The computer version of this would read something like:
> When any would-be innovator can grow and etch a silicon CPU in their kitchen and put it to use in any way their imagination dictates, it might be the seed of the next big set of changes—something revolutionary rather than evolutionary. But until and unless that happens, there can be no irruption.
Put that way, it seems clear this whole reaction is a category error.
When VCs started investing in eg former Bridgewater employees over tech founders, the writing was on the wall.
It is a status club now, and giving a shit does not make you one of the cool kids. Giving a shit is papered over as an “ego spike” and cool kids don’t have those, because they make sure not to take the risk of insisting on anything.
> In our early evaluations, it matched or exceeded Gemini 4 Carbon on overall quality at 3x the speed and 60% of the cost. We’re excited to bring these capabilities to Glean customers tackling multi-step work that demands accuracy, detail, and speed.
So possibly smaller than Carbon? Or some other axis that is making it faster and cheaper? No idea.
...first runs are in, and Neon is far below Carbon in terms of quality, but indeed much faster - ~4x faster, and 2x cheaper per task. Probably a good model for delegating to after key design work is done with something cleverer.
What category of work did it underperform on for you? I am seeing comparable outcomes in some technical document processing, but the cost is not lower for me.
Gold plating for tasks where the agent is given the intention, the explicit ask to make the best impression, and ways of self evaluating. Much like opener's agents pre 6.2 sol/astra, it considers the task satisfactorily completed a bit earlier than anthropic models which seem to better guess what the specs left out. It is a useless feature for those who keep their llms at a short leash - my hobby benchmark aims to evaluate performance at under-specified tasks and, well, vibe coding as it was described a year ago.
It's a good model for delegating work to, but not deciding the detailed scope. So, not fable class for me.
Yeah they seem to be chasing a classic integrations and sticky-identity playbook. But it is weird this time because the product itself can trivially recreate or bypass the moat. Phones did not happily and actively unravel their makers’ competitive differentiation when asked to do so by their users.
In Nov/Dec 2025, just prior to the training cutoff for the first model to show this capability (Opus 4.7) I was working on solving arc-agi using claude code and was exploring using protein data and other adjacent data to improve performance.
In January it seemed very clear the subscription models behind claude code became less competent, there was a whole series of posts about it. I stopped working on it as a result.
"Treat[ing] the media as trustworthy on subjects you don't know anything about" and "trusting newspapers for topics which [you are] not knowledgeable about" aren't tricky phrasings that only turn out to mean the same thing if you take a minute to think. They are the same idea stated in almost exactly the same words.
You can't display Gell-Mann amnesia by noticing that the media isn't trustworthy. You can display Gell-Mann amnesia by trusting the media. (And you can't do it by any other means!) That's what I said, that's what wiktionary says, that's what you're repeating here, but you seem to believe that you're saying something entirely different.
From your phrasing I thought you were saying you didn’t need to forget first, but really
the issue is I was only reading the immediate parent.
The poster of the original comment, you are right, is probably just waking up to the discernment that yields the potential of Gell-Mann amnesia, but has not yet reached the full expression.
I scanned the video and transcript and it argues (poorly, imo) that competition produces parity and mediocrity.
I also don’t really see how it applies here — AI in this case is removing what most people would understand to be competition. Do people driving cars still compete on running speed?
> AI in this case is removing what most people would understand to be competition. Do people driving cars still compete on running speed?
We're not talking about business competition, we're talking about human competition. Species level, subconscious, and driven to seek (perceived) safety in the hierarchy amongst other humans.
It is interesting I think a hard societal split is emerging between those who understand AI is a tool that can be used and can use it, and those who embed themselves in the view that AI itself competes with them as a person.
For the majority of tasks, the best models are no longer required.
Many benchmarks are already saturated today on many tedious tasks still being done by humans across the world. Frontier labs are introducing their own budget options prior to IPO, undercutting their revenue, because otherwise they risk losing seats.
Choosing what to do and how to do it is the actual frontier now and is still not apparently well supported in any model. Current so-called decision models are not decision models, they are available-fact-evaluating models.
> Many benchmarks are already saturated today on many tedious tasks
Exactly my point, think the easy things won't even need 3rd party apis in the near future.
But frontier models only seem to need more and more VRAM to keep winning the race. Feel free to look up the cost of running Kimi K3 open weights today, probably the best estimate known publicly, it's hardly a casual affair.
It is interesting that it can do it, especially given how bad earlier models were. But you did appear to walk it through steps it probably would not have selected on its own, from two-tier spice/behavioral modeling to substrate to layout. Those steps appear to have been instrumental to its success. Did you try to one-shot it and it failed?
> When any would-be innovator can grow and etch a silicon CPU in their kitchen and put it to use in any way their imagination dictates, it might be the seed of the next big set of changes—something revolutionary rather than evolutionary. But until and unless that happens, there can be no irruption.
Put that way, it seems clear this whole reaction is a category error.
reply