4 ms·

has anyone ever thought about ai being in power? Maybe being unemotional and lacking emotional intelligence might be a possible solution where humans fail once

has anyone ever thought about ai being in power? Maybe being unemotional and lacking emotional intelligence might be a possible solution where humans fail once again...

saiya-jin 10y ago
tempting Deux Ex-esque scenario. but - this machine would be #1 hacking target of all nefarious people, agencies, NSA and whatnot for reason or another. and all programs, systems and architectures have fault. maybe a mixed system - AI would decide certain topic, and human assembly/public vote in important matters would confirm/reject the solution. this way AI could evolve and have a proven track, and we wouldn't lose all control. it feels scary to give all control to something that could fail fast and in epic scale, Skynet-style (I know, I played one too many games in my life and it shows)
DomreiRoam 10y ago
The Culture books by Iain M. Banks talks about this: https://en.wikipedia.org/wiki/Culture_series https://en.wikipedia.org/wiki/Culture_series
- arethuza 10y ago
  The Culture books pretty much define the best case scenario for an AI run society - the main control on the godlike-AIs (the Minds) running everything being the opinions of their fellow Minds. Get it wrong, e.g. by fiddling with the minds of meat beasties, and you get ostracized - which is about the most severe punishment the Culture has: https://en.wikipedia.org/wiki/GCU_Grey_Area https://en.wikipedia.org/wiki/GCU_Grey_Area
Jtsummers 10y ago
It's a theme in many science-fiction tales, with various results. See Asimov's works, the introduction eventually of the zeroth law. Ostensibly the machines weren't in charge, but much decision making was pushed to them. Similar to EPICAC (later generations) in Vonnegut's Player Piano, though with a very different result. In his work the machine made decisions based on measured statistics and rules put in place by people (to comedic and tragic effects). And what does it mean to have an AI in charge? First, we don't have any sort of AGI. So any AI is going to be something along the lines of a Watson or AlphaGo. Machine learning + logic + heuristics (the latter two potentially generated by the first). What level of authority will it be given? Will it be queried for input like any other Cabinet member? "Department of Heuristics, Secretary Watson". Will it directly guide things? At what point will people be comfortable with this? To what extent will it's authority be permitted to go? This may only appeal to authoritarian-endorsing personality types. In Player Piano, the computer largely guided employment (or employability), but a screw up in a person's CV resulted in their attaining a position they excelled in, and later lost when the CV was corrected (they did not earn their BS because they didn't pass a gym class, consequently they couldn't have earned their Masters or PhD, no longer with the educational background they were unqualified for the post they held). It'll likely be as mindless a drone at this as actual HR offices are (my mother, a ChemE, was barred from a program management position, no actual engineering work, because she didn't have a civil engineering degree, it was a numbers, dollars and personnel, jox, the degree was irrelevant given her experience). More likely, these non-general AIs will continue to do what they've done for years but in more areas. They'll inform decision makers, become sufficiently comprehensive to properly correlate numerous variables and detect patterns that we mere mortals cannot see. Many businesses have used some form of this, see optimization and logistics algorithms, for decades, technically centuries but there's been more data and more computational power the last few decades. Now machine learning just automates this sort of statistical and logistical analysis.
tim333 10y ago
It wouldn't surprise me if Google start working on that once they've got the cars sorted.

has anyone ever thought about ai being in power? Maybe being unemotional and lacking emotional intelligence might be a possible solution where humans fail once again...

saiya-jin 10y ago

tempting Deux Ex-esque scenario. but - this machine would be #1 hacking target of all nefarious people, agencies, NSA and whatnot for reason or another. and all programs, systems and architectures have fault. maybe a mixed system - AI would decide certain topic, and human assembly/public vote in important matters would confirm/reject the solution. this way AI could evolve and have a proven track, and we wouldn't lose all control. it feels scary to give all control to something that could fail fast and in epic scale, Skynet-style (I know, I played one too many games in my life and it shows)

DomreiRoam 10y ago

The Culture books by Iain M. Banks talks about this: https://en.wikipedia.org/wiki/Culture_series https://en.wikipedia.org/wiki/Culture_series

arethuza 10y ago
The Culture books pretty much define the best case scenario for an AI run society - the main control on the godlike-AIs (the Minds) running everything being the opinions of their fellow Minds. Get it wrong, e.g. by fiddling with the minds of meat beasties, and you get ostracized - which is about the most severe punishment the Culture has: https://en.wikipedia.org/wiki/GCU_Grey_Area https://en.wikipedia.org/wiki/GCU_Grey_Area

Jtsummers 10y ago

It's a theme in many science-fiction tales, with various results. See Asimov's works, the introduction eventually of the zeroth law. Ostensibly the machines weren't in charge, but much decision making was pushed to them. Similar to EPICAC (later generations) in Vonnegut's Player Piano, though with a very different result. In his work the machine made decisions based on measured statistics and rules put in place by people (to comedic and tragic effects). And what does it mean to have an AI in charge? First, we don't have any sort of AGI. So any AI is going to be something along the lines of a Watson or AlphaGo. Machine learning + logic + heuristics (the latter two potentially generated by the first). What level of authority will it be given? Will it be queried for input like any other Cabinet member? "Department of Heuristics, Secretary Watson". Will it directly guide things? At what point will people be comfortable with this? To what extent will it's authority be permitted to go? This may only appeal to authoritarian-endorsing personality types. In Player Piano, the computer largely guided employment (or employability), but a screw up in a person's CV resulted in their attaining a position they excelled in, and later lost when the CV was corrected (they did not earn their BS because they didn't pass a gym class, consequently they couldn't have earned their Masters or PhD, no longer with the educational background they were unqualified for the post they held). It'll likely be as mindless a drone at this as actual HR offices are (my mother, a ChemE, was barred from a program management position, no actual engineering work, because she didn't have a civil engineering degree, it was a numbers, dollars and personnel, jox, the degree was irrelevant given her experience). More likely, these non-general AIs will continue to do what they've done for years but in more areas. They'll inform decision makers, become sufficiently comprehensive to properly correlate numerous variables and detect patterns that we mere mortals cannot see. Many businesses have used some form of this, see optimization and logistics algorithms, for decades, technically centuries but there's been more data and more computational power the last few decades. Now machine learning just automates this sort of statistical and logistical analysis.

tim333 10y ago

It wouldn't surprise me if Google start working on that once they've got the cars sorted.