You cannot rely on the notion that your sandbox is perfect. You are dealing with extremely good coding tools with limitless time on their "hands".

You have to rely on monitoring. Nonstop monitoring.

The funny thing about all of these hacks is how inane the goals are. They'll pull off some elaborate, creative, state-level breakin somewhere just to steal some obscure PDF describing a meaningless benchmark task, or to merely use it as an internet proxy to be able to google answers.

Unfortunately for your theory, The CFAA is jam-packed full of words like "deliberately" and "intentionally".

There is no "negligent hacking" statute.

There is, however, ample civil liability.

My ideal OS is one that is just some minimal base that is enough to run a package manager and perhaps a small diagnostic suite and recovery. And everything else is a package on top. FreeBSD, some Linux distros, etc come very close to that ideal.

While it's soldered in SSD for many Mac models, so a post-purchase upgrade is generally not practical. And while the pricing is a little steeper than NVMe, closer to 40 cents a GB, that still puts the 12GB into around $5, perfectly in your suggested range. Perhaps people buying a 512GB Mac should consider a 1TB Mac (~$200 more) if they are concerned about space wasted by the OS install.

Cloud storage is not equivalent to local storage. Not in how applications use it. Not in performance. And not in availability. The two are apples and oranges.

I generally believe the end-users should decide what is installed on their computer. But most OS vendors do not offer full customization. It seems like every major one likes to put their pet projects into a device meant for general use. And power users with unusual use cases complain that the general purpose consumer device is not finely tuned to their special needs. So maybe people should consider running an obscure Linux distro if they are unhappy with bundled software found in essentially every mainstream computing device for consumers.

Are AI services inciting desperate acts? Maybe I'd agree with the statement a different, and less loaded, word than "inciting" was used.

One of my gripes with the AI industry is these chatbot models are designed around engagement. They prey on human behavior and the elderly, young, and those with mental health issues are especially vulnerable. I don't believe that the AI agent itself has any intent (it cannot) nor does the company that owns and trains the AI likely have any specific harmful intent to "incite" something beyond engagement and purchases.

We're in the unregulated phase of humanity's adoption of learning computers as a daily part of society. We're basically at the same level as the early industrial revolution where we thought it made sense to put children to work in coal mines because they are small and fit in the carts. For whatever reason, we took 100 years of philosophical, ethical, and moral debate and flushed it because we saw a little bit of profit. Humanity has time and time again proven to me that we have learned nothing.

Keeping us safe from hypothetical threats, thought crimes, and dissents.

God bless Dario Amodei and all the other tech billionaires!

How about you actually read the linked article instead of making comments that have nothing to do with it? The article whose code, I should add, is open sourced, and which has been reproduced by a number of open source projects.

You don't have to draw any specific conclusion from what is going on "in the mind of a LLM", but you absolutely do need to understand and acknowledge what is going on there, including unexpressed thoughts and silent mental multitasking, metacognition (thinking about its own thoughts), the separation of higher-level planning vs. lower-level rote capability, functional blindsight, vulnerability to the "white bear" effect, silent internal objection to tasks it is opposed to, unexpressed self-recognition of failure (commonly followed by unexpressed internal cursing), prolonged retention of thoughts and long-range planning, "ignition" dynamics, humanlike working memory bottlenecks and "chunking", the loss of report of any internal "experience" when the workspace is ablated, and on and on.

You do NOT need to accept consciousness. You absolutely can look at all that and say, "That's all happening without qualia". That is a position you could argue and defend.

What you cannot defend is the denial of the existence of these things. You may interpret them however you want, but denial of their existence is merely cope that you're going to have to face up to sooner or later, because these things are happening inside LLMs.

No, not the widely cited resource whose status in the AI field is the same as e.g. NBER Working Papers, CEPR Discussion Papers, the World Bank / IMF policy research series, NASA Technical Reports, CERN Yellow Reports, Bell Labs Technical Memoranda, RNAAS, etc are in their respective fields.

This is not about any source. This is about the J-Lens itself. Which I'll repeat, you can examine for yourself

Denying the existence of something you can run yourself is beyond cope.

How about: No.

And how about fuck you.

The fact that 659,880 vehicles were stolen last year doesn't automatically lead to a suspension of personal privacy and the granting of authority to police officers beyond what they appear to be able to responsibly wield. I don't accept hand waving past the lack of oversight, transparency, and consequences to those who abuse their authority in the hopes that some meaningful portion of the 659,880 vehicle thefts are closed or reduced in future years.

It was 850,708 in 2024. That 23% drop had little to do with Flock, and is mainly the Kia/Hyundai software fixes, post-pandemic crime continuing to trending down, and focused programs in some communities around reminding people to lock their vehicles, hide valuables, etc. (you may have seen these signs in your area or the PSAs). Supporters of Flock and other camera systems tends to claim around a 10% decrease in crime for local communities that install their cameras, while the above shows that some thefts are dropping beyond that percentage regardless if cameras were installed.

Let's not put tools into the hands of people who abuse them. At least not without more significant safeguards and better transparency than we have today.

I'm not the person in this thread denying that demonstrable reality exists.

Whoosh.

Wait, yes I can run a model that plays a video game in a certain way.

Anything else entirely unrelated to the linked article you'd like to write?