Garbage in, garbage out.
We are building AI on the random sewage of the Internet crawl. AI builders can’t conceive of base models built on Reddit wars, Facebook comments and old Twitter Drive-by hot-takes as being problematic. I had the talks, “Brian you are crazy, data is data”.
Calories are calories, there are calories in sewage, most choose not to consume them (willingly 🤓).
But all corporate AI models are built on Internet Sewage (of course there are facts and history too, but low percentage). The ingredients label would show 99% sewage.
And just like the Y2K problem the technicians could not convince that this data, and how it reflected upon the essence of humanity would form a faulty foundation ultimately built on sand.
regarding y2k, at the time, l suggested a means of dealing with it by redesigning compilers esp cobol for instance, such that we could use hexadecimal notation for years. so 1997, 1998, 1999, next is 199A instead of the unavailable ‘2000’. it would *only* apply to the ‘singles digit’ (6 more years to get yer act together) and the compiler would generate the executable code necessary to then deal with the ‘we ran out of dates’ situation.
it would of course only apply to date type variables, a temporary measure until the permanent ’64 bit’ solution was found.
as usual the suggestion was immediately shot down by ‘experts’ who preferred to let everyone panic, then *they* would panic at the prospect of tracking down the original source code, and proceed to throw out perfectly good code otherwise,
stubborn stubborn stubborn.
meh. their loss . . . .
Sure. Parsing digits and treating the 1s column in hex while the rest is in decimal is trivial.
That’s gotta be one of the stupidest ideas I’ve ever heard.
As I remember, Y2K turned out to be a nothingburger, as people made normal fixes, instead of hair-brained BS like you just spouted.
well yw NO CONVERSION NECESSARY ANY TIME THE COMPILER SPOTS THAT
CHAR-ACT-ER STRING IT WOULD PLOP IN THE GIVEN DATE. the users would merely opt for the accurate hexadecimal to ‘indicate’ to the compiler what date was being specified.
199A = 2000
199B = 2001
199C = 2003
199D=2004
199E = 2005
199F = 2006
got it? no calculations or mathematical conversion req’d
a refined and limited measure to cope with the source code language that everyone else but YOU knows ‘had’ no place to PUT year 2000. and the so-called ‘nothing burger’ was NOT so for some using legacy sw and jeepers, didnt have $100,000 to play catch up.
and do feel free to F OFF
So, you gonna replace all years after 1999 with 6 lines of code, but leave the days, months, etc, alone.
Then all the addition, interest rates, etc, gonna be based on some more code you made up, like how many compounding periods between Aug 10, 199B and July 5, 1979, yep, your coding will calculate 199B – 1979, etc. Because all dates after 1999 will be replaced with 6 lines of code assigning new numeric values to strings you made up, where the last digit is hex. Tell us, why stop at F if your just making up strings? Or are your new “years” actually variable names that you assign values to, and even then, why stop at F? You could go through almost the antire ASCII char. set?
Again, the stupidest idea I’ve heard in a long, long time.
wrong agin.
my solution at its most basic uses the fact there are unused bit combinations to allow to indicate in this case years 2000 to 2006. and then the newly created executable code knows well and enough how to handle these newly available representations of year.
all else you describe is merely the fleshing out of operations to deal with the ripple effect on dates as you suggest.
all within the framework of a patch dipping into hex notation.
“with 6 lines of code,”
see this is the flaw in your arguement. what l propose is merely additional code added to the revised compiler that knows how to deal with the additional 7 years. taking however many lines of code in the revised compiler doing the magic. all else shows up in the new executable code exactly as it did in the legacy system because it had no need to change.
way beyond you dude at conceptual level.
The whole issue was using 2 decimal digits for the year.
So, you would have 99 roll over to…9A? Or 99 would roll over to 199A?
If your gonna do that, just as easy to fix the whole 2-digit issue in the first place, and have 99 roll over into 2000, which is pretty much what was done.
“newly created executable code”
Sounds like BS to me, but I figure your expertise is in non-executable code.
“revised compiler” So, writing whole new compilers, now.
OK.
The biggest challenge to the software industry isn’t cheap foreign competition or generative AI. It’s the Dunning-Kruger effect. As witnessed here.
Not only have you fundamentally misunderstood the nature of the problem, your solution would involve significantly more work than just fixing the two-digit year fields in the source code.
It was a financial bopon for Fortran coders 🙂
I did the same thing with y2k as I did with covid, nothing, I ignored it and was correct in doing so.
Regarding Y2K, no one could explain why the computer in my car would care what day, or year, it is.
SAME….
and on the question of AI? l started out in IT in 1970 operating those massive $10,000,000 DEC 10s etc and acquired the usual degree.
l hate and have ZERO faith in where this ‘AI’ is going to go. and the old argument about ‘technology, progress, ’tis a 2 edge sword’, well, we’re gonna see a LOT of the destructive harmful edge imho.
Anyone who takes “AI” seriously, or believes all he sees and hears, is an idiot. This is an ancient golden rule.
Computer programming, nothing more, a computer can only respond to information already programmed in. People seem to think that the machine is sentient, it is not.
You are correct about all we see and hear. Today we have to question everything unless we are directly involved.
In the hands of amoral goons, claiming not just the personal privilege of wealth of power, but all its institutions:
AI=BB.
Can’t wait for Y2K2 or the year 2038 when 32bit flips over to 10000000000….
Pol Pot will turn in his grave. Only 60 years too late. From 2 billion to minus 2 billion. A modern greeny would envy.
The browser I use returns as its first item an AI summary:
e.g. “Internet sewage”
Internet Sewage Infrastructure
The concept of “Internet sewage” refers to the integration of wastewater management systems with Internet of Things (IoT) technology. This approach enables real-time monitoring and prediction of sewage flow, pressure, and quality, allowing for more effective maintenance, reduced costs, and improved environmental outcomes.
Smart Sewers
Several initiatives have demonstrated the potential of IoT-enabled sewage systems. For example, South Bend, Indiana’s sewer system uses sensors to monitor flow and divert water to prevent flooding, while the UK’s sewerage network is exploring the use of IoT technology for real-time prediction of flooding and asset performance analysis.
I’m more concerned about AI being made in the image of its creators. For an example, watch the Star Trek TOS episode “The Ultimate Computer”.
When the intelligence is trained to emulate neurotic, unstable personalities, it doesn’t have much of a chance to be anything else. Why do you think most of the AI chats will only give answers from a California Silicon Valley/ Bay Area perspective?
I hope some random drunk comment I made on the internet twenty years ago has made its way into hundreds of PhD papers.
It’s a search engine, not ai. If it searches the internet to answer questions it isn’t really thinking for itself
“If you understand how it works, its not intelligence.”
“Why would you make self-aware AI? There’s no money there. There’s is nothing there but an ethical can of worms.”
Got some news for ya, people often search the internet to answer questions, too. People think for themselves, don’t they, or are we just search engines too?
We are the sum of all that we have absorbed.
No money here? For ethical canned worms? You could market those by the carload!
One of the weaknesses in the description of Internet information as sewage is the fact that most of the crap on the Internet has been constructed for us and is being pushed at us by the same people developing AI. While we may not know what AI has in store for us, they do.
So basically, AI is like people.
“The trash collected by the eyes gets dumped into the brain,
then it tears into our conscious thoughts, you tell me whose to blame.”
Guns n roses?
It took an ai search to find that.
Yep.
“Don’t Damn Me”, from way back in 1991.
Thinking of overlords.
Bill Gates is trying to take over the planet by buying up all the farmland and controlling the software and all things WHO.
Meanwhile, Elon Musk is taking over the Solar System.
AI models will develop by all viewing the same original data, plus that from AI models. Eventually, all AI models will zero in on the standard AI model, unrelated to facts, because the democracy of AI will all agree as to the facts, true or not.
AI is plagiarism. AI is Blarney Science. AI is lies.
AI is just a name given to a bunch of software, like Perceptrons and Neural Networks.
They have uses, but people think they’ll be some kind of oracle of truth.
You are just as likely to get a correct answer from AIs as you are from Joe the taxi driver.
I’m not given to proselytizing in an angry, scolding manner. I don’t wear a placard that reads: “The End is Near”. I don’t preach doom, gloom, fire, or brimstone. Because God IS LOVE and wants deep communion with each of His creations.
But our culture and people have completely lost our moorings. Once we mocked God and said He is “dead” … we began spinning completely out of control. Worshiping and following AI as the inerrant Book of Woke orthodoxy = certain death. Yeah … AI is a sewer. So is every “fact check” report of disinformation.
The people will eventually wake up from this woke slumber. And their waking won’t be pretty
AI is just programing and will never be anything other than that. Relying or believing computer models is what gave us the covid fraud, the current climate change fraud, and will be used to continue the expanding of control over people.
I doubt that most people understand what they are embracing.
It’s very difficult to edit a book on all the shelves in homes and libraries around the world. Editing that same book in the cloud is a piece of cake.