There are lots of different predictions, and you can always find wrong ones. If I looked hard enough I could probably find one that projected glaciers in London. So you can't prove anything by retrospectively picking a prediction that was wrong. You need to either prospectively predict it was wrong and publish refutable reasons, of do a massive retrospective survey of all published models, which is a task nobody has been up to. The IPCC(?) did a prospective report on a bunch of models that they found acceptable and predicted certain signs of warming that could be expected. The models disagreed about exactly what those signs would be, which is not surprising as they were written by specialists in different areas. They rejected those models that they found too pessimistic (as well as many others), and published their report of what to expect. They were overly optimistic, so their predictions have been refuted, but because of their methodology this doesn't say which models were wrong how. My guess is that part of the mistake was that they rejected models that were too pessimistic, part of the mistake was that different models had interactions between the places where they were correct, and there were also feedback loops that nobody included. So they were all wrong, but not in the direction you appear to be proposing.
Picking one model, or cherry-picking features from a bunch of undocumented models, doesn't prove a goddamn thing.
We should anticipate less so that to compensate for the anticipation effects.
We did. The official reports refused to include projections that were too pessimistic to make the results more politically palatable. To be fair, they also refused to include projections that projected cooling, and there were a couple. Then they took the average of the reports they decided to accept. They had to know they were low-balling the numbers. Unfortunately, it's starting to look like the correct models were some of the ones they rejected for being too pessimistic.
I eat a little broccoli because I like it, but I don't eat much because it's too high in non-fiber carbs.... When I said a really low carb diet I meant it. What I really miss is brussel sprouts, but I can only eat a couple at dinner time. Cucumbers and celery, though, seem pretty safe.
Since you said "alleviate", I've got to assert that a really low carb diet can indeed alleviate type 2 Diabetes. I'm not saying cure. And I'm talking about a really low carb diet with essentially no sugars of any kind (including fruit, though I allow myself 1/8 cup blueberries every morning). I've basically replaced all starches with a mix of oat bran, wheat bran, and psyllium husks, with a bit of xanthan gum form texture. Even salad vegetables need to be a bit limited. And for milk I restrict myself to about 1/2 cup per day, as it has a lot of carbohydrates. Swiss cheese though doesn't. A few other cheeses are also fairly safe.
This is a bit tricky to do in actuality, as it can be hard to balance the vitamins. Nuts are supposed to be reasonably safe, as are kidney beans. I'm hoping that green beans will work out, as they have extra vitamins, and I get tired of spinach and lettuce, but cabbage adds too many carbs.
That said, I also take a minimal dose of metformin. I'm told that you need to have your kidney function checked if you do, but I seem to tolerate it well.
Well, it has risen higher than I thought possible. But as it has no underpinnings so it can collapse totally. It probably won't, there are still people who buy tulips. But it wouldn't surprise me if it did. And it's already at the point where it's extravagantly expensive to operate.
Plausible, but many people think themselves rich off BitCoins, and plan to keep holding them indefinitely as the value keeps increasing. And they think themselves rich *now*. One could say that they got rich off hoarding without doing much violence to the terms.
I think most hoarders, if challenged, would claim that they intend to use the stuff they are hording "someday". This is unlike the multiple cans of beans that I store, with the plan where I actively use first the ones with the oldest "sell by" date. That's stock rotation and reserve rather than hoarding. But I have to cop to hoarding floppy disks, which I planned to use "someday", but never got around to using, and never had a plan for when I would use them.
If you sell something you bought, or try to sell it, I don't think the term is "hoarding" but rather "storing". Hoarders collect things they have no intention of selling. It has all the costs of consuming and none of the benefits (well, except that you don't gain weight).
If people are hoarding ether-cats or whatever these things are, how does that make the world worse? (Well, it does because blockchains require electricity, but I mean in comparison to other things they might hoard.) It's not as if hoarding these ether-cats deprives anyone else of anything. It's more like hoarding original art-work when good quality copies are already in circulation...but it's not even as damaging as that.
Disclaimer: I am a tea drinker: 100 teabags for $4.
Not at a tea house. Coffee is cheap at home too. FWIW, I'm not a gourmet about coffee, but I don't like french roast at all, considering it only slightly superior to italian roast. I prefer coffee that I don't need to add something to to make it vaguely drinkable. (So also no long term sitting on a burner.) Fortunately there are now (since around the 1980s) some decent instant coffees. They aren't great, but as I said I'm not a gourmet.
That said, if a place doesn't offer a reasonable place to sit down and read, I don't want to pay their prices for coffee. Most places I go do offer this, but none of them are chains. Also none of them "feature" piped entertainment of any sort, including muzak. It comes in and I go out.
So I disagree with his analysis. He's just not properly pricing a decent place to sit yourself. You aren't paying for the coffee, that's just the nominal fee bearing part of the package. (But why anyone would pay for most StarBucks places I have no idea. Even in a strange town I prefer to look elsewhere.)
I understand that the author of Eliza didn't think it was an AI program, but it was commonly called one anyway, despite being an intentional attempt to deny the idea. So calling similar things an AI today isn't redefining the term. Abusing it, perhaps, but not redefining it.
If you want a redefined term look at what gets called a robot today and contrast it with any use of the term before 1960.
You need a better memory. There was a time when Eliza was commonly called an AI program. Certainly Arthur Samuel's Checkers program was. Most modern things touted as AIs are considerably advanced over either of those.
Saying they aren't smart depends on your definition of smart.
And the GP criticizing them as scripts caused me to wonder how much of what he does could be considered scripted.
Most discussion of AI that isn't extremely technical is so full of fuzzy terms that it's almost meaningless. What you can say is what it does: This thing learns to do object recognition to a reasonable quality rather more quickly than prior ones did...and it was written by a program designed to write other programs. I suspect that it can't do anything else, but that, alone, is impressive.
FWIW, I don't believe that "general intelligence" exists. I believe that what we think of as intelligence is composed of lots and lots of narrow AIs, some of which specialize in routing problems to the appropriate "program", etc. And from any intelligence I'm rather certain that GÃdel's Incompleteness theorem implies that there are classes of problems that it cannot solve. This isn't a direct implication, as no system is going to have an infinite amount of storage, whatever meaning of infinite you choose to use, but I believe it's an indirect implication. Certainly everyone I've ever met has a limited ability to handle nested recursive problems.
Simple example (Many people can solve this, but it displays the problem): You see a man standing beside a picture and he says to you: "Brothers and sisters have I none, but that man's father is my father's son".
You are clearly right that the degree of hype depends on where you read the article, but most people don't read the articles aimed at specialists.
That said, you seem to be agreeing with my point that the high quality 3-D printers are designed for specific jobs. The one that prints titanium can't print cement, etc. This means that their utility in space is limited, though there are other places where they can be extremely useful. I agree that it's reasonable that the high quality printers be more expensive, but that *is* a factor limiting some of their uses, so it's not only the limitation in the materials that they can handle, the speed they can print, the resolution they can print at, or the maximum/minimum size they can print...though those are also limitations on any particular printer. Yes, this is a reasonable trade-off, but it argues against "just ship a 3-D printer with your space ship to make all the spare parts". (Well, the cost isn't a argument against *that* use case, but the rest are.)
What I meant to imply was that there was a continuous (though not equally dense) gradation between learning to tie your shoes and having your head knocked off, and that the article title was absurd.
That said, while I distinguish between obsessive behavior and addiction, I can't draw a line to separate the edge cases. I suspect that "cell phone addiction" is more nearly obsessive behavior, but it's close enough to the boundary that I'm not sure. Similarly for "gambling addiction". OTOH, tobacco is clearly addictive.
Actually, under current law, I believe that if you are "operating" a self-driving car you need to have a license and be in a legal condition to drive. This may vary from state to state, of course.
You're also dismissing other risks with autonomous vehicle networks coming under attack and being responsible for causing more deaths than we have today
You haven't shown that these risks are real.
He may not have the precise risks correct, but there will be problems that haven't been prepared for. These are complex systems, and it's impossible to thoroughly test them before distribution. But I see no reason to believe that they'll average worse than humans. That, however, is something that only time will reveal.
OTOH, it's nearly certain that they will have unexpected failure modes. This really needs to be prepared for, but we all know it won't be. If there's a major problem and all the autonomous cars in a city need to be shut down, there won't be a sufficient backup system. The only one that occurs to me is a manually operated public transit system, and that would be far too expensive to maintain at even a minimal level of utility. (Note that I did not say use. This is a backup to the normal system, so while it needs good testing, it can't be a part of normal operations.)
A third alternative is that this is an expanded prototype project that got hyped by a marketing team that didn't bother to understand it. Of course, it's also true that what you call "1%" problems is actually a lot rarer than 1%, but that doesn't suffice, as they need to be a lot rarer than 1%.
The times I've checked that kind of thing out, each particular kind of production environment needed a custom designed 3-D printer that essentially only worked in that environment...and was a lot more expensive than the standard 3-D printer.
So I expect the GP was correct for general purpose 3-D printers. This is a long way from saying they're useless, but they appear to be significantly over-hyped.
Given the way that he has (apparently) defined addiction, there is no survivable single method that overcomes all addictions, as some are necessary to the continuance of life. Oxygen, e.g. Or food.
That said, the are processes that can be used to defeat any defined addiction, given that the addict really thoroughly wants to do so. And for most addictions that can be survived. This doesn't mean or imply quick or easy, or without problems along the way. This doesn't mean automatic recovery from damage that the addiction may have caused. A "recovered" alcoholic should never drink even a small quantity of alcohol, as their system has not reset, it's just that the habit pattern that caused them to drink has been sufficiently weakened that they can avoid it, but there are various internal chemical processing pathways that have been permanently damages, and probably will never recover.
In your usage is there a difference between addictive and obsessive? There is in mine, and in my usage while cell-phone usage can be obsessive, I've seen no evidence that it can be addictive.
OTOH, of course it alters the brain. Nearly(?) every experience you have does so. The important questions are: In what way? Are the effects reversible? etc.
I also. Reading short squibs on the Internet is one thing, but reading an article is something else, particularly if I might want to be using my computer while doing so...but also if I'm not at home.
It's too bad, but electronic media are barely tolerable, and usually unusable. And it's not like they hadn't been told that before they made their decision. Perhaps I'll someday get an e-book reader, and that might solve at least part of the problem, but tests have shown that even under optimal conditions people learn more slowly from electronic media than from books. If they've figured out why, I didn't see the results.
Why are you surprised that there are things that are called yetis? This is just a comment about what the evidence shows those things appear to be. Don't be hypnotized by a name, or what fools say about the thing bearing the name.
And *that's* the point. I think I might be willing to accept that the results were probably correct if it produced the same result on the same information when it was included as a part of a large double-blinded sample, where most of the answers had known values and the evaluator was totally independent of all interested parties. I'd even accept the results as grounds for stating the probability of the result. But the evaluator needs to be independent and the test needs to be double blind with both known good and known bad values.
Debugging the code is important, but that's a separate thing from validation. Open code can easily have undetected bugs, and evaluation here is a separate skill from programming, so there could easily be "reasonable assumptions" that are false. So the appropriate methodology is a double blind test where both the percentage of false positives and false negatives are revealed in the same dataset that includes the DNA being tested.
Slashdot Mirror
There are lots of different predictions, and you can always find wrong ones. If I looked hard enough I could probably find one that projected glaciers in London. So you can't prove anything by retrospectively picking a prediction that was wrong. You need to either prospectively predict it was wrong and publish refutable reasons, of do a massive retrospective survey of all published models, which is a task nobody has been up to. The IPCC(?) did a prospective report on a bunch of models that they found acceptable and predicted certain signs of warming that could be expected. The models disagreed about exactly what those signs would be, which is not surprising as they were written by specialists in different areas. They rejected those models that they found too pessimistic (as well as many others), and published their report of what to expect. They were overly optimistic, so their predictions have been refuted, but because of their methodology this doesn't say which models were wrong how. My guess is that part of the mistake was that they rejected models that were too pessimistic, part of the mistake was that different models had interactions between the places where they were correct, and there were also feedback loops that nobody included. So they were all wrong, but not in the direction you appear to be proposing.
Picking one model, or cherry-picking features from a bunch of undocumented models, doesn't prove a goddamn thing.
We should anticipate less so that to compensate for the anticipation effects.
We did. The official reports refused to include projections that were too pessimistic to make the results more politically palatable. To be fair, they also refused to include projections that projected cooling, and there were a couple. Then they took the average of the reports they decided to accept. They had to know they were low-balling the numbers. Unfortunately, it's starting to look like the correct models were some of the ones they rejected for being too pessimistic.
I eat a little broccoli because I like it, but I don't eat much because it's too high in non-fiber carbs. ... When I said a really low carb diet I meant it. What I really miss is brussel sprouts, but I can only eat a couple at dinner time. Cucumbers and celery, though, seem pretty safe.
Since you said "alleviate", I've got to assert that a really low carb diet can indeed alleviate type 2 Diabetes. I'm not saying cure. And I'm talking about a really low carb diet with essentially no sugars of any kind (including fruit, though I allow myself 1/8 cup blueberries every morning). I've basically replaced all starches with a mix of oat bran, wheat bran, and psyllium husks, with a bit of xanthan gum form texture. Even salad vegetables need to be a bit limited. And for milk I restrict myself to about 1/2 cup per day, as it has a lot of carbohydrates. Swiss cheese though doesn't. A few other cheeses are also fairly safe.
This is a bit tricky to do in actuality, as it can be hard to balance the vitamins. Nuts are supposed to be reasonably safe, as are kidney beans. I'm hoping that green beans will work out, as they have extra vitamins, and I get tired of spinach and lettuce, but cabbage adds too many carbs.
That said, I also take a minimal dose of metformin. I'm told that you need to have your kidney function checked if you do, but I seem to tolerate it well.
Well, it has risen higher than I thought possible. But as it has no underpinnings so it can collapse totally. It probably won't, there are still people who buy tulips. But it wouldn't surprise me if it did. And it's already at the point where it's extravagantly expensive to operate.
Plausible, but many people think themselves rich off BitCoins, and plan to keep holding them indefinitely as the value keeps increasing. And they think themselves rich *now*. One could say that they got rich off hoarding without doing much violence to the terms.
I think most hoarders, if challenged, would claim that they intend to use the stuff they are hording "someday". This is unlike the multiple cans of beans that I store, with the plan where I actively use first the ones with the oldest "sell by" date. That's stock rotation and reserve rather than hoarding. But I have to cop to hoarding floppy disks, which I planned to use "someday", but never got around to using, and never had a plan for when I would use them.
If you sell something you bought, or try to sell it, I don't think the term is "hoarding" but rather "storing". Hoarders collect things they have no intention of selling. It has all the costs of consuming and none of the benefits (well, except that you don't gain weight).
If people are hoarding ether-cats or whatever these things are, how does that make the world worse? (Well, it does because blockchains require electricity, but I mean in comparison to other things they might hoard.) It's not as if hoarding these ether-cats deprives anyone else of anything. It's more like hoarding original art-work when good quality copies are already in circulation...but it's not even as damaging as that.
For some reason there are more lolcats than funny dog pix, so I think you're probably wrong. I don't know why, but that's reality.
FWIW, my wife was a dog lover and amateur cartoonist, and she drew a lot more cats than dogs. Go figure.
Disclaimer: I am a tea drinker: 100 teabags for $4.
Not at a tea house. Coffee is cheap at home too.
FWIW, I'm not a gourmet about coffee, but I don't like french roast at all, considering it only slightly superior to italian roast. I prefer coffee that I don't need to add something to to make it vaguely drinkable. (So also no long term sitting on a burner.) Fortunately there are now (since around the 1980s) some decent instant coffees. They aren't great, but as I said I'm not a gourmet.
That said, if a place doesn't offer a reasonable place to sit down and read, I don't want to pay their prices for coffee. Most places I go do offer this, but none of them are chains. Also none of them "feature" piped entertainment of any sort, including muzak. It comes in and I go out.
So I disagree with his analysis. He's just not properly pricing a decent place to sit yourself. You aren't paying for the coffee, that's just the nominal fee bearing part of the package. (But why anyone would pay for most StarBucks places I have no idea. Even in a strange town I prefer to look elsewhere.)
I understand that the author of Eliza didn't think it was an AI program, but it was commonly called one anyway, despite being an intentional attempt to deny the idea. So calling similar things an AI today isn't redefining the term. Abusing it, perhaps, but not redefining it.
If you want a redefined term look at what gets called a robot today and contrast it with any use of the term before 1960.
Yep. That was where I started.
You need a better memory. There was a time when Eliza was commonly called an AI program. Certainly Arthur Samuel's Checkers program was. Most modern things touted as AIs are considerably advanced over either of those.
Saying they aren't smart depends on your definition of smart.
And the GP criticizing them as scripts caused me to wonder how much of what he does could be considered scripted.
Most discussion of AI that isn't extremely technical is so full of fuzzy terms that it's almost meaningless. What you can say is what it does:
This thing learns to do object recognition to a reasonable quality rather more quickly than prior ones did...and it was written by a program designed to write other programs. I suspect that it can't do anything else, but that, alone, is impressive.
FWIW, I don't believe that "general intelligence" exists. I believe that what we think of as intelligence is composed of lots and lots of narrow AIs, some of which specialize in routing problems to the appropriate "program", etc. And from any intelligence I'm rather certain that GÃdel's Incompleteness theorem implies that there are classes of problems that it cannot solve. This isn't a direct implication, as no system is going to have an infinite amount of storage, whatever meaning of infinite you choose to use, but I believe it's an indirect implication. Certainly everyone I've ever met has a limited ability to handle nested recursive problems.
Simple example (Many people can solve this, but it displays the problem):
You see a man standing beside a picture and he says to you:
"Brothers and sisters have I none, but that man's father is my father's son".
You are clearly right that the degree of hype depends on where you read the article, but most people don't read the articles aimed at specialists.
That said, you seem to be agreeing with my point that the high quality 3-D printers are designed for specific jobs. The one that prints titanium can't print cement, etc. This means that their utility in space is limited, though there are other places where they can be extremely useful. I agree that it's reasonable that the high quality printers be more expensive, but that *is* a factor limiting some of their uses, so it's not only the limitation in the materials that they can handle, the speed they can print, the resolution they can print at, or the maximum/minimum size they can print...though those are also limitations on any particular printer. Yes, this is a reasonable trade-off, but it argues against "just ship a 3-D printer with your space ship to make all the spare parts". (Well, the cost isn't a argument against *that* use case, but the rest are.)
What I meant to imply was that there was a continuous
(though not equally dense) gradation between learning to tie your shoes and having your head knocked off, and that the article title was absurd.
That said, while I distinguish between obsessive behavior and addiction, I can't draw a line to separate the edge cases. I suspect that "cell phone addiction" is more nearly obsessive behavior, but it's close enough to the boundary that I'm not sure. Similarly for "gambling addiction". OTOH, tobacco is clearly addictive.
Actually, under current law, I believe that if you are "operating" a self-driving car you need to have a license and be in a legal condition to drive. This may vary from state to state, of course.
You're also dismissing other risks with autonomous vehicle networks coming under attack and being responsible for causing more deaths than we have today
You haven't shown that these risks are real.
He may not have the precise risks correct, but there will be problems that haven't been prepared for. These are complex systems, and it's impossible to thoroughly test them before distribution. But I see no reason to believe that they'll average worse than humans. That, however, is something that only time will reveal.
OTOH, it's nearly certain that they will have unexpected failure modes. This really needs to be prepared for, but we all know it won't be. If there's a major problem and all the autonomous cars in a city need to be shut down, there won't be a sufficient backup system. The only one that occurs to me is a manually operated public transit system, and that would be far too expensive to maintain at even a minimal level of utility. (Note that I did not say use. This is a backup to the normal system, so while it needs good testing, it can't be a part of normal operations.)
A third alternative is that this is an expanded prototype project that got hyped by a marketing team that didn't bother to understand it. Of course, it's also true that what you call "1%" problems is actually a lot rarer than 1%, but that doesn't suffice, as they need to be a lot rarer than 1%.
The times I've checked that kind of thing out, each particular kind of production environment needed a custom designed 3-D printer that essentially only worked in that environment...and was a lot more expensive than the standard 3-D printer.
So I expect the GP was correct for general purpose 3-D printers. This is a long way from saying they're useless, but they appear to be significantly over-hyped.
Given the way that he has (apparently) defined addiction, there is no survivable single method that overcomes all addictions, as some are necessary to the continuance of life. Oxygen, e.g. Or food.
That said, the are processes that can be used to defeat any defined addiction, given that the addict really thoroughly wants to do so. And for most addictions that can be survived. This doesn't mean or imply quick or easy, or without problems along the way. This doesn't mean automatic recovery from damage that the addiction may have caused. A "recovered" alcoholic should never drink even a small quantity of alcohol, as their system has not reset, it's just that the habit pattern that caused them to drink has been sufficiently weakened that they can avoid it, but there are various internal chemical processing pathways that have been permanently damages, and probably will never recover.
In your usage is there a difference between addictive and obsessive? There is in mine, and in my usage while cell-phone usage can be obsessive, I've seen no evidence that it can be addictive.
OTOH, of course it alters the brain. Nearly(?) every experience you have does so. The important questions are: In what way? Are the effects reversible? etc.
I also. Reading short squibs on the Internet is one thing, but reading an article is something else, particularly if I might want to be using my computer while doing so...but also if I'm not at home.
It's too bad, but electronic media are barely tolerable, and usually unusable. And it's not like they hadn't been told that before they made their decision. Perhaps I'll someday get an e-book reader, and that might solve at least part of the problem, but tests have shown that even under optimal conditions people learn more slowly from electronic media than from books. If they've figured out why, I didn't see the results.
Why are you surprised that there are things that are called yetis? This is just a comment about what the evidence shows those things appear to be. Don't be hypnotized by a name, or what fools say about the thing bearing the name.
And *that's* the point. I think I might be willing to accept that the results were probably correct if it produced the same result on the same information when it was included as a part of a large double-blinded sample, where most of the answers had known values and the evaluator was totally independent of all interested parties. I'd even accept the results as grounds for stating the probability of the result. But the evaluator needs to be independent and the test needs to be double blind with both known good and known bad values.
Debugging the code is important, but that's a separate thing from validation. Open code can easily have undetected bugs, and evaluation here is a separate skill from programming, so there could easily be "reasonable assumptions" that are false. So the appropriate methodology is a double blind test where both the percentage of false positives and false negatives are revealed in the same dataset that includes the DNA being tested.