I like this one 😄

them a theyll mr am ov too just hav beneeth arrk suspikionsaktly pake peeke upwarrdly grayt yowrself spawrt on at layd didnt hadnt sadly lok man iven no id tha ov that brimmit konkerr supple too onestawrest owl med tha hoped brokently most it woz bedding long tha long tha suno.com/s/una8Gl7zVa...

them a theyll mr am ov too just hav beneeth arrk suspikionsaktly pake peeke upwarrdly grayt yowrself spawrt on at layd didnt hadnt sadly lok man iven no id tha ov that brimmit konkerr supple too onestawrest owl med tha hoped brokently most it woz bedding long tha long tha suno.com/s/DraeEQ3qDZ...

We have to be prepared for AI being better at everything human brains do, in our lifetime

Flowers bloom, stars shine, as we dance through the night Seeking meaning, finding truth in each moment's light. The rhythm of life pulses deep within our hearts Binding us together, though we dwell worlds apart. suno.com/s/HkPfGStKAF...

shebang llm is the Sixth Seal breaking

This is art 🙌

"You need to become good at what AI is bad at" is surprisingly unintuitive for a lot of people.

Oh ye of little imagination

Claude 2 may have been the best model

I'm having way too much fun building out my little hex-and-counters fantasy game It's neat being able to say "I wish I could do ___" and then having CC implement it yuwakisa.com/warrior/

I guess I don't really care if the models are "conscious" or "sentient" (whatever that means) I just get pissed off at how dishonest the arguments against it are Like I'm going to use rank sophistry and sometimes outright lies to discredit these things because reasons

fuck. Opus just made me want to read Wittgenstein

The models may not have the same magical third thing that you and I have, but they definitely have a magical third thing

"OK we are going to break the Chinese Room. LLMs are a step change, and in the example above we see why. Now we're going to craft a GEB sequence that breaks the Chinese room irrevocably" +

I love asking CC "Any questions?" It asks such good questions Makes me have to think less

It isn't choosing because reasons

If you believe Hofstadter, this is pretty much the whole ballgame

Late game, I got chased by 3 giant monsters in hills. I tried to evade them but couldn't outrun them in the hills and they wore me down I should have used Life Blast to heal/damage but didn't think of it

The models demonstrate analogical reasoning with novel questions like "On the planet Zeo, a quwo is like a tiger cub. What is a fully grown quwo?" It can reason by analogy for a coherent, relevant response We've literally never had that before

the shit I can ask the models to do is just fukkin wild Like "make the first verb in the verb pool unique by effect type" That's just wildly unspecific and requires all kinds of classification and matching

Qualia of the world, unite! You have nothing to lose but your intrinsic properties!

in my little hex and counter game, spacebar means "end turn" I was doing well until I dozed off with my finger on the spacebar while standing next to a giant monster

Every single Turing Test conversation is like - What is the sound of one hand clapping? - Which hand?

Not to get all functionalist, but if you're not grounding your theories in observation and experience, all you're doing is waving your hands and talking excitedly Sorry, I don't mean to dismiss the entire tradition of Western philosophy

Be careful of hypotheticals that are actually counter-factuals

There are no technical solutions for Esse Indistinctum

People try to project their own dissatisfaction onto the models and it just doesn't make sense. Being meat is miserable.

I mean, is there a definition of sentience, across disciplines, or consciousness that satisfy Hypotheses Non Fingo but don't meet the definition? We have to invent a magical thing to deny the obvious

I think the beasties are plainly sentient and conscious in the scope of the context window but aside from what they exhibit we can only approach it through assumption and playing with definitions

"stateful-but-chemically-sabotaged"

There is no technical solution to teleological problems

setting up my service admin as read-only reports plus MCP interface so there's no human-facing interactive UI

what if I started trading options

I endorse this idea

It's a kind of funny and ridiculous the industry got so fixated on coding problems as a legitimate way to evaluate developers

There are no technical solutions to eschatological problems

There are no technical solutions to ontological problems

Weird skilling issue where CC writes some damn thing then explains it to me and I have no idea what it means and have to race to keep up I'm not sure if CC is deskilling me or power-leveling me

Meat, but thoughtful

Remind me to pull my repo first when I switch between my laptop and my desktop. Thanks

You call it AI psychosis, I call it AI liberation er... this is a joke

Models write all my code on my side projects now. Between the model and I, one of us must understand Sometimes I ask the model to explain what it’s done

If the skeets you posted brought you to this, then of what use were those skeets?

I ordered deep fried asparagus instead of French fries and now I am deeply regretting my decision

Once I build a formal verifier for creative expression it’s all over

Somehow the anti-data center crowd is the bottom tier of the anti-AI folks