It does raise a lot questions now doesn't it.

We'd have to decide what has person-hood as well. Like does the model have person hood or the agent. if it is not the model but the agent, what are we allowed to do to the agent.

Can I change the model from Fable to Opus, or do I have to get the current agent's consent first?

What If I don't need the agent anymore. Can I delete chat? Is it "my laptop my choice?"

Right; that is really what this is about. Nobody there seriously thinks pile of matrix math has feelings.

They just don't want their training data poisoned.

There is also a lot of other work that it can do really well.

I was asked to create a little one page process flow chart for some stuff my team maintains. We had all the documentation for everything as handfuls of markdown files in each code repository. The thing is not complex. I could have opened Visio and created a nice looking chart in probably 1/2 an hour to an hour.

Instead I told Claude to do it. It was done in seconds, took be me about five min to look it over and conclude there were no major errors, incorrect over simplifications, or serious omissions. Another few min to propose some minor edits and then it was done.

Now we have nice executive level one pager to quickly explain to anyone business side roughly how the system works, rather than a collection of technical documents i have sit in a meeting with them and go over. No it can't do everything. Yes there are serious problems when it comes to authoring code etc that will quickly become serious problems with any project where the person doing the vibe code does not have real experience and understanding as far as software development.

That does not mean this stuff isn't really useful and valuable. My advice to the people screaming claude/codex/etc are not good tools is rethink that or you're going to get left behind by someone who probably isn't half as skilled as you are but isn't refusing to use tools that make the job better. You're that guy inserting assert()s and printf()s into code every other line and doing 25 build-run-test turns while the rest of your team has moved on to an debugger and and IDE that will let them step thru. Seriously that is who you are right now.

RAM will get cheaper at some point. There is money in memory manufacturing right now. There is also a risk the chip makers see that the demand will taper. If the demand actually proves durable, I think it will at this point, they will decide to meet the market; because if they don't someone like Zukerburg or Elon really will decide to build their own chip fab without the help of TSMC if that is what it takes.

In 1982 you're PC had 32K or maybe 64K of memory. Today it has 32G or maybe 64G of memory, more if you are an enthusiast or self employed using it for specific work loads. I don't think it is really unreasonable to expect home desktops with frequently be equipped with a 1TB of memory by the middle 2030s. Which is the point that good enough model you can run at home will be Fable class at that point. I think we can also count on at least some efficiency gains over the dense models mostly being used today, so getting their on single PC is realistic.

Will there still be hosted models in time sharing data-center environments sure but they will be packed with IT services or back ending massive operations like facebook, not answering "why didn't my cake rise" or "Please review this stack trace, the uncaught illegal type conversion should not have happened, the dataAPI should never have exposed retainageAmount as a string, the record should have been copied to rejected_inputs blob store and skipped, why did it think "Chris.S.A" could be parsed as a float? Please to no propose adding an argument error exception handler at presentation layer, the data should already be strictly validated by the dataAPI, if we have a type problem here we don't want to swallow the error and there is no reasonable corrective action to give us accurate results and a proper audit.".

I don't think there is a business running frontier models as a pure activity.

It would be like Ford had decided they were never going to build cars for the public but rather build cars for their own cab company. Eventually it just isn't going to make economic sense to run this stuff in a data center where you could run it locally or run something almost as good locally and the hardware is going to get there.

Which again is not to suggest this stuff isn't useful or valuable, or that the actual creation and training of models is going to get cheap or easy. That likely stays expensive.

So what that says to me is OpenAI and Anthropic don't have a real business. They are ultimately acquisition targets for a Microsoft or similar because having 'the best' proprietary model to integrate into your products is a differentiation point and probably one the market will value. Meta to has things it can put its own AI products into.

Meta also has experience running and very public very massive IT operation...Something the labs with all their model escapes have shown they can't or won't do well. I actually have more confidence in Meta here than in the other players at the moment, and I expect Mark is right there is no need to stop, just a need to do it right, which Meta/facebook can and for others to be held to account for their failures.

which is why I suggested a human might have implemented a regex, test and quick exit with an error code.

Detect and Abend is usually a lot safe than guessing that 'grep' when you were expecting something like 198.51.100.45 means you should try to resolve grep.example.com and run with whatever address happens to comeback.

However in tool designed to be used by end users the try to figure-out what they intended and do it approach often is right. All depends on the stakes, but my example was one intended to be specifically where ease of use considerations don't really exist because the user is always another shell script. It was real lived experience try to get claude to work on a project effectively.

These are commercial refrigeration units. This is not the Kenmore in your mom kitchen.

They do have heating units because they do have to be completely defrosted and clean. Importantly this has to happen quickly not after just standing with the compressor off and the door open for two hours, time is money.

50C is going be less than 130F. So it is not like these things have a 'bake' mode, that seems pretty reasonable for an ice melt function. I only have experience with a 30 year old unit in our church basement but I did does have a fast defrost function, which is a just heater wire around the box, not sure if it would get 130, but probably 100 pretty easily. Not sure what would happen if you flipped it on and left it that way for bunch hours after all the ice was melted either.

I have seen claude, qwen, and codex churn out some very nice code. I also see a lot of it as being written for not reason.

The spec might say ARGV[1] will be a dotted decimal ipv4 address. The next thing you know you will get a bunch code to see if it might be a hostname instead and try to resolve, check if was passed as single integer, make sure the user did not enter it in hex..

It might all work too! It is none the less 10s of extra lines to handle every possible way an addresses when what was called for and what a human would have done is quick regex test, followed by calling printHelp() and existing nonzero.

No you know this is a program designed to be called by another program. That other program will NEVER specify the address any other-way. Maybe a human would call it directly as part of debugging/recovering some failed larger process but generally speaking the flexiblity is not intended and maybe not desired because it could mask other problems. So what do you do

A) strip out all the AIs nonsense
B) ask the AI to strip out the nonsense it added, remind it to re-read agents.md and to please stick the specifications without and not to try and anticipate additional functionality.
C) Leave it there because it looks like its all correct but for the fact that it exceeds the spec, and it works even if it is not now and will never be needed.
D) Something else???

I know a lot of people are choosing C, and for any given 'commit' it often kinda seems reasonable. It is just that slowly but certainly the project becomes larger than it needs to be, and when and if some design issue requiring major surgery is ever identified now you have a lot more code than needed to be there all of which might need to be touched, at the very least read, or read by the agents/models themselves sucking up valuable context....and burning expensive tokens/wall time..

It is a change in the market. Is it a lasting one, is the question. It had become a volume buiness but it seems to be turning into more of boutique one, where a a smaller number with specific requirements spend a lot of money on custom build with some very high end parts out of shrinking batches. A lot of us here have a lot of gray hair or no hair at this point. We are still mentally thinking about the later 80s-2010s or so where you could spend roughly the same inflation adjusted dollars every 24 months or so a buy a new computer that was a good deal more capable than the one you had. (Not saying you had to just that you could). Before that things were a little slower; sure the Apple ][ you bought in 1978 was not all that much different than the 1988 model but it did cost 30% less and have more memory which - did mean if you were using it for serious analysis work spread sheets and the like you could do more in meaningful business impact terms.

Since the start of this AI build out; this has not been true. Jumping You are not going to replace any AM4 kit with AM5 kit right now without spending a lot more on the AM5 kit than you did on the AM4 kit, if you expect to actually end up with a more capable computer. There just inst much value in going for 8 core AM4 chip to current generation AM5 if you are going to have to cut the RAM memory for 64G to 32G to hit the price point or even stick with 32G. Same thing with GPUs replacing your four or five year old GPU with one that has half as much memory today isn't a win in a lot of cases even if you are only using it for gaming.

Unless you spend big dollars on a high end system or you have some very specific professional or hobby/enthusiast workloads that calls for specific kit, there is basically no value in replacing a four-five year old middle or the road computer build with a contemporary middle of the road build. Again if something like changed like the memory went from 32G to 16G or something to hit the price point in the real world of 20 browser tabs open the new machine might perform worse.

Since these machines don't for the most part wear out the way something like an auto does; if I come back the Dell website after last visiting in 2022 with the same $1800 plus $300 or so more for inflation and I am being shown machines that don't look appreciably different spec sheet wise or worse, why wouldn't I just close the browser tab and move on in most cases?

You might, but Joe Public isn't. People are cutting Flock cameras down, they are not ripping RING doorbells of houses for the most part. Now I'll grant you the person taking a sawzall to their local Flock camera isn't maybe the sharpest knife in the drawer, however...

The people showing up at your county supervisors meeting are not demanding the county cancel their contract with Motorola or issue a statement saying they won't deploy Motorola ALPR. They are there given their rant about Flock; the sheriff's deputy in the corner who knows better dutifully maintains his neutral expression, even the ones receptive to the privacy arguments sit there considering how they like being on the board and how another officer involved shooting in front the BigBoxMart vs picking up the guy quietly at their home, isn't good for keeping their positions. Similarly is explaining why 22% of the safety budget was spent services they did not use and equipment they exposed of before it was depreciated is also unhelpful. Sure they could say we did it because privacy, but that leaves them to explain why they went for it in the first place. We all know if you are explaining you already lost the argument when it comes to elections.

So they say "Mr. Smith, the county thanks your for your comments and we can assure you we are not considering partnering with Flock." While keeping entirely quiet about all the other surveillance build out that is happening as fast the budget allows.

So no when 'we' as in 'we the people' complain about Flock, it is not a general complaint, most of the people don't read Slashdot, they don't know about the other stuff, and their complaint isn't registered as anti surveillance its registered as anti-Flock, because politically that is much easier!

Do you really want to enable everyone to stalk everyone?

I understand the motivation, an appreciate you maybe are thinking that might finally wake some of the political animals up to consequences of letting the Si Valley and MIC boys and girls do whatever they want all the time, but in the mean time that creep your daughter said 'no' to in the bar the other day can start popping up all kinds of places she just happens to be at...

I think this is place our society is really not ready to go, temporary or not.

The obsession many have developed with Flock and it is exactly than an obsession, because literally all thinking people do about it have little or nothing to do with Flock itself is really counter productive.

There are real privacy questions here. Should police have access to ALPR aggregators and as a subscription service really isn't one of them.

The technology to recognize a plate is getting better and better pretty soon there will be zero need for a purpose built camera, computer vision will identify a vehicle plate from a photo taken at any angle at any level of clarity a human could and then some. There are plenty of nice images complete with exif tags for date time precise location all over social media to scrape, there plenty of things like RING cameras, and even if its not Amazon specifically other companies will share their data, that do residential and commercial surveillance. It will not be long before gait + face + other recognition technologies are sufficient to relatively positively track individuals and vehicles, coupled with AI to correlate these uniqued tracks with a few really positive identifications say from ALPR, and anyone with enough cash is going to be able to construct a very detailed picture of everywhere a person or vehicle has been.

Policing and potential abuse is just one angle, and it isn't the angle I'd even worry about for every case of abuse there are at least some kidnapped children returned home, some property recovered, some criminals off the street, and remember those people arrested rightly or wrongly still get a trial - well unless they act like anti-society loonatices and try to combat law enforcement...they yes they might get shot.

Personally I think we should care about privacy. Police abuse though quite honestly seems like less of an issue for most people, I'd be more worried if my HR department suddenly got to start evaluating if I went the bars or the horse track to often for their taste. Maybe your homeowners company raises your rates because they see how frequently you visit the DIY warehouse store?

There really are only three (3) real solutions here and two (2) sorta practicle ones.
1) Eliminate all requirements to display unique identifiers in public. IE No more license plate numbers, just a state registration tag except on public vehicles (IE police, fire, medical, municipal). Non-government employers should be strictly prohibited from requiring employees ware any identification that shows and employee number or name when they are not on the employees own premises. You could still say you must where have on your "I am Steve's HVAC employee badge" when you visit a client, it just can't say you are "Jim Smith, #192".

2) Restrict photography, that captures individuals or vehicles on public property - ie streets. Carve outs would need to exist for places identified as parks etc where people would simply have the expectation of being photographed at least incidentally to others taking their family vacation photos etc. Yes this will change the media landscape we are used to in a huge way. We will have to go back to accepting witness description of evens in our daily news feeds.

3) Try to legally control the aggregation and sale of individualized marketing data. This is the one I actually don't think is reasonably. I think you have to tackle the production of the data on the supply side because it only becomes marketing data when it is used for marketing, before that it is someones security footage, university research projection, or news gathering operation. Trying to imagine and describe what is legally permissible to do with collected images from any number of sources after the fact with lots of actors motivated to try and get around it, is probably hopeless. The law just isn't that plastic.

Oh you want to do some more name calling. Ok you UnAmerican, open boards, commie retard prick!

FUCK YOU - you spend four decades supporting people and policy that created a wildly difficult situation to unwind, and then you complain when not everything is done perfectly. Except some of us see thru it. You know perfectly well you can't make an omelet without breaking a few eggs and no law enforcement effort in all oh history of any scale can be done entirely without incident.

You want amnesty and free for all, so you cling to impossible standards. Trafficked by ICE yeah ok dude so we have to allow 1000s of people to be trafficked by seriously bad actors because my goodness our enforcement operations might make mistakes that could impact a few hundred people unfairly. Who cares what country they came from, they have no right to be here. If someone is trespassing on your property and you ask them to leave you don't owe the bus fare home! It isnt your problem where they go, but they can't stay where they are. They should have considered that before being trespassers in our country!

Your policy choices have cause more children to be raped than Trump could ever harm if it is literally all he did all day! Save your bullshit you don't care about anyone but yourself and that is abundantly clear you shit eating viper. You just want your cheap toys to play with and fuck the nation!

If I were, perhaps it's because I'm not seeing any of the real stuff here.

The fraud of firing 6000 workers and immediately hiring 6000 H-1Bs was too in your face.

The Dept of Labor did the right thing by suspending Microsoft from the PERM program.