I'm sorry, it's comparatively early here and so I haven't had enough coffee. Have you just edited the Wikipedia article so that it agrees with your figures (and then intend to use the Wikipedia article to support your argument) ?
Super Artificial Intelligence, a naive approach
- Thread starter ProgrammingGodJordan
- Start date
I'm sorry, it's comparatively early here and so I haven't had enough coffee. Have you just edited the Wikipedia article so that it agrees with your figures (and then intend to use the Wikipedia article to support your argument) ?
So, you've got a computer geek's 1989 paper as your prime source, and then Wiki entries, not directly on the subject. Your original source is where you picked up your awful "roughly" habit of guessing withing a range of 2 orders of magnitude:
You should also have noted that your original source gave two very different figures for the size of the brain. This alone should have stopped you stating brain size here unequivocally.
At least you have acknowledged your cock up. That's a step forward. It may encourage you to be more careful with your claims in future. What it hasn't done, though, is get you to use Wiki properly. Go to the noted source for the figure, in this case:
You'll find that here.
Read it, to find out if it says what Wiki claims it says. You see, with people like you editing Wiki, there can easily be inaccuracies.
Finally, just put "number of synapses in human brain" into your favourite search engine, and you might come up with some interesting stuff. For instance, did you know that males have a much larger number of synapses than females? And that the number of synapses declines with age? Here. Thus any one single figure for the number of synapses in a brain is inevitably wrong for most cases. That undergraduate paper, by the way, gives the total number of synapses in the neo-cortex alone (ie only about 75% of the brain) as between 0.85 x 10^14 and 1.7 x 10^14, with the first figure being of young males, and the latter figure being for 90 year old females.
It also seems that you build and lose synapses all the time, and by that I mean on a daily basis. There are, it seems, variations of between 35 and 50% in the number of synapses in the brain when you are asleep compared with when you are awake. Here is a paper on that subject, but not the source of the above figure.
I don't have the time to do this more thoroughly now, but what this shows is that any simplistic figure for brain size (in synapse terms) is wrong. It is much more accurate to describe the total number of neurons in the brain, rather than the temporary and transient synapses, which are constantly making and remaking themselves (you've hear of "brain plasticity" no doubt).
You should also have noted that your original source gave two very different figures for the size of the brain. This alone should have stopped you stating brain size here unequivocally.
At least you have acknowledged your cock up. That's a step forward. It may encourage you to be more careful with your claims in future. What it hasn't done, though, is get you to use Wiki properly. Go to the noted source for the figure, in this case:
You'll find that here.
Read it, to find out if it says what Wiki claims it says. You see, with people like you editing Wiki, there can easily be inaccuracies.
Finally, just put "number of synapses in human brain" into your favourite search engine, and you might come up with some interesting stuff. For instance, did you know that males have a much larger number of synapses than females? And that the number of synapses declines with age? Here. Thus any one single figure for the number of synapses in a brain is inevitably wrong for most cases. That undergraduate paper, by the way, gives the total number of synapses in the neo-cortex alone (ie only about 75% of the brain) as between 0.85 x 10^14 and 1.7 x 10^14, with the first figure being of young males, and the latter figure being for 90 year old females.
It also seems that you build and lose synapses all the time, and by that I mean on a daily basis. There are, it seems, variations of between 35 and 50% in the number of synapses in the brain when you are asleep compared with when you are awake. Here is a paper on that subject, but not the source of the above figure.
I don't have the time to do this more thoroughly now, but what this shows is that any simplistic figure for brain size (in synapse terms) is wrong. It is much more accurate to describe the total number of neurons in the brain, rather than the temporary and transient synapses, which are constantly making and remaking themselves (you've hear of "brain plasticity" no doubt).
Last edited:
A 32 year old paper on the brain? Really!?! You'd give that credence? They hadn't even started MRI scanning the brain in 1985. (Which, by the way, is how they quickly disposed of the left brain/ right brain theory). It wouldn't take you long to find later papers on the subject.
ProgrammingGodJordan
Banned
(1)
I had already mentioned that I briefly read the first source, and related all instances of 10^14 to ibm's chip.
Even now, you may still find sources that report 10^15 overall.
But I use wikipedia's neuron url as final reference.
(2)
A more general approximation is roughly 10^15 for children, and roughly 10^14 for adults...
(3)
I had already encountered the koch link https://christofkoch.com/biophysics-book/ and also the same book via google books, long before you had posted them.
I could not find any pertinent data.
(4)
The wikpedia correction was made by me, but the previous value was not written by me.
This is clearly seen in edit history, as I mentioned in post #234.
(5)
I created the wikipedia account today, to make the correction.
It appears people like you had not noticed the error, and so it would have gone unchecked, if not for beings like me.
Last edited:
Porpoise of Life
Illuminator
- Joined
- Oct 5, 2014
- Messages
- 4,950
You just edited that article to fit your beliefs! You don't get to edit an article to agree with you and then claim it as a source for claims you made before the edit.
You're being very dishonest.
Squeegee Beckenheim
Penultimate Amazing
- Joined
- Dec 29, 2010
- Messages
- 32,124
Yes. Can you please explain exactly what sources you used for the figures of 1015, 1016 and 10, please? A source for all three, please.
Just to summarise the position......
ProgrammingGodJordan is suggesting that we're less than three years away from having human-comparable A.I.. The basis for this is a measure of the processing capacity of the human brain, the processing capacity of current A.I. systems and an application of Moore's Law to get from the latter to the former.
As I see it, the major problems ProgrammingGodJordan's assertion has are as follows:
It seems that we have a whole lot of nothing here - unless of course I have misunderstood the cut and thrust of this thread.
ProgrammingGodJordan is suggesting that we're less than three years away from having human-comparable A.I.. The basis for this is a measure of the processing capacity of the human brain, the processing capacity of current A.I. systems and an application of Moore's Law to get from the latter to the former.
As I see it, the major problems ProgrammingGodJordan's assertion has are as follows:
- Attempts to quantify the processing capacity of the human brain are speculative at best.
- The processing capacity of current A.I. quoted was vastly overstated (by several orders of magnitude)
- Even by his own, flawed, methods of calculation and dodgy sources, the human brain is between 100 and 10000 times more powerful than the current A.I., instead ProgrammingGodJordan has used a factor of 3
- As the physical limits of chips are being approached, it's very likely that Moore's Law no longer applies
It seems that we have a whole lot of nothing here - unless of course I have misunderstood the cut and thrust of this thread.
Last edited:
ProgrammingGodJordan
Banned
No.
(A)
The article was already off, before I got there.
ProgrammingGodJordan said:
(B)
If you pay a little more attention, you would realize that 1014 is now there instead of the prior 1015.
Last edited:
ProgrammingGodJordan
Banned
Sources were already provided.
ProgrammingGodJordan
Banned
Yes, like your comrades' writings, this last writing of yours shows misunderstandings.
(1)
I had stated modern AI as 10^14 synapses, according to IBM source. This was not 'vastly overstated', because IBM reported roughly 10^14 synapses.
(2)
By my references, IBM had actually already achieved human level numbers, of roughly 10^14 synapses, in 2012 (but these are not synapses to exactness).
Last edited:
Squeegee Beckenheim
Penultimate Amazing
- Joined
- Dec 29, 2010
- Messages
- 32,124
So, basically, you skimmed the first paragraph, saw what you thought was information which backed up what you'd like to believe is true, and then cited it as fact? That doesn't sound like a rigorous methodology to me. Why should anybody take you seriously after such an admission?
If you now believe that the information in that paper can be dismissed, then why are you using it as a source of 10 synapse operations a second?
Squeegee Beckenheim
Penultimate Amazing
- Joined
- Dec 29, 2010
- Messages
- 32,124
If that's the source of the information, then that is what you should cite. If you want to use that paper as a source, you will have to link to that.
ProgrammingGodJordan
Banned
Small nitpick: 10 impulses per second.
(A)
Anyway,
I didn't need to believe any of the data.
I observed the data to be probably valid.
Also, I couldn't find any later data regarding impulses per second.
(B)
Apart from IBM, other regions are approaching human level brain numbers by 2020.
(C)
I actually read through the entire paper, but I related 10^14 with ibm's chip.
Last edited:
Porpoise of Life
Illuminator
- Joined
- Oct 5, 2014
- Messages
- 4,950
The man admits in his article that it's a guess. He even uses the word guess, and there is no reference to any research there.
You just cherrypick what you want to believe and ignore the rest, and sometimes you edit your own sources to agree with you.
You have nothing.
Errr... no
See this exchange - 1014 seems to be overstated by several orders of magnitude.
ProgrammingGodJordan
Banned
(A)
I see that you are late as usual, in multiple ways.
The guy referenced work by neuroscientists:
Principles of Neural Science, by Eric R. Kandel and James H. Schwartz, 2nd edition, Elsevier, 1985.
(B)
The source already said 10^15 when I got there.
It was I who edited it TODAY, to the correct approximation, 10^14.
Last edited:
Squeegee Beckenheim
Penultimate Amazing
- Joined
- Dec 29, 2010
- Messages
- 32,124
Yup. And replaced a figure taken from an undergraduate and graduate university neuroscience textbook with one taken from what is essentially a promotional article by IBM.
Mojo
Mostly harmless
What do these neuroscientists say about the number of operations per second?
ProgrammingGodJordan
Banned
Wrong again.
(A)
The 10^14 would actually disregard my initial calculation, as it would mean that IBM had already computed human level brain numbers. (although those synapses would still be crude)
(B)
I edited wikipedia, today based on other wikipedia data:
https://en.wikipedia.org/wiki/Neuron
https://en.wikipedia.org/wiki/List_of_animals_by_number_of_neurons
Also, Wikipedia had already made reference to Koch (2004), which is entailed to express 10^14. I simply updated the 10^15 to reflect the initial reference. So, the IBM edit reference is valid.
Last edited:
Squeegee Beckenheim
Penultimate Amazing
- Joined
- Dec 29, 2010
- Messages
- 32,124
The whole issue seems remarkably unclear. For the sake of clarity, please provide a source for each.