There is a long history of artists "pushing the envelope… by self-deprecating and making “statements” about how ridiculous the market for art is.

You don’t get to shove Dada, a Campbell’s soup can, canned “artist’s shit”, a black square on a canvas, a pile of trash, a shirt on a coat hanger, a self-shredding copy of a graffiti, and thousands of similar examples, all under the same label as the Louvre, illustrators, writers, etc… then magically have people value all artistic work.

There has been continuous, serious, self-damage done to the concept of “art”, and now come the consequences: the general public, thinks the value of art is close to $0, and that’s what they’re willing to pay.

Coincidentally, AI companies allow anyone to get “better than literal shit” art done for $0… so… 🤷

Customer perception of the value of art in general. You don’t get to call it all art, without muddying the perception.

“artists don’t deserve a living wage” or “they don’t do real art”

Blame the “Modern Art” movement, where a black square, or canned literal shit, can be worth millions.

At some point, people realized that they don’t need real art, just to decorate an office or house space with “artsy looking” stuff, so they went on a chase to the bottom. That used to be random people from third world countries working for peanuts… now it’s LLMs and GenAI working for fractions of peanuts.

Same thing has been going in all areas. Who needs a slab of real ice (from a mountain) for their fridge, when they can get fake ice much easier, for a fraction of the cost.

The new kernel.split_lock_mitigate knob, if set to zero, will disable the penalization of processes using split locking (while retaining the warning sent to the system log)

Sounds to me like it’s fixed. WINE could follow dmesg, and show a popup with recommendations when it detects one of its processes is getting throttled.

There are two sides to that story.

There is not enough gold to match the increases in both population and productivity of the last 70 years, and you don’t want just a handful of people holding gold that spikes in value through the roof.

Smart people invest in companies that pay dividends. Speculators invest in… whatever, tulip bulbs.

There is a reason why people keep asking “How do you spell it?” when being told a name in English. The counterpart is, “How do you pronounce it?”.

Even with “long a”, I still can’t tell how would you want to pronounce “Rach”. I can come up with 4 different pronounciations right now: “Ra-ah-ch”, “Ra-ah”, “Ra-sh”, “Ra-kh”.

It’s not about capitalism:

• 1 human can talk to 1 human
• 1 chatbot can talk to 8 billion humans

Human therapy will be more expensive, for as long as we value human time more than machine time.

Don’t like Fortnite? Think Roblox is bad? …and all the remaining points?

Wait until you see the future of gaming, coming to Xbox:

Muse, a Generative AI Model for Gameplay Ideation

https://news.xbox.com/en-us/2025/02/19/muse-ai-xbox-empowering-creators-and-players/

The complaint is: Narcissistic incompetent dev spreads FUD while putting vulnerable people at risk.

There is an experimental distributed open source search engine: https://dawnsearch.org/

It has a series of issues of its own, though.

Per-user weighting was out of the reach of hardware 20 years ago… and is still out of the reach of anything other than very large distributed systems. No single machine is currently capable of holding even the index for the ~200 million active websites, much less the ~800 billion webpages in the Wayback Machine. Multiple page attributes… yes, that would be great, but again things escalate quickly. The closest “hope”, would be some sort of LLM on the scale of hundreds of trillions of parameters… and even that might fall short.

Distributed indexes, with queries getting shared among peers, mean that privacy goes out the window. Homomorphic encryption could potentially help with that, but that requires even more hardware.

TL;DR: it’s being researched, but it’s hard.

The basic algorithm is quite straightforward, it’s the scale and edge cases that make it hard to compete.

“Ideally”, from a pure data perspective, everybody would have all the data and all the processing power to search through it on their own with whatever algorithm they prefer, like a massive P2P network of per-person datacenters.

Back to reality, that’s pretty much insanely impossible. So we get a few search engines, with huge entry costs, offering more value the larger they get… which leads to lock-in, trying to game their algorithms, filtering, monetization, and all the other issues.

There’s a good commentary about that in here:

AWS CEO Matt Garman just said what everyone is thinking about AI replacing software developers

“That’s like, one of the dumbest things I’ve ever heard,” he said. “They’re probably the least expensive employees you have, they’re the most leaned into your AI tools.”

“How’s that going to work when ten years in the future you have no one that has learned anything,”

https://www.itpro.com/software/development/aws-ceo-matt-garman-just-said-what-everyone-is-thinking-about-ai-replacing-software-developers

US-based nodes

Tor has nodes all over the world: https://tormap.org/

Probably more helpful to say “Stop using VPNs to watch porn”… helpful for VPN providers’ sales, I mean.

Education is supposed to teach “how to learn to learn”.

Left to his own devices, then, without knowing quite what to ask or how to interpret the responses, the man in this case study “did his own research”

The whole thing with “do your own research”, is kind of funny:

• some use it to avoid explaining their points
• others use it to come up with a lot of nonsense
• while the proper way to begin any “research”, is to… ask an expert.

Nobody has ended up in a psych hold, just by reading a bunch of Wikipedia articles, asking ChatGPT… then consulting a doctor.

Good news: advances in medicine have reduced “physical” natural selection so much, that “intellectual” natural selection is overtaking it.

Now, if only all countries could say the same.

Keywords: NPU, unified RAM

Apple is doing it, AMD is doing it, phones are doing it.

GPUs with dedicated VRAM are an inefficient way of doing inference. They’ve been great for research purposes, into what type of NPU may be the best one, but that’s been answered already for LLMs. Current step is, achieving mass production.

5 years sounds realistic, unless WW3.

If the current rate of execution of Project 2025 is an indicator… no, there are no “decades” left.

chain-of-thought models

There are no “CoT LLMs”, a CoT means externally iterating an LLM. The strength of CoT, resides in its ability to pull up external resources at each iteration, not in dogfooding the LLM its own outputs.

“Researchers” didn’t “find out” this now, it was known from day one.

As for who needs to hear it… well, apparently people unable to tell apart an LLM from an AI.

It still is: https://www.google.com/search?q=tic+tac+toe+ai

Plenty of examples out there.