The Lemon Fool

Sorcery wrote:My mea culpa involves :

Then there are the first two lines
long bit_idx = sbit % 8;
BYTE mask = (1 << bit_idx) - 1;

So if bit_idx is non-zero it definitely doesn't work, the 1 gets left shifted to make zero and the following subtraction of -1 makes the mask =255 for all non zero cases. Think its probable that the bit_idx = zero, case is'nt right either.

The last explanation is wrong, I am mixing up left shift with right shift. A left shift is a multiply, a right is division. It might be wrong but not for the reason given.

So what was that again about "twits" who use divide by 8 rather than right shifts?

XFool wrote:

Sorcery wrote:My mea culpa involves :

Then there are the first two lines
long bit_idx = sbit % 8;
BYTE mask = (1 << bit_idx) - 1;

So if bit_idx is non-zero it definitely doesn't work, the 1 gets left shifted to make zero and the following subtraction of -1 makes the mask =255 for all non zero cases. Think its probable that the bit_idx = zero, case is'nt right either.
The last explanation is wrong, I am mixing up left shift with right shift. A left shift is a multiply, a right is division. It might be wrong but not for the reason given.

So what was that again about "twits" who use divide by 8 rather than right shifts?

This partly explains why one shouldn't mix decimals with bits

THAT ^ there is a modulo not a division
the fact it is a modulo makes the lack of range checking on the result forgiveable as it cannot be greater than 7 ...assuming BYTE is an 8 bit unsigned integer
if it had been a division replacing it with 3 right shifts would possibly not have done what is expected given long is signed (and we don't normally treat bit fields as signed)

the left shift and -1 is a pretty "standard" way of generating a field of ones in a bit mask where bit_idx would be the number of bits set to the right hand side (LSB) end

but having it lying around makes this feel like a homework (or job interview) exercise rather than part of a proper code base but at least it's going in to a variable called mask
doesn't look like you can ever get a mask of 0xFF (8 bits set) which might be what is wanted

and until ChatGPT can explain this stuff better we should be wary of what it suggests

servodude wrote:doesn't look like you can ever get a mask of 0xFF (8 bits set) which might be what is wanted

and until ChatGPT can explain this stuff better we should be wary of what it suggests

Clearly ChatGPT is not the only one suffering from Dunning–Kruger.

You need to review the operators and standard integer promotions, amongst other things. These are not in the curriculum and job interviews just for giggles, but because if you don't know them you can't code correctly (in C, if you want to code in lisp you need to know different things).

9873210 wrote:

servodude wrote:doesn't look like you can ever get a mask of 0xFF (8 bits set) which might be what is wanted

and until ChatGPT can explain this stuff better we should be wary of what it suggests

Clearly ChatGPT is not the only one suffering from Dunning–Kruger.

You need to review the operators and standard integer promotions, amongst other things. These are not in the curriculum and job interviews just for giggles, but because if you don't know them you can't code correctly (in C, if you want to code in lisp you need to know different things).

I must have missed something.
Can you point me to where it can happen with those two lines.

Interesting video on whether open source will dominate closed source for AI (11mins 30 seconds):

https://youtu.be/URvja3IyMDo

Regards,

BT plans to replace 10,000 jobs with AI by 2030.

https://www.theguardian.com/business/2023/may/18/bt-cut-jobs-telecoms-group-workforce

BT has said it will become a “leaner business” as it announced plans to reduce its workforce by as much as 55,000 by 2030, more than 40% of its global employee base, including about 10,000 jobs replaced by artificial intelligence.
.
.
.
For a company like BT there is a huge opportunity to use AI to be more efficient,” he said. “There is a sort of 10,000 reduction from that sort of automated digitisation, we will be a huge beneficiary of AI. I believe generative AI is a huge leap forward; yes, we have to be careful, but it is a massive change.”

servodude wrote:

9873210 wrote:
Clearly ChatGPT is not the only one suffering from Dunning–Kruger.

You need to review the operators and standard integer promotions, amongst other things. These are not in the curriculum and job interviews just for giggles, but because if you don't know them you can't code correctly (in C, if you want to code in lisp you need to know different things).

I must have missed something.
Can you point me to where it can happen with those two lines.

He's disappeared off the face of the earth

GS

Very good results from Nvidia tonight. Up $81 or 26.5% after hours. AMD going up in sympathy, up $10.70 or 9.9%
These are 2 of the most promising stocks tackling AI from the hardware side. On the software side I guess 2 of the most promising stocks would be Microsoft and Google. Both up after hours but much more moderately.

Astonishing rises though.

Sorcery wrote:Very good results from Nvidia tonight. Up $81 or 26.5% after hours. AMD going up in sympathy, up $10.70 or 9.9%
These are 2 of the most promising stocks tackling AI from the hardware side. On the software side I guess 2 of the most promising stocks would be Microsoft and Google. Both up after hours but much more moderately.

Astonishing rises though.

As is Nvidia have the only off the shelf chips for AI, not ideal for the job, but the best currently available in potential volume although apparently demand is making their chips difficult to come by and they have contracted large amounts of the fab space in the major Taiwan plants for production.

The hilarious part was that Wall Street totally missed all of this when their were obvious clues. When Musk said x/Twitter were buying thousands of Nvidia chips it was very clear that Nvidia would beat huge, but right up to the latest results Wall Street wasn’t pushing Nvidia, but the ones I bought on the Musk announcement of Twitter buying Nvidia chips have been a blessing to my p&l.

Maybe Wall Street needs to use AI, although a ten year old could have seen this coming.

Regards,

odysseus2000 wrote:

Sorcery wrote:Very good results from Nvidia tonight. Up $81 or 26.5% after hours. AMD going up in sympathy, up $10.70 or 9.9%
These are 2 of the most promising stocks tackling AI from the hardware side. On the software side I guess 2 of the most promising stocks would be Microsoft and Google. Both up after hours but much more moderately.

Astonishing rises though.

As is Nvidia have the only off the shelf chips for AI, not ideal for the job, but the best currently available in potential volume although apparently demand is making their chips difficult to come by and they have contracted large amounts of the fab space in the major Taiwan plants for production.

The hilarious part was that Wall Street totally missed all of this when their were obvious clues. When Musk said x/Twitter were buying thousands of Nvidia chips it was very clear that Nvidia would beat huge, but right up to the latest results Wall Street wasn’t pushing Nvidia, but the ones I bought on the Musk announcement of Twitter buying Nvidia chips have been a blessing to my p&l.

Maybe Wall Street needs to use AI, although a ten year old could have seen this coming.

Regards,

The trigger for me buying NVDA was finding out Microsoft bought circa $230m of Nvidia accelerator boards to develop ChatGPT4.

New superbug-killing antibiotic discovered using AI

https://www.bbc.co.uk/news/health-65709834

a winner for AI

The morons have been let loose
https://www.bbc.co.uk/news/uk-65746524
his face alone suggests he's not the full ticket

pje16 wrote:The morons have been let loose
https://www.bbc.co.uk/news/uk-65746524
his face alone suggests he's not the full ticket

That's a pretty rude way to talk about Geoffrey Hinton, who is, after all, one of the world's leading experts on AI...

mc2fool wrote:

pje16 wrote:The morons have been let loose
https://www.bbc.co.uk/news/uk-65746524
his face alone suggests he's not the full ticket

That's a pretty rude way to talk about Geoffrey Hinton, who is, after all, one of the world's leading experts on AI...

i just speak as i find

Fascinating polarisation between the experts and long standing practitioners in the field who are warning and many who don't have the expertise who are saying this is all over blown.

As far as I can tell this is the most powerful technology ever developed, with its roots in evolution, the most powerful force known.

I find it impossible to believe that there is not enough potential in this technology to lead to extinction of the current dominant life form and its replacement with a newer and more powerful one, like has happened many times before if you believe in evolution.

Regards,

NVIDIA reaches $Trillion today on the back of AI.

Is this just the beginning.....Have I missed the boat or should I pile in ?

odysseus2000 wrote:Fascinating polarisation between the experts and long standing practitioners in the field who are warning and many who don't have the expertise who are saying this is all over blown.

As far as I can tell this is the most powerful technology ever developed, with its roots in evolution, the most powerful force known.

I find it impossible to believe that there is not enough potential in this technology to lead to extinction of the current dominant life form and its replacement with a newer and more powerful one, like has happened many times before if you believe in evolution.

Regards,

are you a really enthusiastic bot?
Doesn't bother me if you are... everyone's welcome

The big risk with these at the moment is believing the hyperbole, leading to a lot more "Computer says no!" fiascos
I believe you've recently had that kind of thing in the UK with Post Offices being accused incorrectly of fraud
Now that kind of edict will be delivered by something that sounds like a "nice but dim" genial colleage rather than the output of a calculation
- so much easier to swallow but still GIGO

We're not quite at Demon Seed territory - but i CAN imagine folk having to argue with their cars in the forseable future

I really do share your enthusiasm for evolutionary based tech though - I posted at length about it a while ago (might have been on TMF actually! I've been a huge fan for a long time - Genetic Algorithms were my main PhD topic <cough>-teen years ago). It can be really useful - but like real evolution it is hugely wasteful (vs targetted training); we normally need to play the role of intelligent designers to get really practical results

-sd

In the meanwhile, lawyers in a civil case in the US use Artificial Intelligence (ChatGPT) to find cases and extracts from those judgements to support their arguments, and then demonstrate Natural Stupidity in presenting them in their case without checking them first.

Of course, ChatGPT, as it often does, hallucinated wildly—it invented several supporting cases out of thin air and the lawyers are now facing sanctions. viewtopic.php?p=592179#p592179

scotview wrote:NVIDIA reaches $Trillion today on the back of AI.

Is this just the beginning.....Have I missed the boat or should I pile in ?

Have you asked ChatGPT?

odysseus2000 wrote:Fascinating polarisation between the experts and long standing practitioners in the field who are warning and many who don't have the expertise who are saying this is all over blown.

As far as I can tell this is the most powerful technology ever developed, with its roots in evolution, the most powerful force known.

I find it impossible to believe that there is not enough potential in this technology to lead to extinction of the current dominant life form and its replacement with a newer and more powerful one, like has happened many times before if you believe in evolution.

Regards,

I am more sanguine. Not sure it can evolve except with a human to tell it where it's gone wrong. It could be damaging but only if it's given power e.g setting bank rates without oversight. At the moment it can't produce code that works and I don't think I would rely on it's stock picks. As long as it's supervised it's OK, better if it's supervised by layers of decision makers and tested where possible, then we ought to be safe. It should never have the power to do anything where outcomes are beyond it's supervisor's power. I'm thinking of it being used as a doctor's assistant or armed forces fire control. It needs qualified supervision, with the supervisor the only one empowered to actually do anything. Suppose that rules out driverless cars though ...