Now image that smart phones are "always online" and could be hacked such they simply wouldn't start anymore. But who would be silly enough to propose that phones should be online?
I want to create, systems about which I can say "this is known to be absolutely correct; it has been mathematically proven correct".
At best, you can hope that you can mathematically prove that the system conforms to the specification. Whether the specification is actually what you (or the customer) wanted is still unsure.
If we can shift that 10 Million from NASA to NOAA, and NOAA orders the satellite from NASA and uses the leftover cash for more climate studies instead of hiring climate experts (which NASA would have to do. NOAA already has experts)
NASA already has experts too. You wouldn't actually save any money. You still need the same work to be done. By moving the project from one place to another, you would even incur extra costs and inefficiencies during the transition.
But all of that is completely irrelevant. The project and budget isn't shifted. It's shafted.
That ends up with rainbow alliances where getting a majority large enough to form a government ends up with complex compromise agreements trading policies in order to get into power
Marcus said Facebook didn’t have plans to integrate cryptocurrency into its apps anytime soon.
What ? That seems to be only logical use case for them, and they're not doing it ? Makes no sense.
"Payments using crypto right now is just very expensive, super slow, so the various communities running the different blockchains and the different assets need to fix all the issues, and then when we get there someday, maybe we'll do something,” Marcus said.
That's stupid too. Facebook can slash transaction costs by keeping them off-chain, just moving virtual balances from one account to the other, and then only occasionally settle on-chain.
That's what they said about some schmuck writing down mercury thermometer measurements every day in his old notebook a century ago.
Now these measurements allow us to create a nice graph of the temperature changes. Collecting data is useful, even if you don't see an immediate need for it.
NOAA was not monitoring the same thing that NASA was doing. You would have a point if the NASA CMS program was moved to NOAA, keeping the same funding, but it's been completely cancelled with nothing to replace it.
But in both cases, any new application requires a whole new engine with lots and lots of work by humans looking at its mistakes and writing little modifiers to the algorithm to make it better than what an average human can do with training
You are apparently not familiar with the newest developments.
If you are a chess player, try out Leela Chess Zero: http://play.lczero.org/ (apparently, there's also an option to play it on lichess)
You can select easy/normal/hard mode. In normal mode, it will look at 50 different positions before making a move. That's not "brute force". All of the chess knowledge was discovered by LC0 by playing itself. There's no human input, except for putting in the rules of the game, and creating the self-learning frame work.
The Global Monitoring Division of NOAA/Earth System Research Laboratory has measured carbon dioxide and other greenhouse gases for several decades at a globally distributed network of air sampling sites
These are measurements from a couple of dozen fixed ground stations. While the form a nice background check, they do not provide much fine detail over where the carbon is coming from, and how it moves through the atmosphere.
The software knows from analyzing millions of games that humans have played what winning strategies are, and combines that with brute force strength to know where to optimize its searches.
Not quite. Alpha Zero learned the winning strategies by starting from scratch and playing only against itself.
But a neural net is ultimately just a (complicated) nonlinear function that produces a deterministic output depending on it's input. It's completely algorithmic
Nobody in AI research is suggesting artificial neural networks are going to achieve consciousness.
There are plenty of people in the AI research community, and I doubt you speak for all of them.
Artificial neural nets are just algorithmic ways to generate a nonlinear function for classifying things.
There's no reason why these functions could only be used for classification. We have neural nets that can generate images, provide translations from one language to another, convert written text to realistic speech, learn to play computer games and many other things.
Slashdot Mirror
Humans and ML work fundamentally different when they "recognize" an image, so one cannot tell the other how it was done
Depends on the image. If you spot a family member in a crowd, you can't explain how you did it either.
Now image that smart phones are "always online" and could be hacked such they simply wouldn't start anymore. But who would be silly enough to propose that phones should be online?
A versioning file system is not the same as a backup mechanism.
I want to create, systems about which I can say "this is known to be absolutely correct; it has been mathematically proven correct".
At best, you can hope that you can mathematically prove that the system conforms to the specification. Whether the specification is actually what you (or the customer) wanted is still unsure.
A lot of people live in apartment buildings where they can't plug in the car at night. Plugging it in at the office would be a good alternative.
A human would give the correct answer
You hang around with nerds too much if you expect the average human to know the answer to life, the universe and everything.
this essentially groups junk food with cigarettes
With heart disease and stroke both outnumbering lung cancer as national diseases, that's not a bad idea.
Junk food is only dangerous by over indulgence
Junk food is carefully designed to promote over indulgence.
But I'm torn...is it really the government's business to play nanny, and try to prevent stupid people from behaving stupidly?
If you expect the government to pay for the consequences, they should also have the right to prevent.
Okay, but my point is that the data could be in a true vacuum when it is first collected.
If we can shift that 10 Million from NASA to NOAA, and NOAA orders the satellite from NASA and uses the leftover cash for more climate studies instead of hiring climate experts (which NASA would have to do. NOAA already has experts)
NASA already has experts too. You wouldn't actually save any money. You still need the same work to be done. By moving the project from one place to another, you would even incur extra costs and inefficiencies during the transition.
But all of that is completely irrelevant. The project and budget isn't shifted. It's shafted.
or in unique markets with no competitor (SpaceX)
You're funny.
That ends up with rainbow alliances where getting a majority large enough to form a government ends up with complex compromise agreements trading policies in order to get into power
Making compromises is not actually a bad thing.
Marcus said Facebook didn’t have plans to integrate cryptocurrency into its apps anytime soon.
What ? That seems to be only logical use case for them, and they're not doing it ? Makes no sense.
"Payments using crypto right now is just very expensive, super slow, so the various communities running the different blockchains and the different assets need to fix all the issues, and then when we get there someday, maybe we'll do something,” Marcus said.
That's stupid too. Facebook can slash transaction costs by keeping them off-chain, just moving virtual balances from one account to the other, and then only occasionally settle on-chain.
That's what they said about some schmuck writing down mercury thermometer measurements every day in his old notebook a century ago.
Now these measurements allow us to create a nice graph of the temperature changes. Collecting data is useful, even if you don't see an immediate need for it.
NOAA was not monitoring the same thing that NASA was doing. You would have a point if the NASA CMS program was moved to NOAA, keeping the same funding, but it's been completely cancelled with nothing to replace it.
But in both cases, any new application requires a whole new engine with lots and lots of work by humans looking at its mistakes and writing little modifiers to the algorithm to make it better than what an average human can do with training
You are apparently not familiar with the newest developments.
If you are a chess player, try out Leela Chess Zero: http://play.lczero.org/ (apparently, there's also an option to play it on lichess)
You can select easy/normal/hard mode. In normal mode, it will look at 50 different positions before making a move. That's not "brute force". All of the chess knowledge was discovered by LC0 by playing itself. There's no human input, except for putting in the rules of the game, and creating the self-learning frame work.
The Global Monitoring Division of NOAA/Earth System Research Laboratory has measured carbon dioxide and other greenhouse gases for several decades at a globally distributed network of air sampling sites
These are measurements from a couple of dozen fixed ground stations. While the form a nice background check, they do not provide much fine detail over where the carbon is coming from, and how it moves through the atmosphere.
The software knows from analyzing millions of games that humans have played what winning strategies are, and combines that with brute force strength to know where to optimize its searches.
Not quite. Alpha Zero learned the winning strategies by starting from scratch and playing only against itself.
But a neural net is ultimately just a (complicated) nonlinear function that produces a deterministic output depending on it's input. It's completely algorithmic
Sure, but so is most of our brain.
Nobody in AI research is suggesting artificial neural networks are going to achieve consciousness.
There are plenty of people in the AI research community, and I doubt you speak for all of them.
Artificial neural nets are just algorithmic ways to generate a nonlinear function for classifying things.
There's no reason why these functions could only be used for classification. We have neural nets that can generate images, provide translations from one language to another, convert written text to realistic speech, learn to play computer games and many other things.
I'm pretty sure that Google isn't going to let you make mass prank calls.
They're not faking the voice I think, instead it's a set of pre-recorded voices
Neither, it's a neural net trained on real voices.
it's either a live human being making the appointment or reservation or you hang up on it. Could be a prank
If I wanted to make a prank call to set up an appointment, I could just call your business myself.
Even better when you write the letter in all caps.
Like this https://www.youtube.com/watch?...