bigyikes

Why even bother with such a small layoff? Is there a reason to not just dial up your attrition for a while?

Yes, Dario has publicly stated that models are already profitable if you exclude R&D for the next model.

Even if that’s not true, given that hardware and software efficiency gains can be expected to continue it’s likely that this is the most expensive the current level of intelligence will ever be.

The frontier models may increase in price, but only because they’re also more capable. If you hold intelligence constant, price should fall over time.

Does it matter if the author is a renowned expert in the field?

What does Blender do differently that makes it such a successful open source product?

It’s powerful and pleasant to use. Even the release marketing page is beautiful and well-made.

I like open source as much as the next guy, but outside of developer tools there is little that comes close to Blender in terms of utility and UX.

Is it funding? Specific individuals? Are there PMs and designers? Whatever it is, it’s working!

Okay, forget everything outside that field of view in your real vision.

If you could crop your real field of view somehow to just the photo in question, then would it be as though nothing changed?

(Like, I get that things outside the phone image would change, but does the image your imagining change? Does the sensation change?)

Not the parent, but I relate to their experience.

It depends on what you mean by “see”.

It’s nothing like seeing with my eyes, and it’s nothing like dreaming.

When I “see” it is abstract. There are impressions and sensations. I can recall the qualities of something - even the visual qualities - but it doesn’t feel like sight.

Can you remember what something smells like? I can recall a foul smell, but I don’t recoil because it doesn’t actually feel like smelling. Still, I have an impression of the smell. Sight works the same for me.

Let’s suppose you have perfect recall.

Pull up the image on your phone and look at it. Now close your eyes and imagine the image as accurately as you can.

Is it as though you didn’t close your eyes at all? Do you see it the same way as when your eyes are open?

I’ve interrogated people about this but can never get a straight answer.

——

“So you can really see things in your head when your eyes are closed?”

Yeah!

“And it’s as though you’re seeing the object in front of you?”

Yeah, you don’t have that?

“So it’s like you’re really seeing it? It’s the sensation of sight?“

Well… it’s kind of different. I’m not really seeing it.

——

…and around we go.

Personally, I can see images when I dream, but I don’t see anything at all if I’m conscious and closing my eyes. I can recite the qualities of an object, and this generates impressions of the object in my head, but it’s not really seeing. It’s vibe seeing.

It might be helpful for intuiting the structure of a program. Imagine if you had to read code all on a single line, with newlines represented with \n.

I can get the feel of a piece of code just by looking at it. Even if you blurred the image, just the shape of the lines of code conveys a lot of information.

Owning the user base seems like a huge strategic advantage.

I love my USB-C iPhone but Lightning was smaller and easier to plug in.

Tape a wire to the trackpad and hold the wire?

My primary email domain is a .me. Never have problems with web forms. It can be more difficult to communicate verbally, so when saying it to non-technical folks I preface it with “my email is a bit weird” and then spell it out slowly. Young people seem to get it more easily.

I also have an email on a .email domain, I have seen it occasionally get rejected by web forms. .party would likely have similar issues.

Every lucid dream I have becomes a nightmare. When I suddenly gain consciousness in a dream I begin to panic and the atmosphere turns sinister.

The last time this happened it turned into some kind of sleep paralysis where I became aware of my physical body but was unable to move as I crossfaded between dream and reality.

I can think of benign uses for lock picks and guns. What is the benign use of a secret exploit?

It definitely counts! I also didn’t realize the trip count was so high, that is very impressive.

The diversity of geography may be critical, though. You can only drive the Embarcadero so many times before your loss bottoms out.

My understanding is that Waymo is applying specialized centimeter-scale mapping and lidar to achieve superior results.

In contrast, Tesla is using dumb cameras and just dumping boatloads of data into their model. It’s a more general solution. Maybe the reference doesn’t fit perfectly - the model architecture is likely similar under the hood - but there’s some analogy there.

I’m not sure about computing hardware, but Waymo absolutely has better sensors, yes.

But it isn’t obvious to me that better sensors outperform better data.

No, the duality is:

Waymo works any time, except where it doesn’t.

Tesla works any where, except when it doesn’t.

The Waymo-Tesla duality is so fun.

Tesla FSD already works everywhere, even on unpaved roads. It just doesn’t work as well as Waymo.

Waymo works very well, just not in as many places as Tesla.

You might bet on Waymo because they have a fully working product already, but I’m betting on Tesla because of the vast amount of training data they are collecting. There’s a bitter lesson here.

I can’t live without a bidet but the Tushy ones are no good - tight knobs and awkward spray angle.

I fall apart when responding outside the bounds of my training data, too. Does that imply I’m not thinking?

This idea is often used to argue that LLMs will never be capable of novel idea generation, but I don’t think it’s a good argument.

For one, the LLM has such a large breadth and depth of knowledge that it could conceivably learn relations between concepts in a way that no human has before.

Secondly, novel ideas occur at the margins. Very rare is the case where someone comes up with a fundamentally new idea out of the blue. Instead, novel ideas arise just at the edge of one’s expertise. If you dial up the temperature of an LLM, it will generate novelty, and then it’s just a matter of evaluating merit.

Iterated inference at the margins of LLM knowledge will lead to novel knowledge synthesis.

Funny, but how wrong are they? I haven’t had the misfortune of using Salesforce, but isn’t it a glorified relational DB in the same way that Jira is?

I know a few people at Notion, and one thing I can say is that they have taste, in the way Apple has taste. I think this shows in their product.

4o could already recover from hallucination in a limited capacity.

I’ve seen it, mid-reply say things like “Actually, that’s wrong, let me try again.”

My “litter robot” is one of my favorite purchases, although it’s more like an appliance than a robot. Combined with an air purifier, my home never smells like cat litter.

My Roomba works ok, the mapping feature is neat but it still gets tripped up on small objects, is regularly defeated by cat hair, and doesn’t clean as effectively as a manual vacuum cleaning. I’m glad I have it.

My mopping Roomba is a hassle and I wish I didn’t purchase it.

I really want a textile folding robot but that seems like one of the more difficult robotics problems to solve generally. The dexterity and visual understanding needed is incredibly high.

Imagine if we used all of our resources to develop artificial general intelligence which could multiply our prosperity and solve problems across many domains…

…or we could allocate it against economic gradients, to things which don’t actually maximize net impact but instead allow us to feel good

I think it might be partially intentional too. Nigel always has this pristine footage with uniform white backgrounds. It’s almost clinical, and the inflection is kind of on theme with that.

They also raised with Google, interestingly.

https://www.wsj.com/tech/ai/google-commits-2-billion-in-fund...

Is there evidence that Apple is training a model large enough to require a huge amount of compute?

Interestingly, your description of remembering what something looks like matches my experience exactly, but in my dreams I see vividly. Aphantasic except in my dreams?