> The urgency to build with the new A.I. was crystallized in an internal email sent last month by Sam Schillace, a technology executive at Microsoft. He wrote in the email, which was viewed by The New York Times, that it was an “absolutely fatal error in this moment to worry about things that can be fixed later.”
I wonder what the list of "things that can be fixed later" is and when would "latter" be.
Mentally ill people will be triggered to do things by it. As an older guy, this reminds when LSD was introduced. We learned to manage that after we had a bunch of acid causalities and Charles Manson cults, we'll learn to manage this
A key difference here is that the effects of "smart" AI on networks binds tightly to, and in a way that is hard to detect integrates with, online channels of communication. Online channels of communication are quite the topic right now, considering phones use as social endpoint.
IMO- that ill-advised publicity and "party" campaign by Timothy Leary caused a lot of damage to a lot of people.
Well, I hooked it into the power grid voltage monitoring system because there was a legacy meter that didn't put out the right signal, but the analog meter showed correct. The image recognition part of the model could tell when the sensor was about to go high, and then it issued an API command to the generator to slow down...
Using LLL-based image classification to read an anlog meter and then using that as the source of truth. OK, this just utter nonsense.
On second thought it sounds like the type of thing that very ill-advised people could come up with.
Um, Sir!... Nuclear war or a supervolcano or a superbug or climate doom are killing the entire planet because our compute just isn't powerful enough yet to run the "fix these problems" script faster than 1,000 years!!!
Any technology that is that powerful to solve those problems with a fast and correct solution can also kill you a billion times over a trillion times faster...
Not solving those problems with a fast and correct solution could kill everyone, not to mention that nuclear technology misapplied is already at risk of ending everything a billion or a trillion times faster than the other risks. Including risks from AI.
if microsoft and (particularly) google are "speedy" in this, then what the open (open source, etc.) projects (like stable diffusion) are? the level of 'whatever, we don't really care, whatever happens, happens, just ship it' that stable diffusion and the explosion of sd-based projects/apps/services (and other similar open source projects that are easy to utilize/fork/replicate/etc.) that will just 'do stuff' without much care or concern - how could that be described? 'scorching, incinerating speed'?
at this point, why even really 'care' or 'be concerned about' these ai projects, when existing projects just move ahead and don't give a damn about things? creators of such projects certainly don't, and people...kinda no longer do (if only due to helplessness, since projects will just move ahead (along with people's data they scraped, which people still can't do much about), without much concern).
if anything, there might be a vague worry about the eventual openly available chatgpt-like thing that would really be at the level and power of it. but I don't really think people are gonna be that 'concerned'. certainly not the creators of them, who just move along and make more and more of their AI apps and services, without stuff like 'caution' on their roadmap. things continue to get implemented in their lackadaisical non-approach to actually implementing systems that'd let people manage ai in better ways (or, like, at all), and have some control and accountability over it. (such as, 'being able to pluck out data and opt-out from a system/model/whatever on request' - but who cares about that, right. so, currently there's no substantial, workable development of such system, not from the first parties (the actual companies that create and train ai systems like that), and "third party" systems remain somewhat of an non-functioning smokescreen that those companies get to point at like 'we did something'.)
so, whatever. i guess it doesn't really matter, and we're just gonna move along and find out later when something crashes and burns, cause that's the general approach here. and it might not even be all that wrong.
Another perspective: If a company that invests in this would ever even remotely claim they have a sustainability and decarbonization plan they should be laughed out of the room from now on.
17 comments
[ 5.0 ms ] story [ 51.2 ms ] threadI wonder what the list of "things that can be fixed later" is and when would "latter" be.
IMO- that ill-advised publicity and "party" campaign by Timothy Leary caused a lot of damage to a lot of people.
1/ It can output offensive things.
2/ It can output incorrect information that someone takes at face value, like "drink bleach to fix your headache".
3/ It can leak secret information, if it was trained on secret information.
None of these sound like a really big deal TBH.
Well, I hooked it into the power grid voltage monitoring system because there was a legacy meter that didn't put out the right signal, but the analog meter showed correct. The image recognition part of the model could tell when the sensor was about to go high, and then it issued an API command to the generator to slow down...
So, um, sorry about the blackout.
at this point, why even really 'care' or 'be concerned about' these ai projects, when existing projects just move ahead and don't give a damn about things? creators of such projects certainly don't, and people...kinda no longer do (if only due to helplessness, since projects will just move ahead (along with people's data they scraped, which people still can't do much about), without much concern).
if anything, there might be a vague worry about the eventual openly available chatgpt-like thing that would really be at the level and power of it. but I don't really think people are gonna be that 'concerned'. certainly not the creators of them, who just move along and make more and more of their AI apps and services, without stuff like 'caution' on their roadmap. things continue to get implemented in their lackadaisical non-approach to actually implementing systems that'd let people manage ai in better ways (or, like, at all), and have some control and accountability over it. (such as, 'being able to pluck out data and opt-out from a system/model/whatever on request' - but who cares about that, right. so, currently there's no substantial, workable development of such system, not from the first parties (the actual companies that create and train ai systems like that), and "third party" systems remain somewhat of an non-functioning smokescreen that those companies get to point at like 'we did something'.)
so, whatever. i guess it doesn't really matter, and we're just gonna move along and find out later when something crashes and burns, cause that's the general approach here. and it might not even be all that wrong.