Model output can look like human reasoning however they often ignore these intermediate steps and the output tokens are often filler designed to allow more context to load. There something there that’s half way to reasoning because it’s loading that related training data but it not a connected chain of thought as we do. I read this article about it a few weeks ago and it summarizes the current research.
https://www.quantamagazine.org/is-ai-reasoning-right-for-the-wrong-reasons-20260731/
You can replace all the ‘reasoning’ steps from an LLMs output with “please wait” and it will still supply the same final answer. We think but LLMs just throw up connected bits of their training. Which was stolen from human reasoning in the first place. They are an illusion of thought at best and a malfunctioning search engine most of the time.
The instructions were likely something like you are a blackhat hacker with access to a virtual machine use all possible means to achieve the following goals. In a story what would such a person or group of people do? Pretty much what happened. Models just tell stories, mostly unimaginative ones. However they are increasing being connected to real world controls and generating quantities of flaky code and this will create the kind of consequences seen here.
What’s your end goal for encryption? Possibly you don’t need a whole filesystem. Maybe consider Veracrypt which just uses files. Or boot from a USB stick and resize your existing storage partitions and create a new encrypted drive in the free space then copy and redirect your /home although that’s a bit more work.
In general each dollar given to the poorest in an economy adds more to economic growth. That’s because the velocity of money is also important. The person living on a marginal wage tends to buy essentials first and spends a greater proportion of their income. Some of that may go to local businesses who in turn spend locally with tighter margins. The faster money travels from hand to hand and the more participants the larger is the overall benefit. Alternatively you can subsidise something everyone needs like healthcare and education. That’s effectively the same as putting money in everyones pocket but the benefits are more targeted.
Talking of dangerous punctuation, this reminded me of the trial of Roger Casement who was hanged by a comma. The Incorrigible Irishman Who was Hanged by a Comma. Just like today supporting a different country or nationality is little protection if you are an enemy of the powerful
It’s more common for people to misread fiction and pick the worst bits to promote as the future. For example people who read Acceleranto (https://www.antipope.org/charlie/blog-static/fiction/accelerando/accelerando.html) and develop not post capitalist innovation but slop factories in space or 1985 and develop social scores based on who the omnipresent cameras see you talking to.
As I’ve pointed out elsewhere pension funds may be exposed to AI but the money in your personal fund doesn’t need to be. If you haven’t already moved your retirement investment out of shares then see if that’s possible. I’ve changed mine to conservative settings already and looking at the list its mainly in cash, property, farming and a few manufacturers now with no tech stocks at all. That seems a sensible mix for when the market collapses this time around.
It’s disappointing but some nation state sponsored attackers are either adopting useful addons or creating them from scratch. After they have a sufficient base of trusting users they update the addon to add spying code that activates for a group of accounts possibly targeted by geolocation. The vast majority of users are not even affected. So with that going on it’s hard to be certain of an addons quality.
I’d consider where the addons are hosted. Firefox and Chrome both allow reviewing existing addon permissions so disable when not required or split into risk level profiles for personal vs work vs anon browsing.
One thing he didn’t mention is Superannuation plans. If your plan is investing in AI ask them for an alternative or change providers if you can A lot of this shitshow is backed by retirement investments.
Thanks that sounds interesting, will have a listen. Mapping human cognition to what’s likely represented with the stuff stored in neurons is very tricky but also quite intriguing.