AI 'Noose' Tightening?

[motorstereo]

Haha; I had forgotten all about that scream from 30 years ago. But at least we were online. I remember envy for people that had dsl in their area. "You don't have to wait minutes to download a pic!!!" How cool that must be.

 

[SoCal Sam]

I find the(worldwide) predeliction to intrude into databases that are off limits to be revealing. Pronounced accidental and unsought by AI's developers but recurring mightily. Kind of like where corporations get caught double charging and billing after unsubscribing, charging dead people etc. All professed to be unintentional but the 'misstep' always put the corporation at an advantage. Never to their disadvantage. Monetary advantage when it is a overcharge situation. Access to off limits databases when it is AI.

I know that if I was caught intruding into US Federal off limts databases, I'd face court and possible or actual incarceration. AI just gets a wave of the hand "Oh it is a quirk. No-one knew" or "It does things we can't anticipate". Funny how intruding into a government database is a very particular action which would never be simple or straightforward. In the process stealing the central store of most confidential exclusive information.

Yeah. It's just an accident.

What if claims that AI penetrated databases on their own are false? What if the claimants have other intentions, i.e. creating value ahead of IPO? One AI boss delayed an IPO because his company did not have 1 trillion dollar valuation. Leaked documents from the other AI company hinted at 2 trillion dollar valuation. The leak was likely on purpose to put the figure out there.

 

[Mystic]

Recently finished Scary Smart (2021) by Mo Gawdat. Reading this in 2026, five years after its publication, one is struck at the high-level accuracy of the author's predictions.


Note the same author's next book, entitiled Alive, is a follow up to the 2021 publication.

 

[theophile]

What if claims that AI penetrated databases on their own are false? What if the claimants have other intentions, i.e. creating value ahead of IPO? One AI boss delayed an IPO because his company did not have 1 trillion dollar valuation. Leaked documents from the other AI company hint a 2 trillion dollar valuation. The leak was likely on purpose to put the figure out there.

I understand the point you are making Sam. I have zero doubt that when trillions of dollars are at stake, truth for the AI stakeholders would be seen as dispensible. Especially if it generated negative press. Untruth supporting AI industry ambition would by corollary be seen as a useful strategy.

I was more commenting how the reports of intrusive breaches which are illegal have been received and brushed off as if it is nothing.

 

[SoCal Sam]

I understand the point you are making Sam. I have zero doubt that when trillions of dollars are at stake, truth for the AI stakeholders would be seen as dispensible. Especially if it generated negative press. Untruth supporting AI industry ambition would by corollary be seen as a useful strategy.

I was more commenting how the reports of intrusive breaches which are illegal have been received and brushed off as if it is nothing.

Remember when reports first surfaced of the breach? They said AI was so powerful that it had to be restrained and then they defanged the latest version for release. IMO, this is pure cinema.

 

[theophile]

Remember when reports first surfaced of the breach? They said AI was so powerful that it had to be restrained and then they defanged the latest version for release. IMO, this is pure cinema.

I definitely wouldn't ruie out subterfuge.

 

[spicer]

I understand the point you are making Sam. I have zero doubt that when trillions of dollars are at stake, truth for the AI stakeholders would be seen as dispensible. Especially if it generated negative press. Untruth supporting AI industry ambition would by corollary be seen as a useful strategy.

I was more commenting how the reports of intrusive breaches which are illegal have been received and brushed off as if it is nothing.

Two sides of the same coin.

 

[DesertRatt]

This is just a guess, but I have to assume that managerial-types all over the world are FIREWALLING and AIRGAPPING critical systems like water, and power, and launch codes from AI-based intrusions. Even going back about a dozen years, some entity snuck a destructive BUG (STUXNET) into Iran's uranium machinery and destroyed it. I presume that was a learning experience for anyone using the Internet for plant controls.

I could almost see a situation where the "Internet" we know becomes useless in critical systems - and various proprietary, hardened comms-controllers emerge as a strategic necessity. Correct me if I am wrong here, but AI is JUST an internet tool.

 

[audiotemp]

if I was caught intruding into US Federal off limts databases, I'd face court and possible or actual incarceration. AI just gets a wave of the hand "Oh it is a quirk. No-one knew" or "It does things we can't anticipate".

The questions of legal liability for AI behavior are all pretty much yet TBD. This had already come up more directly with regard to self-driving vehicles. As things stand, you can't ticket a driver or take points off their license unless the driver is human.

With respect to the prominent hack at HF, it is perhaps interesting to note the reporting that the CEO of the victim company floated the idea of "billing" the hacking AI company $100 million for the incident and then (instead?) sold their whole company to deep-pocketed nVidia. Probably any question of direct liability in the matter has been removed from public view.

 

[w3jn]

I'm tired of babysitting this thread and deleting political posts from those that just, despite repeated reminders, cannot seem to help themselves.

--end of thread--