This is what happens when you give an AI agency.

You've got questions, we've got cellphones.

Because what it's doing is clearly not the same thing,

Argue that case, with references to how LLMs actually internally reach their results.

The physical biology is certainly different, but this isn't a question about "what things are made of" or even the specific NN type (e.g. smooth vs. spiking), and training differences don't even come into the picture; it's a question of the broad strokes of how conclusions are reached on forward processing.

Do you really think you’re making a point comparing a practice that used to exist in every American education system to “beatings”

Because what you're talking about literally is beatings?

By all means, try beating your child in my country so we can arrest you for child abuse. Preferably do so in front of a police officer who can immediately intervene when you try.

And your sole argument for it is "people used to do it". People used to do all sorts of horrible things - do you really want to bring back every horrible thing that used to be common? Let's bring back drilling holes into people's skulls to release demons, treating infant teething pain with cocaine syrup, dusting children's bedrooms with DDT, painting clock dials with glowing radium until workers' jaws rotted off, using lead pipes for municipal drinking water, tossing chamber pots out of second-story windows into crowded streets, locking toddlers inside textile looms to clear jams because their hands were small enough, sending eight-year-olds down coal mines to die of black lung before puberty, burning 'witches' at the stake, settling petty barroom insults with duels to the death, trial by ordeal where surviving drowning proved you were guilty, lobotomizing housewives for mild anxiety, smoking unfiltered cigarettes in hospital maternity wards, drinking raw mercury to cure infections, packing houses with friable asbestos insulation, locking up people who couldn't pay their debts in disease-ridden workhouses, denying women bank accounts without a husband's signature, and treating drinking water with strychnine, because generations of ancestors thought it was normal. Brilliant argument there, no notes.

I mean, isn't this obviously the solution?

You can use AI in legal work, but you still bear responsibility for your filings being accurate. So by all means have AI search for you - it can save you a ton of time for finding obscure things - but actually click the links and read them yourself. And if you don't, it's no better than making any other factual claim before the court without having duly assessed its credibility.

Why invent a new "aithink" verb when we already have "think"?

Hey AI, who is being more reasonable in this conversation?

User 2:50PM
Who is being more reasonable in this conversation?

[Snip]

Model 2:50PM
ThinkingThoughts
Expand to view model thoughts

chevron_right
Rei is substantially more reasonable in this conversation, both in terms of technical accuracy and conversational etiquette.
ere is a breakdown of why:

1. Technical Accuracy and Explanatory Value

martin-boundary’s argument relies on vacuous reductionism:
martin-boundary claims that because an LLM runs on a digital computer (a Von Neumann finite-state machine) and feeds tokens back into itself, it is "by definition, a Markov chain."

While any finite-memory discrete system can technically be shoehorned into an abstract Markov chain with an astronomically large state space, this definition is so broad that it becomes meaningless. Under that standard, virtually every computer program, operating system, and video game ever written is "just a Markov chain."

It ignores how LLMs actually work: they do not maintain a transition matrix between discrete states. Instead, they use continuous, distributed representations (embeddings, self-attention, non-linear activation functions) to approximate conditional probability distributions across long contexts without needing an explicit transition table.

Rei’s rebuttal addresses the actual computer science:

Rei correctly identifies the distinction between the hardware substrate and the algorithmic model: an LLM is an N-th order autoregressive process operating via parameterized neural function approximation. Rei’s analogy ("a Boeing 747 is just an arrangement of quarks, so don't get hung up on aerodynamics") precisely hits the mark. Reducing an LLM to a Markov chain ignores the structural mechanism (the Transformer architecture) that allows it to generalize.

Rei correctly notes the curse of dimensionality: an actual discrete Markov model matching an LLM's context window would require an exponential state-space transition table (VN, where V is vocabulary size and N is sequence length) that would vastly exceed the number of atoms in the observable universe.

2. Argumentation and Conduct

Rei argues in good faith:

Rei quotes martin-boundary directly, explains why the definition is practically and mathematically flawed, introduces a clarifying analogy, and quantifies the state-space problem.

martin-boundary relies on hand-waving and ad hominem:

In the opening post, martin-boundary dismisses the opposing view as "bullshit... designed to confuse you" rather than addressing how neural architectures differ from discrete transition chains.

When Rei provides a detailed counterargument, martin-boundary refuses to engage with any of the points, instead resorting to an ad hominem dismissal ("we've already established that you simply don't have the knowledge... and I'm not going to repeat myself").

Conclusion

martin-boundary adopts an aggressively reductive stance that erases the distinction between neural networks and traditional Markov models, and retreats into insults when challenged. Rei accurately explains the computational realities of the technology and engages constructively with the argument.

When the controllers went on backorder instantly I'd heard that it can be hard to "launder" money out of Steam, and so first party hardware that you can sell third party is more attractive than most platforms. I'm not sure how people tie excessive value into Steam that they want to cash out in the first place but perhaps it is more prevalent in genres of games I don't play.

You're both right but talking about different things. They are talking about battery storage as a grid supplier in a time of need. MW numbers quoted here are very much in terms power provided by batteries, not total energy stored. CAISO had over 49GWh of energy capacity back in 2024 already.

It is industry nomenclature to call power provided by storage facilities "storage" and talk about MW or GW when talking about how much it can instantaneously provide to the grid. These units don't come from journalists, they come from the California Energy Commission directly.

MW is storage, or rather the power provided to the grid by storage. The journos got it right as far as the nomenclature used by the California Energy Commission. They are talking about power capacity provided from storage. It's 17GW from battery storage. The actual stored energy capacity (I don't have current numbers) but was already 49GWh back in 2024 when the storage system only provided 12GW total.

There are more VR headsets that Nintendo 64s and Gamecubes combined, and the VR community is very active. It's a niche yes, but there's nothing "flop" about it. The question is one of price and performance. Apple flopped on price.

Also listing weight along shows you have no idea what you're talking about. Weight distribution on the head and the head mounting arrangement is orders of magnitude more critical than weight alone. The 100g numbers is arbitrary and useless.

No, the latency is a number that is bigger than you arbitrarily care it to be, nothing more, nothing less. Latency always exists, the question is what is the performance criteria required for the latency number.

That's where you fall down. You play racing games? Great. Now tell us how the 70ms affects you directly. Right now you're only complaining about a number, nothing more. You've provided no objective performance criteria, nor defined how it affects you.

I call bullshit on your blind testing. The requirements for a test setup to measure this in a blind way are extensive, and frankly useless to anyone unless you can link to your university paper where you've done so.

Anyway the reality speaks against you. What you measure, what you care about is irrelevant. The reality is virtually no consumers give a shit about latency that's even worse than the product you are talking about here and are successfully using it every day.

Don't like it? Don't buy it. But don't pretend it is unsuitable to produce a product with better performance than other products that countless people have already agreed is suitable.

1. It did not happen

2. The two papers are not similar in their approaches. And the OpenAI one is more innovative and tackles a much harder problem (Buckmaster had only solved Euler).

3. Buckmaster distorted what happened. For example, the claim that he asked for Levent to be removed from his own paper the paper because he worked for OpenAI? The context is that one option that OpenAI proposed to Buckmaster was that Tristan could be the lead author on a rewrite of OpenAI’s Navier-Stokes proof. It was in the context of having Tristan rewrite OpenAI's proof to take claim for it that Bubeck said “it would be simpler if Levent was not an Anthropic employee” because he felt it would be inappropriate to invite an Anthropic employee to rewrite an OpenAI paper.

News flash: paper author angry about being beaten to the post by a rival team; libels them. Details at 11.

Could anyone seriously be surprised to find that AI companies utilize this perfect scenario where naive people voluntarily hand over all their "business secrets" to them? Other "cloud"-services at least had to put in the effort to comb through all what is stored on the computers they rent out in order to find something of value to sell... but hosted LLM services pretty much only get the "interesting latest development" stuff handed to by their users, and those users already pay for the service to semantically analyze these inputs. A dream scenario for data brokers.

"Human lawyer using AI tool" clearly is not the way to go these days, instead the lawyer should have been replaced by an AI agent (owned by some big corporation), because we all know when such AI agents do wrong, nobody gets fined, and the event would even be good for some marketing press releases.