I can't hear you over the sound of goalposts being moved.

Every time someone makes a supremely confident prediction like this, machine intelligence overtakes another domain previously sacrosanct to humans.

Well, Japan isn't politically radicalized so that may have something to do with it.

On a beach sipping strawberry daiquiri, hopefully.

I mostly just don't want to go to work next week.

You are not weird and you are not alone. I can't say whether it's good or bad but I 100% expect the majority of people to unironically consider an AI their best friend by 2030.

My money is on Gato being or being closer to "true" AGI than anything else at the time it's made public.