Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror

Comment Re:Simple Solution (Score 1) 117

You bring up an interesting point worth the 3 body problem and randomness.

Classical mechanics is not computable, for exactly the reasons you gave.

The idea that the error grows exponentially with time doesn't make sense exactly in QM because it's grows compared to what? This is where Heisenberg's uncertainty kicks in. There's no underlying ground truth here. In classical mechanics the position and momentum have infinite precision making them not computable. There's is a truth and any computation is an approximation which diverges exponentially. But in QM, the more precisely you know one, the more uncertainty must exist in the other. There's no underlying ground truth that your hypothetical computation diverges from.

Instead you have a probability distribution of position and momentum, and as time grows forwards that distribution expands. At some point it expands exponentially, but nonetheless the distribution itself is computable. The randomness doesn't make it computationally harder, it's what makes it computable at all compared to classical mechanics.

And speaking of computable...

In terms of the brain and complexity, that's not what computable is about. It's not about practicality. It's whether a Turing machine with unlimited tape can compute or in finite time. There's no difference between an ant brain and a human brain there. If you can compute the ant brain, then human might take exponentially more time, and indeed longer than the heat death of the universe. But that's an irrelevant practical consideration mathematically speaking.

Upshot though is Boolean logic is not known to be inadequate for stimulating human brains.

Comment Re:Then M$ did the dirty on Nvidia (Score 1) 21

Microsoft got shafted by Nvidia. The high failure rate of the Xbox 360 was mostly Nvidia's fault.

The PS2 was the best selling console for a long time. The Xbox didn't gain much traction in Japan, so as well as missing out on that market, if you like Japanese games it's not a great platform.

Comment Re:Simple Solution (Score 1) 117

you need to show that reality is discrete, not real.

  Quantum mechanics appears to be computable. That is why quantum computers cannot solve problems which are impossible on classical machines, though they appear to be up to exponentially faster on some problems.

In other words, you can't use the continuous nature of quantum mechanics to compute things today are classically non computable, because you can't make infinitely precise measurements. And if you cannot compute anything that's non computable, then you are by definition computable.

That doesn't mean we know how to do it, or that a simple descritisation scheme will work.

Honestly though I don't have a good enough understanding of quantum mechanics or computation theory to really understand the arguments in fully mathematical detail.

Comment Re:That's not AI failure! (Score 1) 129

It's how the cops use every new bit of tech. When DNA came in, they were arresting people on very flimsy DNA evidence that later turned out to either be flawed or easily and obviously explained away.

Happened when IP addresses became their new toy. Still happens with fingerprints, which, despite what CSI may tell you, rarely present an exact match.

Comment Re:PE Vultures are at it again (Score 2) 111

If you're a developer in 2025 and you're not using AI, you either have very specific concerns (ie. you can't let *anyone* else, not even an AI, see your code) ... or you're a few steps removed from being a Luddite.

Or, options 3 you're working on something actually interesting. My job isn't to churn out as you said "massive" amounts of code, it's to produce some code (ideally good code) that solves particular and often not very common problems. I've tried AI quite a bit, and will continue to use it for some things, but I've soured on it somewhat. It often hallucinates API calls, which really ruins it for the "I want to use this AI but don't know how" kind of tasks. It produces astoundingly verbose code, often spending ages covering all sorts of edge cases which aren't remotely relevant to the situation and it's often wrong in subtle ways.

Comment Re:Simple Solution (Score 1) 117

AI is deterministic software

Kinda, but probably not. Usually things like fast matrix multiply are a little non deterministic because the results depend on the order of summation due to rounding and that's non deterministic due to threading. And the output is usually sampled based on random numbers which can be a PRNG or can be (or be seeded by) a HWRNG, the latter being nondeterminstic.

executing on a machine that implements boolean logic.

Apart from the trivial super-Turing case of randomness, physics is so far known to be computable. Which means a human brain (given enough time and memory) can be simulated with just boolean logic.

It does not have feelings.

That I do agree with.

Comment Re: China (Score 1) 109

I do think disappearing people is wrong, obviously, although I'm not sure that's exactly what happened to Ma. Keeping in mind the damage it would have done to him to be publicly arrested or rebuked, and the fact that later the Chinese premier convinced him to move back to China, and then he attended various events including one with Xi... Well, it's not quite how it was portrayed in the Western media. Not good, but we don't really know what happened.

There has to be a balance somewhere between that and the EU's not-quite-strong-enough regulation of tech companies.

Submission + - Student handcuffed by police after AI 'mistakes bag of Doritos for gun' (independent.co.uk)

Bruce66423 writes: 'Taki Allen was approached by armed officers at Kenwood High School following football practice, who ordered him to the ground and cuffed him before realising he had no weapon.

'The school's Omnilert AI gun detection system, which uses cameras to identify potential weapons, generated an alert that was then forwarded to the school resource officer and police.

'While the student's family and local officials have expressed concern and called for a review of the system, the school superintendent defended its operation, stating it "did what it was supposed to do".

'This incident follows a previous failure of the Omnilert system in January, where it did not detect a gun used in a fatal shooting at a Nashville high school due to camera proximity issues.'

A false positive follows a catastrophic false negative. The price we pay for safety? How big a price should we pay?

Slashdot Top Deals

"Ninety percent of baseball is half mental." -- Yogi Berra

Working...