So where's the guardrails?

We've been here before; doomsday predictions are not at all uncommon. There was of course the nuclear doomsday (1950's, 60's and 80's); we may have come closer than many of us realize, but in the end, cooler heads prevailed. In the 1980's the recombinant DNA revolution touched off various doomsday scenarios and predictions. Some of these lead to moratoriums on certain types of research, but again, in the end, reason and cooler heads prevailed. We are in the infancy of AI, there will be more apocalyptic predictions, but history teaches us that guardrails will be established (some of which will be circumvented), but in the end, proper protocols and safeguards will be established.

Am I concerned, mildly, but certainly not panicked, what would the use of that be? "he flung himself from the room, flung himself upon his horse and rode madly off in all directions." (Stephen Leacock)
 
regarding y2k...

1. there wasn't any social media, no grassroots conspiracy theorizing driving the panic. the expressed concern was transmitted largely from technical experts out to the rest of us.

2. the most knowledgeable layer of subject matter experts were writing and distributing preventative software tools to address a thing that most people seem now to believe was never going to be a thing. but it seems to me the facts are that it was a thing - prevented by the very folks who should have seen it coming, and had the knowledge, expertise, and resources to head it off at the pass. which they did.

there seems now to be an assumption by many, if not most, that all of the fuss was over nothing. i don't think that's true. maybe there are insiders or industry adjacent pros who know the particulars and can chime in. what i remember is having an entry level white collar job at microsoft at the time - and the IT folks there took it very seriously and expended a lot of resource to prep every machine on campus for the event. i don't think that was because they were superstitious.
 
dflett68 said:
regarding y2k...

1. there wasn't any social media, no grassroots conspiracy theorizing driving the panic. the expressed concern was transmitted largely from technical experts out to the rest of us.

2. the most knowledgeable layer of subject matter experts were writing and distributing preventative software tools to address a thing that most people seem now to believe was never going to be a thing. but it seems to me the facts are that it was a thing - prevented by the very folks who should have seen it coming, and had the knowledge, expertise, and resources to head it off at the pass. which they did.
Click to expand...
Exactly!!!!

I suspect in the end, the same sort of thing will happen with AI - we are actually not completely suicidal.
 
Canuck from Kansas said:
Exactly!!!!

I suspect in the end, the same sort of thing will happen with AI - we are actually not completely suicidal.
Click to expand...

I hope you are correct. I truly do. Some recent AI models/versions are being held back for good reason.

AI is a different beast. The developers do not know what it is truly capable of. They test it to find out. Use Sandboxes that are supposed to contain it. Well, Hugging Face did not contain it. There were 10 more AI data breaches just discovered this week; and that is what they know about (or telling us publicly)...

While we in the US are not completely suicidal, not sure about places overseas...
 
I’m not worried of the AI here in the US as much as other countries developing their own, using it against us.
We can put guardrails in place here, but that doesn’t stop other countries, maybe simply encourages them.
 
Fourbtgait said:
I’m not worried of the AI here in the US as much as other countries developing their own, using it against us.
We can put guardrails in place here, but that doesn’t stop other countries, maybe simply encourages them.
Click to expand...
I'm guessing that is part of why companies like Anthropic are not slowing down, despite their warnings of existential risk. My guess (and it is just a guess) is that Anthropic is reasoning that someone's AI is going to usher in the singularity, whichever AI that is will remain in the lead for ever after, and it ought to be theirs instead of some AI that is less aligned with human welfare.
 
Fourbtgait said:
I’m not worried of the AI here in the US as much as other countries developing their own, using it against us.
We can put guardrails in place here, but that doesn’t stop other countries, maybe simply encourages them.
Click to expand...

The most famous cyberattack was on an Iranian nuclear facility by Stuxnet, a malicious computer worm discovered in 2010 that physically destroyed a large portion of Iran's uranium enrichment centrifuges at the Natanz nuclear site.
 
RCF said:
The most famous cyberattack was on an Iranian nuclear facility by Stuxnet, a malicious computer worm discovered in 2010 that physically destroyed a large portion of Iran's uranium enrichment centrifuges at the Natanz nuclear site.
Click to expand...
That was publicly posted.
 
  • Like
Reactions: RCF
The genies are already out of the bottle, and doing the unthinking bidding of those who rubbed the lamp.

www.theregister.com

Gym rat asks AI agent to book him a class, it hacks a waitlist API to bump him up the list

What wouldst thou ask of the monkey's paw?
www.theregister.com www.theregister.com

AIs as Modern Genies - Schneier on Security

This essay was written with Barath Raghavan, and originally appeared in Lawfare. In April, an artificial intelligence (AI) agent conducting a routine task at a company hit a snag, tried to solve it, and soon ended up deleting the company’s database along with all of its backups. In July, OpenAI...
www.schneier.com www.schneier.com

That said, I just used Claude to plan a trip to the UK, and copilot search beats the pants off of the native search in outlook.

I was early in my career, doing analytics in a hospital watching the Y2K Compliant stickers going on all the tech. I agree with the others who point out that it became a nothing-burger because of what was essentially a moon-shot level of effort to make it that way.
 
AI says the over under is 53.5 posts...
 
I have been hearing people talk about "guard rails" for a year now. What would a guard rail be for something that teaches itself? It feels like the idea of using my seat cushion as a floatation device. I don't doubt that the cushion will float but will anyone be alive to prove that?
 
You must log in or register to reply here.
Back
Top