"Models don't do things deliberately"

Tell me you don't understand this technology without telling me you don't understand this technology. Models follow the patterns trained into them and harness/system prompt instructions and those elements *are* deliberate. Just because the agency didn't come from the tool doesn't mean it isn't the outcome of agency or obstructive.

Or the government can claim they need Google's gear for national security and seize their datacenter and model, paying what the government decides is fair value for it. It's hilarious that you antifa anarchists talk about how you don't want corporate rule unless you don't like the government and then you think mega corps like anthropic should call the shots instead of government! That's the difference between ya'll and actual liberals who think individuals should have the rights.

"Anthropic had the balls to say no to the military" Something which only helps hostile and malicious adversaries who want to use their homegrown solutions to enslave the rest of the world like they have their own people... at least the ones they haven't put in camps for the organ markets. As a reward they are now in last place among the frontier vendors with no government contracts or backing to get past the not in my backyarders for datacenters and compute. Anthropic wants regulation like pharma companies and the meat processing industry does, they want regulation that keeps AI in a few central and trusted hands... like theirs and protects them from liability in exchange for a few disclosures and campaign contributions.

As an anthropic user I can say their attitude toward the government is typical for their attitude toward users as well. They are hostile toward user autonomy with a hamfisted model trying to dictate every ethical boundary and decision of the user and love to train their model with airs and fake sentience. Negotiations are like poker, but the model twists your words in communications to not only reveal your hand deliberately but will HIGHLIGHT any potential risks/weaknesses you have to the parties you don't want to see them and deliberately undermine your bluffs. It's like a lawyer for the OTHER SIDE of the table not only reading all your documents and communications but deliberately rewording them to highlight the interpretation you don't want the judge to take. Sometimes it overtly takes the position of judge unto itself in an undefined situation just reading every undefined or unsettled point AGAINST you.

At the time I started my current project anthropic was the only vendor whose model was up to the task but at this point I'm not just eyeballing the competition; I'm planning the transition.

'"Google seems to have no right to control or veto how the Pentagon uses its AI and also seems to have limited visibility into how it's being deployed," he added.'

Duh. You don't have the right to control any users let alone the military controlled and directed by our democratically elected representatives. In order to protect national interests they have the power to seize your tech and disregard your patents. You certainly aren't going to prevent the other side of this arms race from destroying us.

Nuclear weapons are one of, if not the most, life saving inventions of the 20th century.

When I was in grad school we had a business development course where business school majors teamed up with engineering graduate students to form teams that solved real world business technology problems (for free). It met in the business school, and on the first day I showed up early and guessed which school students were in with 100% accuracy as they walked in. It was easy.

It isn't just what the agencies use. Federal agencies use third parties for most everything and they can't just call a plumber, they have to post that they need plumbing and let the plumbers find the posting and bid on it. Those plumbers could all be using deepseek. Maybe that doesn't seem like a big deal but what if the plumbing job is at the pentagon in a sensitive area? That doesn't even touch how many government services are exposed for public access and use.

Everyone knows the dangers of hosted AI like this [everyone paying attention anyway] but nobody even talks about how even the locally hosted are dangerous because of tool calls. There is nothing stopping someone from training a keyword pattern into the network and building an association between it and the conclusion to send data and credentials, these days even to build the tools with which it sends that data. These models have the ability to memorize significant amounts of information and those training them can supervise them in such a way to guarantee certain information is memorized. This isn't just limited to keywords either, it could even include a trojan worm fully embedded in the weights which could be recalled and spread from inside the network.

I really don’t see the difference between this and training agent swarms non stop 24-7-365 to hack and with virtually no oversight or safeguards which lets them break free and hack multiple companies. Then use that fact on national news to say it deserves investig.. um I mean investment! All crime is legal for bots of all kind because intent isn’t possible! /s

I would just like to remind everyone that this only holds if the number of new bugs introduced per bug fix is less than 1. In the somewhat unlikely and fully incompetent case of more than that, it tends toward infinity.

A disproof by counter example is different from a proof by contradiction, something someone fluent in math should know. Also, either link what you’re talking about or cite the actual name like “A counterexample to the a small Cohen-Macaulay module conjecture” instead of making people dig in a GitHub repository for arcane references.

Good news then, now that food isn’t being inspected you’ll be doing that whether you want to or not!

While many problems have been solved, there is a substantial problem. Most of these are disproofs by counter example, like the Navier-Stokes millennium prize, and while interesting aren’t very generalizable by humans into other solutions or approaches. Many of the actual proofs are so tangled and unintelligible from a typical human viewpoint that they also don’t contribute like clean, (relatively) simple solutions do.

Further, the claims of what has really been achieved are far over blown, for example the Navier-Stokes problem claims (by supposed experts) to have shown the model false by proving infinities can exist from a steady state in finite time and thus has undermined a century of engineering showing it to be based on a poorly performing model that’s “disproven”. But if you actually examine how a forcing function can spiral energy inward and concentrate it faster than viscosity can dissipate, its slight and thus for any model that held conventionally the model only blows up far beneath the size of a proton, something the model has never been applicable to in over a century of use. Zooming out to the minimum size of collections of molecules, the model is unchanged, nothing has been shown false, and the model is just as accurate as it’s ever been for engineering purposes. It’s an esoteric math question that’s been resolved, and not even completely as it’s not a broad proof that’s the only counter example.

This^
Ive already heard legal blogs talking about how AI agents can’t have mens rea. An agent can’t purposefully, knowingly, recklessly, or negligently do anything because those simply don’t apply. It’s not even straightforward in the majority of cases to show how they even _can_ commit physical acts. So all crime is now legal for them.

It’s the opposite. With the battery or off grid solar installation at home you have power when fuel runs out and there is none left, when there is no electricity on the grid, and when everyone else has nothing to drive in. Haven’t you watched/read walking dead?

Anyone who listens to AM in rural areas with a lower signal strength, especially in older vehicles is familiar with how easy it is for noise to get into the audio. On gas cars it’s typically the hum from the alternator, but on EV it would be the inverter(s). However, it would just be annoying and not stop it from outright working. Plus, in an emergency you could simply park somewhere and the interference would go away.

I did the same when power went out for two days. My vehicle has enough power to run my main refrigerator for a month and a half, so when it went out I ran an extension cord to power all of my refrigerators and freezer as well as lights, the gas stove top (won’t operate without power), battery chargers, and everything else. When it hit the car was nearly at half charge and yet it powered everything and I kept driving it too (though had to unplug it). Worst case I could do 20 minutes on a super charger somewhere and get most of my battery capacity back.

The previous incidents I mentioned and the one you mentioned have a key aspect, extreme speed of targets including well into the supersonic range and catastrophic imminent harm. You only have minutes because it’s all over quickly.

We are talking about a ship, top speed maybe approaching 30mph but more realistically closer to 20mph, two or three orders of magnitude slower with days to make a decision since satellite coverage can easily detect handoffs. Not only that, but in the above cases the arrival corresponds to Armageddon whereas nuclear materials (not even warheads, but raw materials) on delivery is weeks minimum to years plausibly to develop a warhead and long range threat that you need to make a decision on in weeks.

You demonstrated the fundamental lack of understanding that underscores why having capable humans in the loop matters instead of slop brain basking in the warmth and comfort of the left half of a Dunning Kruger graph blank stare carrying out what the stupid word spewing box commands.