I have been stress-testing AIs with increasingly complex projects for some time. The Chinese AIs struggle, but actually do a FAR better job of handling massively complex tasks than Grok, and Gemini just rolls over and whimpers at anything above a very low level of complexity.

What I've found is that the Chinese AIs tend to be sycophant but do "understand" complex projects properly in that you can ask specific technical questions and the answers will be generally very accurate. Any sort of critical analysis is beyond them, though. (Ether that, or I'm a mega-genius. Which....doesn't sound terribly likely.)

Of the "Top AIs", ChatGPT is good on basics but is incapable of any kind of detailed generation. Claude is brilliant at detailed generation, but overloads with anything but a tiny data set.

I've been putting up the projects on Gitlab for a while, so anyone who wants to see an AI break down and cry in despair is able to do so.

The secret tools don't bother me - they'll have long understood how to use Big Data and Analysis of Competing Hypotheses. AI isn't going to find out any more than combinations of those tools will, because that's basically all AI is - a Big Data classification system.

So you're telling Claude something vague and washy, then Claude invents a prompt that might vaguely possibly be somehow related to what you want along with a drink that is almost but not entirely quite unlike tea. Claude then recurses through this until it has a Celtic knot so intricate that it has its own Hausdorff dimension. What burps out is a product that is completely useless and patented to the Sirius Cybernetics Corporation.

They won't be OceanGate-grade. They used PRCF that had already been rejected by quality control. That's a non-starter in aircraft.

As production has ended, if the A380 is genuinely necessary, then the economics shift somewhat. That doesn't mean they CAN be replaced, from the sounds of it they can't* (at least in many cases), but the inability to replace the aircraft would mean that options that aren't rational become necessary.

*I have to be careful here. If the wing is designed to be the absolute minimum weight possible, then I don't see how they could be without fully disassembling the entire wing and then reconstructing it from the ground up. And adhesives/welding might mean that just can't be done. At all. On the other hand, there's no obvious reason why you couldn't design a wing to have far more structural support than actually needed AND make spars deliberately maintainable and replaceable. I don't have an A380 handbook in front of me, so can't say how Airbus approached this. But it seems improbable that they're built to be swapped.

The Trash 80s? Had a Commodore PET 3032. A whole 1 megahertz. On the other hand, the IEEE 488 meant that I could send a command to one disk drive to transfer to a second disk drive, whilst printing, with the computer then totally free to actually do other stuff. SCSI it wasn't, but for the time, it was an ingenious solution to a lot of problems.

Sheer luxury, lad! Sheer luxury!

We had to make our own bits and push them uphill! Both ways! In the snow!

Pah, youngsters.

However, some of you might be old enough to remember the April 1st when the front page turned pink and a link to Cute Overload caused absolute chaos. Did CmdrTaco ever get a pony, btw?

And the number of false convictions (roughly 25% of those convicted in the US) doesn't cause a problem for you?

Then you're part of the problem.

US: 66% (Wall Street's numbers aren't those found in official statistics)
UK: 28.9%
Holland: 23%
Norway: 16%
China: 6%

US' conclusion: The rate is a complete mystery, we've no idea how to decrease it, let's do more of what we're currently doing differently to everyone else.

There is a slight possibility this may be flawed.

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. :P

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.