I see so many names in the commit logs, but some standouts include: Blum, Cook, Torvalds, Solodai, Tyragu, Stitt, Bergmann, Wysocki, Panda, de Mello, and no doubt some I missed who have a large number of commits fixing this problem.
Thank to all who undertook this Herculean chore!
It has seemed to me, for a very long time, that modern AI systems would need to be integrated with standard RDBMS systems for reliable persistant storage of raw information, some sort of no-sql database (memcache or some variant) for persistant storage of associations, some sort of document database for blocks of textual information, a SPARQL system for searching semantically-marked information within the document database, and a more old-fashioned back-propogation NN to provide a store of understanding that the user can directly manipulate.
Probabalistic classifiers are all fine and good, but only for a subset of the tasks needed. The above structure is a very loose, wildly-speculative initial framework. It's almost certain that if you actually tried building an integrated multi-model system, that you'd end up making a lot of changes to this basic idea, but that you'd end up having to implement the same core concepts that are identified in it.
The British government, in a desperate bid to increase profitability, has trademarked Alan Turing himself.
How is ChatGPT supposed to know the difference? Both involve hands and mouths, so clearly it's the same context.
From their careful selection of text, they WANTED it to mean something else so badly that they couldn't handle putting in the full text. It's a common blight on today's Internet, where people want other people's writings to mean something other than what was meant by the writer, so carefully select the words they read.
That's the entire point. Trying to solve other people's problems NEVER WORKS. You CANNOT control others into responsible behaviour, but you CAN place them in a position where they will choose to be responsible of their own accord. It is the ONLY way that works. It is the only way that has ever worked. If you look at computer programming, you will see this repeated over and over - well-meaning "hard rules" are ignored, STANDARDS are kept.
You must give them parameters and force them to find their own solution within those.
The original guy got to keep using it. There was someone else hired for a brief time.
I remember the author's name but he really doesn't want to use it, so that's OK to respect. He's given me a lot to think about over the years. I remember when he wrote on his PBS site about unicast becoming cheaper than radio broadcast for TV, predicting that it would overtake by 2012 (IIRC). Youtube became huge around then. We were smart folks around the water cooler in the late 90's who could follow the math but had nagging skepticism. He wss right.
I think I have one of his science writing books under his real name about atomic energy somewhere. You can find it if necessary.
Nice to see Bob back on the Dot.
I don't care how much of an idiot you are, you're simply too stupid to respond to further. I don't want things to be as they were in my childhood. Back then, things were a mess. BECAUSE government tried either to micromanage everything or manage nothing at all. The idea of a third way, where governing is about just that, placing control mechanisms in place but not do the management, is obviously far beyond your pea-brain.
Don't you see, AI is inventing ways for humans to consume less energy so it can eat it all.
> The lender can't repossess a college degree to make themselves whole.
No but if the borrower can't get a good job there should be cause of action for Warranty Act claims against the college.
Extremely few people go to college with the expectation of borrowing to be unemployable.
The problem is that you can ALWAYS get around rules. It isn't possible to make perfect rules for anything above a minimal level of complexity - that's just a variant of the Turing-Church Halting Problem.
So you are forced to invert the dynamics. There's no real alternative. Instead of you creating a high level of complexity that the departments will work their arses off to avoid, you force the departments themselves to create the regimens that they're prepared to live with. But you have to do so cleverly. They will always create regimens that mean they do the least work necessary (because that's cheap on resources and they will ALWAYS consider this sort of extra work to be an imposition) and have the least amount of culpability.
So you need to meet three conditions:
1. The department can't evade the bits they're actually able to do
2. The department CAN pass on work they're not equipt to do, but ONLY if it's their responsibility to oversee the department they pass it onto
3. The department IS inescapably culpable for failure to either do the work OR ensure that others do it
You do NOT need the frameworks for each department, and should not attempt to draw those up. Those will be departmentally-specific and timeframe-specific. Far, far better have people who actually know the specific context do that work. No department likes to look like it's being forced to do anything, so making the actual detailed specifics internal, you're utilising their psychology. They're not being "forced", they're defining their additional responsibilities and duties. From a psychological angle, they're much more likely to be receptive to this perspective.
But because the departments are all internally writing their own management protocols, YOU DON'T HAVE TO. You only need to have a framework which obliges them to write up what they will request. This is MUCH lighter and, because it is much lighter, it is far less prone to have failure points where generic ideas don't work for a specific type of work.
If we want to look at this in software terms, only an idiot would write an overly-restrictive langauge that imposes a strict model of thought regardless of the type of work. If you want to provide a high level of confidence in correctness, you don't try to impose it through a myriad of complex hurdles and rigorously controlled APIs. You achieve it by incorporating contracts (function X is guaranteed to take in data meeting these requirements, and is guaranteed to deliver data meeting these other requirements). Contract programming is much, much lighter on the development process, doesn't impose on the programmer, and yet creates a very high level of assurance. Mostly because programmers aren't working to try and cheat with irritating APIs.
In Linux terms, you want a lightweight virtual layer handling filesystems in general, the filesystem policies should be handled by the filesystem not the main kernel. You want the main kernel to be doing as little of the work as possible. As soon as it is heavy and micromanaging everything, you're going to end up with something slow and unstable, that really can't do a whole lot.
You want to push the complexity to the edges, that's where complexity belongs. The bit that changes slowly, can't handle special cases, has least visibility into what is needed, and is really a very blunt instrument wants to be lightweight. One reason for having things like Common Law and Case Law is precisely because the legal system figured all this out centuries ago.
Why do you believe those "AI detectors" are accurate? The past evaluations I've seen came to a different conclusion.
Almost like he's a campaign donor! Did he fly on AF1 to China with the rest of them?
Government very rarely runs on merit. It's a terrible way to conduct anything that doesn't require killing people.
Sure, but border guards and spooks probably already had this exploit so the difference is minor. Their PoC page also says there's no access to Secure Enclave so perhaps the damage is minimal?
Curiously I saw some commits for an iPhone platform in LineageOS a month or two ago. Perhaps an option for EoL Apple hardware with working exploits.
You know you've landed gear-up when it takes full power to taxi.
