lettergram

I'm running an M5 max with 128Gb of RAM right now on a 14" macbook pro

Am I reading this correct? Max of 32Gb of RAM?

> With 16GB of unified memory standard, configurable up to 32GB, and higher memory bandwidth up to 170GB/s, MacBook Pro makes multitasking and working with demanding apps and large projects even faster.

So they're basically removing the ability to run local models effectively. I wonder if they'll bifurcate their market to enable those.

I actually wrote about this recently. There's a quote somewhere along the lines of:

> What is possible is unevenly distributed. As knowledge and technology compound, the distance between those closest to the source and everyone else also exponentially increases.

I actually argue the continuum that forms singularity has been progressing before we realized, but we're at the point where in 5-10 years people not using AI will be equivalent of the amish.

Unfortunately, as the rate of change increases, the distribution of knowledge wont be as fast. So those closest to the source of innovation will have exponentially more knowledge, capability and in-effect power.

https://ipcopilot.ai/blog/the-singularity-already-happened

It's not "corruption" if it's the way it should be done. Corruption would imply some sort of nefarious means.

> appoint expert groups to collect information

So... you'll have the representatives make their own lobbying groups? Also, each "expert" will have their own bias, so in doing so you'll inadvertently (or intentionally) block hearing from the wider community.

In the past I FOIA requested 100 universities for their grades - https://austingwalters.com/foia-requesting-100-universities/

There are some really hilarious responses when asking for student grade distributions (i.e. records they 100% have):

> Purdue University does not have such a record in existence. One would have to be created in order to fulfill your request. Indiana’s Access to Public Records Act puts no obligation on a public agency to create a record in order to satisfy a records request. Therefore, Purdue has no records to provide.

> The University of Utah received your GRAMA request. In that request you seek grade distributions for all formal courses at the University of Utah for the last five years. You requested this information in a digital CSV format. We can only provide a paper format. The report for the last five years is 9,714 pages. The reasonable copying fee for documents responsive to most GRAMA requests is $.25 per copy. Therefore, your cost for this report will be $2,428.50.

Often you can request it in the format they have available, etc. You can also apply pressure by submitting a different request, i.e. all the invoices or secondary records, then have AI scan them.

The first example I saw was making a shopping list for school for pick up.

Do the same thing for recipes and buying groceries.

Or "order me take out from a bunch of places I haven't been that you think I'd like." The AI can ask follow up questions and see your ratings.

"Make a music playlist for my daughter when she enters the car based on the conversation last month. I remember vaguely her mentioning what she liked"

These are things that 100% people want AI to do -- imagine if you had a personal assistant, is something people have asked for since forever.

Now, I kind of agree on the minimal use compared to business, because in business everything has always been highly integrated. That said, that's changing rapidly.

Literally every campaign promise he's went back on. So, "Saying that maybe this is actually what it means to be an American could hold some truth" I don't really think is accurate lol

I know a lot of folks who voted for the man, almost all of them would love for a competent alternative. Unfortunately, in the last election, neither were competent, so a lot of it was a coin toss.

If you like this, you may like the "Tasting History" YouTube channel - https://www.youtube.com/@TastingHistory

He basically covers a different historic dish every week ranging from the oldest recorded dishes to cafeteria food from 1990; or what they served on the Titanic, etc.

It's a great channel, and covers history well.

Judges at this level often do things like this, and given it was brought in California likely the next level would also block the administration. That said, the Supreme Court has been pretty clear that the executive branch has broad and the only authority to designate supply chain risk. That said, they do need to justify it, not sure they did it well enough here, but this is far from an impartial judge.

Also, judges are political appointees -- This one has ruled against the Trump administration repeatedly and it's why this case was brought here. She was appointed by Joe Biden, was born/lives/works in the Bay Area and has a track record.

This would likely be overturned if the federal government needs to push it. At the time Anthropic was saying publicly that they oppose certain actions and their model is restricted in certain ways. The executive branch viewed that as a risk and has the ultimate say over it's vendors, regardless of the judges ruling. The supreme court has been pretty clear on this.

This is also a reminder that, contrary to popular understanding, judges are indeed political appointees. This one was appointed by Joe Biden, works in the Bay Area and definitely has a political bias in both the track record and in activism.

https://grokipedia.com/page/Rita_F._Lin

Ah yes, the claim free money would save lives!

The truth is once systems like this are implemented they change the dynamics of the study and as such could hurting lives.

As a simple example, the biggest cost to the healthcare system in the US is dialysis. Many of those folks already have health issues, and by both dialysis and for all the issues that caused the need for dialysis in the first place is a massive strain. The direct cost of dialysis is $33B annually [1], but indirectly is probably at least another doubling.

Further, the way the government has imposted monopolies in the space already has caused costs to balloon. While making it "universal" could mean the government gets control, and can limit the explosion. It can also could result in weird incentives like Canada, where they're promoting MAID.

[1] https://pmc.ncbi.nlm.nih.gov/articles/PMC8729831/

I think you’re missing the real purpose vs the stated purpose.

The point is monitoring everyone, age verification requires you to share who you are.

After reading the story, it seems like lack of training. I think training should assume GPS goes out (which it does), in this case they didn’t pull up when they saw “lights going out” as a mountain became between them and the airfield.

Air traffic control also failed here.

Objectively?

> Democratic Party platform is aligned with national polling on every one.

It depends on the pollster and where you're polling. I guarantee you rural Tennessee will not agree with downtown Washington DC on any of these issues. In contrast, rural California will likely agree with rural Tennessee. It's not as cut and dry as a homogeneous national poll of 2500 people. Every state, city, county are different. That's why there are federal, state, city, and county governments.

For instance, Abortions are legal nationally. States can individually decide how, or if, they wish to restrict it. This is as the constitution intends under the 10th amendment:

> powers not delegated to the federal government nor prohibited to the states are reserved to the states or the people.

This allows for democracy to take place at the local level, rather than having particular regions thousands of miles away from each other ultimately oppress the other.

To the point on LLMs, I think it's abundantly clear they will be used to propagandize and similar to social media will lock people in a bubble without alternative opinions. It'll be the worst of both worlds, the question is who's the puppet master. At some point soon, I imagine it'll be the AI.

> My guess is that, in the US, the right will cynically adopt manipulation to great effect and the left will take a moral stand against shady practices and lose elections.

I think that statement may itself highlight how prevalent manipulation is.

I fully anticipate all groups to continue maximal manipulation they can. One thing with LLMs is that it'll be a far less unified view, so a "divide and conquer" strategy is what I anticipate.

It was a false positive for what it was worth - https://www.nbcnews.com/health/health-news/fda-says-taylor-f...

Unfortunately, the AI being locked down and proprietary is the winning strategy for these companies.

My company hosts its own models. Some customers require us to use either US / EU models, while others are fine with us using any model.

As such, we have two GPU clusters, the general AI cluster runs a Chinese model as it's the most accurate and robust. The US/EU required ones have a few percentage points lower on our accuracy metrics and we provide them those that require it for an extra fee.

Why host at all? Because it enables us to get much higher margins than competitors, while reducing costs. Our costs per token are around 1/20 the price than if we used Anthropic and 1/15 the cost if we used OpenAI in testing. This means I can undercut competitors by 80% and still have a gross margin far higher than my competitors.

In reality, these US AI providers are jacking up the prices and trying to implement regulatory capture. I'm actually fairly confident they'll succeed. At some point, I'm expecting the US / EU administration(s) to block foreign based model, at the same time, they'll probably invest in Anthropic and OpenAI.

What Anthropic and OpenAI are doing is using "safety" as a wedge, just like large corporations used "environmentalism" or "food safety" or "workers safety" as a wedge to regulate smaller competitors out of the picture. Then they jack up rates, sue and/or buy anyone who can potentially be a threat. It's the #1 threat to our business model.

Our competitors are giving half of their margin over to these large AI service providers, we keep the vast majority of ours. Eventually the AI service provider will be able to squeeze them even more until the margin just isn't there and either they are purchased or replaced via internal tools at the company they sell to.

If you're self-hosting a model as a startup (e.g. using GPUs), you're almost certainly using a Chinese model. If your a startup outsourcing (e.g. using tokens), you're going to be using a US based model

As a tractor owner. Two things, the DPF & SCR (>=75hp) on a tractor is not a great idea --

1) Tractors are typically owned by low margin businesses (i.e. farmers) that need to be repaired in the field AND need to be repaired quickly, else you loose a crop. Adding complexity to tractors literally can cost the farm.

2) The actual emissions reduced is questionable. Tractors run significantly less than a truck, like 50-100x less often. Further there are at least 2x more trucks sold per year

3) To run the SCR system, the engine had to run hot for like 20 minutes burning extra fuel and required DEF (yet more input costs)

3) The emissions they are trying to reduce with the these are likely not excessively harmful from a tractor; largely because most tractors who need an SCR system is >75hp, which also means they're typically used on a large farm (100+ acres). Which dissipates the risks substantially.

For reference my 2022 Kubota tractor repeatedly had issues with the DPF / SCR system, mostly the software to enforce environmental rules. This lost us ~$20k one year due to the tractor being knocked out for a week (I was mid-cut for 140 acre hay, rained & rotted in the field post-cut).

For reference, I was very much ready to bypass the SCR system, but decided against it to keep the warranty. It had nothing to do about "right to repair", I figured out exactly how to bypass it.

The stats come from the government agency in the US which tracks driver fatalities.

Here's the source document: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/...

Regarding "distracted driving" causing cyclist fatalities, my point was more that bicyclists have a higher risk rate. If we're already going down this path for safety, banning bicycling would have a higher impact than reducing fatalities ~10% which is what the EU claimed it could prevent.

Regarding the claim that "pretty much all fatalities on bikes are the result of being killed by drivers", that's simply not true. It also largely depends on laws and location.

https://www.bikeattorney.com/bike-accident-common-causes.htm...

https://www.npr.org/sections/health-shots/2011/05/20/1364622...

> incredible amount of killing from distracted drivers

> 1.10 fatalities per 100 million vehicle miles traveled

https://www.transportation.gov/briefing-room/trumps-transpor...

For reference, fatalities on bikes:

> 9.32 fatalities per 100 million cyclist miles (6 deaths per 100M kilometers)

https://www.calbike.org/urban-transportation-research-bike-f...

We actually found the Mistral Small 4, quantized to 4bit was comparable to Qwen 3.6 27B and is roughly the same size. At least from our experience on our use cases, the quantization of the Mistral model worked far better than trying to quantize the Qwen family.

Fully agree to your point though, Mistral in general is far behind where I'd expect and Qwen in particular is crushing it at the smaller sizes.

Personally, I'd consider anything 20B params and above a "medium" model. Small being <20B and large >100B. I think obviously we can get to the huge 1-2T param models, but frankly the margin of accuracy improvement for the speed hit is kinda insane (1-2% for many metrics).

Yup, that's pretty much what I was saying for the above. I had a Kubota and can indeed attest to the issue(s).

They're both considered medium tractors with similar weights for a similar purpose. I forget the official weights, but it's something like 100-175HP is mid-tier.

You can get a kubota M5-111 with a closed cab for $70k-100k, cheaper than these. Plus zero percent financing though then for 5-7 years. Well built and a comparable class in terms of weight and horse power.

People aren’t buying them for price, but the first sentence discusses it as if it’s relevant.

My assumption are farmers are trying to skirt the eco rules for vehicles in some way. Which by the way is insanely annoying and has caused issues for all the farmers I know at one point or another. Worse, you can’t fix the ecosystems on your own so you have to get them serviced costing quite a bit and importantly putting your tractor out of commission for a while. It’s why older tractors have a premium

Clicking the link splits off the ".", which is interesting but necessary.

You can buy books on how to make and obtain chemicals on your own.

Hell here's an Internet Archive book on making explosives

https://archive.org/details/saxon-kurt.-fireworks-explosives....

If you ever chat with older folks pre-90's much of this information was accessible fairly easily. It only changed with the push by the government to crackdown on Waco, Oklahoma City bombing, militias and other related groups. There was then a campaign to make it "normal" to limit free speech on the subjects, where as these books were available before.

I think the whole thing where AI should make information less available is a difficult battle and one which I personally oppose, but do understand. Free speech and information isn't the problem, it's the people, actions and substances they create.

After the age of the internet, I think it's been a forever loosing battle to limit information, it's why we couldn't stop cryptography, nuclear weapon proliferation, gun distribution, drug distribution, etc. The AI is just another battle ground, one which, if they actually do manage to control could definitely create some walls to this information, but not stop it.

More scary, is that the AI as it becomes pervasive and stop people from asking certain questions, because they don't know they should ask... but that's unrelated to the risk of mass death.

There's still plenty of moats frankly, same moats as before. What isn't the moat is the software development time.

In our case, we're building a tool that has a moat from: integrations, multiple parties connecting, and others

It's very sticky once we get in, and has nothing to do with the software so much as legal, company policy and inter party communication