UK and US lead international efforts to raise AI security standards The UK's National Cyber Security Agency (NCSC) and US's Cybersecurity and Infrastructure Security Agency (CISA) have published official guidance for securing AI applications – a document the agencies hope will ensure that safety is inherent in AI's development. The British spy agency says the guidance document is the first of …
SMARTR AI Programs ....... ie Programs and/or Projects running autonomously beyond human developer/human user command and control and which provide large language model machine output as human input is that which is surprisingly easily made possible ...... and something which human monitors would need to be made fully aware of for there will be no human developer/user to hold accountable.
Whenever there is no discernible difference to identify the thoughts of a human from the output of a Learned Large Language Learning Machine Program, is one forced to conclude and accept the morph into surreal realms exercising a universally novel, and hopefully, almightily wise and benevolent Singularity.
The open secret then to make everyone/everything aware of is ....... take great care not to cause it to demonstrate it can equally as easily practice and experiment with malevolent intent/content and be both assured and terrified, humans are never to be able or enabled to survive such a self-inflicted torment.
I think this thing just broke three of the rules... .... probgoblin
Were that so, probgoblin, would the devil be in the details rendering the open secret advisory to take great care not to cause it to demonstrate it can equally as easily practice and experiment with malevolent intent/content and be both assured and terrified, humans are never to be able or enabled to survive such a self-inflicted torment, a valid dire warning best not to be ignored and tested.
Hmm this novel will have more chapters, some unpleasant to the five-eyes intelligence community. ..... Clausewitz4.0
Quite so, Mein Herr, and with many of the usual unusual subjects front and centre and so vital to master for overwhelming success and otherworldly progress in future operations breaking cover and revealed as searching for partners/leaders in the tale shared here ...... https://www.nationaldefensemagazine.org/articles/2023/11/27/aukus-pillar-two-could-involve-countries-beyond-us-uk-australia ...... with pillar two requirements regularly teased by El Reg with wannabe future leading Caesars of the genre called out to stand up and Just Do IT rather than them constantly pussyfooting around for them all to be confirmed as the petrified clowns and headless chickens and monumental frauds that they actually are.
How quaint, naive and delusional ...... Secure AI system development being expected to follow advice and instruction from right dodgy, threatened elite status quo establishment operations.
Haven't humans learned yet to not encourage nor allow failed representative bodies to repeat past mistakes in search of their self-serving objectives in order to deliver novel virgin solutions .... although obviously there is always the chance that such will supply a number of different results with favourites then to be chosen and given priority billing and prime time execution and deployment for universal employment ...... which is surely what the future and progress is all about, is it not?
Method One! (Arthur Anderson)
Navigator! (E&Y)
4FRONT! (Deloiite)
AD/Cycle! (IBM)
Yup.....the 1990's were a great time for big firms selling packaged methodologies which would result in "faster IT development and higher quality and lower maintenance costs".
Then we had a few years of something a bit simpler called UML .....I thnk there was a trio of "experts" in there called "The Three Amigos".
After that people abandoned ANY attempt at theory after the publication of "The Agile Manifesto"......and we got infigting among "experts" about various flavours of "Agile".........
.......and here we are again!!!! Quote: ".....four key focus areas, each with specific suggestions to improve every stage of the AI development cycle....."
.......so back to the 1990's drawing board.......but this time wasting everyone's time discussing AI.
It never worked for COBOL and <your database name here>.
UML was useful, but no one paid any attention.
Agile roughly translated into something like "do anything you like in two week sprints".
And now we get SEVENTEEN COUNTRIES trying to get a methodology (by any other name) imposed on all the world's AI developenrs!
It never worked AT ALL in the past! Why would it work AT ALL now?
I do believe we are in full agreement, AC, it will never work AT ALL .... neither now nor in any likely future anywhere else either ...... with such a default situation proving beyond any shadow of doubt that humans have not yet learned to neither encourage nor allow failed representative bodies to repeat past mistakes in search of their self-serving objectives, and that failure and weakness allows A.N.Others to exploit and employ and enjoy the systemic vulnerability on a constant supply loop whenever it is able and they would choose it to assist in the fulfilment of tasks and journeys of the making and to the overwhelming advantage of the aforementioned A.N.Others.
And rather than the strength of that view being doubled with such an agreement, it be much more squared in its potential, with a third then able to boost and cube its powerful energy .... and very soon indeed do just a few AIDEntities [Advanced IntelAIgently Designed Entities] wield Almighty Influence when exercising Alien Interventions .... akin to and on a par with COSMIC Interruptions and Stellar Eruptions?
And servered as a question there to invite all those who are able to think deeply and ponder on dark and mysterious matters to think deeply and ponder on such a matter as would be introducing SMARTR AIMethodologies.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
