Even the smaller subreddits on niche topics seem to be dying. I suppose the lost good will (and loss of third party mod tools, or course) caused many of the superusers to leave.

Illustrators. Actors. Animators. Writers. Editors. Directors. Let’s make art impossible to sell so we can get back to proper starving, errr… I mean… making art as a form of expression rather than commerce.

I think it’s more likely a compound sigmoid (don’t Google that). LLMs are composed of distinct technologies working together. As we’ve reached the inflection point of the scaling for one, we’ve pivoted implementations to get back on track. Notably, context windows are no longer an issue. But the most recent pivot came just this week, allowing for a huge jump in performance. There are more promising stepping stones coming into view. Is the exponential curve just a series of sigmoids stacked too close together? In any case, the article’s correct - just adding more compute to the same exact implementation hasn’t enabled scaling exponentially.

I did some source digging to hopefully best address your observations. Science journalism (even when internal and likely done in concert with the authors) is fundamentally a game of telephone. But looking at the source papers:

They say it in an incredibly formal way, but they do seem to come to the conclusion that the LLM develops understanding. The paper makes that case within an incredibly narrow context, but it does include:

We anticipate that this technique may be generally applicable to a broad range of semantic probing experiments. We argue that the observed semantic content cannot be fully attributed to a retrieval-like process, and instead requires the LM to perform some degree of generalization over the semantics. More broadly, we see programs and their precise formal semantics as a promising direction for working toward a deeper understanding of the behavior of LMs, such as whether or how LMs acquire and use semantic representations of the underlying domain more generally.

With it now clear that the generalized case is not shown: the specific type of understanding that they have shown is non-trivial.

Conclusion: This paper presents empirical evidence that LMs of code can acquire the formal semantics of programs from next token prediction.

A foundational topic in the theory of programming languages, formal semantics (Winskel, 1993) is the study of how to formally specify the meaning of programs.

From Winskel: The Formal Semantics of Programming Languages provides the basic mathematical techniques necessary for those who are beginning a study of the semantics and logics of programming languages. These techniques will allow students to invent, formalize, and justify rules with which to reason about a variety of programming languages.

Also notable but unrelated: Jin and Rinard’s paper was supported, in part, by grants from the U.S. Defense Advanced Research Projects Agency (DARPA).

You’re a moron.

Pres·i·dent /ˈprezəˌdent/

On a scale of 1-10, with 1 being Inconsequential and 10 being Paramount, how important is hierarchy to the healthy functioning of a group?

I honestly don’t know why more people aren’t excited about A.I. succeeding us.

Claude for President

As a Large Language Model, Better not tell you now

Keep it in your pants, I’m still voting for the incumbent geriatric. That does not change the fact that Biden demonstrably should step down.

Creating the narrative of seeing Biden with my eyes and hearing him with my ears? There’s no way he’s good for another 4 years. I’m all for him resigning right now. I’ll vote for Kamala.

Yes. Replace him with AI.

Plants with more flexible and responsive genetic systems were better able to adapt to changing environments and thus more likely to survive and reproduce, so yeah. However, the basic building blocks of these systems - DNA replication, gene expression, and the fundamental biological processes arose from simpler chemical and physical interactions that were likely governed by principles of self-assembly and thermodynamics. The primary drivers are different at different levels of abstraction and complexity, and there’s dynamic interaction across levels.

Thermodynamics -> Natural Selection -> Responsive (Epi)Genetics -> Memetics -> Metamemetics (probably?)

We “boil things down” to Natural Selection or Thermodynamics as is convenient for communication, but the higher levels affect the lower as well. So we can’t really reduce them like that without losing important information.

In our effort to disillusion people of the idea that evolution has a purpose or conscious hand, we over-simplify things, though. Plants actively (but not consciously) shape their own evolution through complex molecular and genetic mechanisms. They can respond to environmental stresses by altering their DNA methylation patterns, potentially priming future generations for similar conditions. Plants also engage in niche construction, modifying their surroundings in ways that influence their evolutionary trajectory. For instance, they can change soil chemistry through root exudates, creating new selective pressures for themselves and their offspring. Plants participate in intricate co-evolutionary relationships with pollinators, herbivores, and other organisms. These interactions create dynamic fitness landscapes that drive reciprocal evolutionary changes. While not “inventing” traits in a deliberate sense, plants possess sophisticated genetic tools - such as whole genome duplications, transposable elements, and adaptable gene networks - that allow for rapid evolutionary innovations. These mechanisms enable plants to continually adapt and evolve, even without conscious intent or direct feedback.

deleted by creator

Honestly, I’d get on-board with just about anytime 2000 to 2010. The enshittification of the internet and social-media-driven comment culture didn’t start in earnest until smart phones took off.

The year 2000 was peak human technology. It’s been downhill in every way since, until generative AI - which is f’in amazing. But let’s be real, the future belongs to the bots.

/stares in smart glasses

I asked Claude to re-write the beginning of Ulysses at a 5th grade level:

Buck Mulligan, a big man, came up the stairs. He was carrying a bowl with shaving cream, a mirror, and a razor. He wore a yellow robe that wasn’t tied. He held up the bowl and said some words in Latin.

He stopped and called down the stairs: “Come up, Kinch! Come up, you scared priest!”

Buck went to the edge of the tower. He turned around and made blessing motions three times towards the tower, the land, and the mountains. Then he saw Stephen Dedalus. Buck made quick crosses in the air and shook his head at Stephen. Stephen didn’t like this. He was tired and put his arms on the stairs, looking at Buck’s face and hair.

Buck looked under the mirror quickly and then covered the bowl. “Back to work!” he said seriously. He added in a funny voice: “This is the real thing, friends. Body and soul and blood. Slow music, please. Close your eyes, guys. One moment. A little problem with those white blood cells. Quiet, everyone.”

He whistled long and slow, then waited. His white teeth had some gold in them. Two loud whistles answered back. “Thanks, buddy,” he said quickly. “That’s good. Turn off the power, will you?” He jumped down and looked at Stephen. His face looked like an important church person from long ago. He smiled. “What a joke!” he said happily. “Your funny name is from ancient Greek!” He pointed at Stephen in a friendly way and went to the edge, laughing. Stephen followed him halfway and sat down, watching as Buck set up his mirror and started to shave.