Comment No, it isn't obvious (Score 1) 18
They could just not run that AI. I wonder if that has dawned on any of them.
Not running a malicious AI doesn't do anything to protect YOU.
You are at risk from OTHER people running a malicious AI.
They could just not run that AI. I wonder if that has dawned on any of them.
Not running a malicious AI doesn't do anything to protect YOU.
You are at risk from OTHER people running a malicious AI.
The US has enjoyed amazing economic success since WW II because it recognized that there's no better investment not just in terms of monetary ROI, but societal ROI.
And the ROI is really, really impressive. Each dollar in the NIH budget gets returned as about 5 (depending on the study) in increased GDP. To my knowledge, there is no other billion-dollar scale investment that has consistent, long-term returns like that.
Congress knows this fact. They commissioned many of the studies showing it. It is difficult to believe the Executive Branch is not aware of it. If we really wanted to be America First, rather than reducing the NIH budget, we'd be promoting another run of annual increases with the intent of doubling that allocation in real dollars over the coming decade. Contact your congressional representatives today to advocate for the idea.
So don't allow your model a tool to do that?
Do you even know how any of this works? Because it seems you don't know how any of this works. You can restrict what commands a model can run, so give it a setup that denies any access to `ip` or `networkmanager` or whatever.
Try not being a paranoid ignorant.
You could try Fsck'n off. Restricting the commands seemed to work SO well for the people who wrote the agents. I am sure YOU can fix it all and everything will be peachy. In the mean time I again suggest Fsck'n off
Next up will be a "national emergency" to prevent or disallow the results of the general election.
Tell me again how badly America needs to pass the SAVE Act.
No, no. Don't let anyone tell you otherwise. Doubling down with long-TDS worked out great. I think Democrats should triple down next time.
That'll really teach 'em.
And pay no attention to those screaming at you for tossing the ball back through your own goalposts. They're just jealous.
What the hell does any of this even mean?
I have a Tesla Model Y which has (supervised) FSD, but I've never had to intervene critically and it's been thousands of miles. You realize I just get in the car enter an address and it goes there parking-spot to parking-spot -- zero intervention of any kind in all kinds of traffic -- it handles roundabouts that humans have trouble with. Yet nobody who doesn't actually have Tesla FSD believes it works. Nothing we can do or say makes you guys believe it really and truly works. We can show all the stats
To claim that the gap to level 5 is "astronomical" is plain wrong. If Tesla removes the "supervised" label from their FSD
The value gap is indeed astronomical as I stated. Self driving is a long tail problem. Technical progression along the tail doesn't have much effect on the underlying value proposition.
In a Tesla model-y a licensed driver is required to sit in the drivers seat with their hand on the wheel. The driver is required at all times to retain full situational awareness and be ready at a moments notice to intervene. A Tesla model-y can't pick you up from work. It can't run errands for you. It can't be sent off to people who you want to be transported in it.
A level 5 vehicle doesn't need a licensed driver to operate. It doesn't even need a single human being to be present inside of it. It doesn't require any passenger to be paying any attention to anything. During a commute to work the vehicle can be used as a mobile office. When the passenger needs to return it can pick them up without care for parking or where they were originally dropped off.
Waymo only works in specific areas and is professionally managed and monitored.
A country full of level 5 vehicles is one in which the relationship between people and vehicles is fundamentally altered and with it society. Millions of professional drivers are rendered obsolete. Nobody pays for parking. Everyone expects to be able to send their vehicles out to run errands and gather shit for them. Insurance and licensing is severely disrupted. Traffic engineering is dramatically changed as automation enables what was once infeasible. The follow on effects of economically feasible level 5 vehicles is enormous.... just getting 90% of the way there. 95%... 99% or 99.9% doesn't translate into substantive value. You actually have to deliver in full for the technology to actually matter.
This isn't the text of the study but it at least provides useful details instead of the words "statistically significant" followed by meaningless percentages.
Further, actually having at least one dosimeter on board when they are so cheap, light, and compact would go a long way to actually showing what exposures happen, but that would simply create liabilities so it doesnâ(TM)t happen.
It can always be crowd sourced using an app on ordinary cell phones so long as they are stored in the dark or tape is used to cover the camera.
Meaningfully shielding entire (or I guess at least the upper half of) airplane would also make em so much heavier that they'd inevitably end up slower (particularly in climb), which'd in turn mean you'd spend longer time exposed to the radiation and eat into the benefit and further make the boondoggle even more expensive.
You can always use fuel as a radiation shield. Instead of keeping it all in the wings have some of it sloshing around above the cabin.
It's dumb to try to stifle it when it's just developing. For one thing, AI can't take anyone's job. Increasing production has never in history resulted in job loss.
The one thing I'll give Billy credit for is avoidance of this same inductive trap:
"Another reason people underestimate AI is that analogies to the effects of past innovations are misleading. We have no experience with a technology that can be adopted quickly or that can think and move like a human. "
Think about it, cars are already capable of autonomous driving
There has never been even a single level 5 self driving car that has ever existed. The difference in value (and associated disruption) between cars that can mostly drive themselves and cars that can fully drive themselves is astronomical.
My favorite line from his entire post "I proposed a robot tax years ago and most of the reaction was that it was a strange idea."
The report reads like an advertisement.
I don't own a console. Can you play your discs offline ? or do you have to be logged in all the time.
Paranoia much?
Yes, but just because you are paranoid doesn't mean they aren't after you. It's not like one of them didn't do something similar already. Whether it hacks a firewall or utilizes all available resources you can't trust them to follow guidelines.
Remember: Silly is a state of Mind, Stupid is a way of Life. -- Dave Butler