HOME | DD

Description

Part 1: www.deviantart.com/cactuschowd…
[Part 2]
Part 3: www.deviantart.com/cactuschowd…
Part 4: www.deviantart.com/cactuschowd…

Part 5: www.deviantart.com/cactuschowd…

Part 6: www.deviantart.com/cactuschowd…
Part 7: www.deviantart.com/cactuschowd…
Part 8: www.deviantart.com/cactuschowd…
Part 9: www.deviantart.com/cactuschowd…
(Some kind of story with this premise is growing in my mind like a stalactite)

I remember reading somewhere that training AI to do tasks can be a challenge because sometimes the AI will find a way to optimize exactly what you technically told it optimize, but in a way that sidesteps everything you wanted it to actually do in the process. If people trained to think like computers replaced computers, I think they’d become very good at finding ways to do less work, especially if they were trained to disregard all ethics.

Image description:
_____________________________________

A two-panel comic featuring Piter de Vries and Baron Vladimir Harkonnen from Dune.

Panel 1:

Baron (with nonchalant entitlement, holding the ‘world’s best boss’ mug from The Office and standing in front of a background that looks like Michael’s office from The Office):

You see, my mentat’s training was built around rejecting ethics as a consideration. How long will it take him to complete a task? Well... half as long as you might think it should, right? Those mentats are very smart. And mine doesn’t have to take the time to make sure he’s being moral. That’s efficiency. That’s basic efficiency.

Is he going to hurt somebody? Sure. Is that my problem? Ha ha ha. No. No, it is not.

Panel 2:

Piter (derisively and annoyed, about to cut off a lock of his hair with a knife, standing in front of a background that looks like the wall of the conference room from The Office):

My boss likely sent you here for some kind of garish, violent plan that I take obvious delight in, and I must disappoint you. First, all of my delight is gone, thanks to this awful time crunch. Second, the simplest solution does not have to be spectacle. Sometimes it is merely sense.

Developing artificial intelligence is supposed to be one of the greatest taboos of this world, but I cannot find a single sensible downside. If I can build a thinking machine, I estimate that I will be able to secretly offload at least 30% of my daily worklad onto it, unknown to my boss, for this week and every week after it.

I have known for some time of a private collector who possesses a thinking machine schematic, and I would trust myself to steal it if I tried. However, there are enough unknowns that I can't take the risk. So I am going to move around some numbers in the Baron's expenses and find the funds to order a few clones of myself, who can attempt the heist while I get started on the week's drudgery as a failsafe.

Description

Part 1: www.deviantart.com/cactuschowd…
[Part 2]
Part 3: www.deviantart.com/cactuschowd…
Part 4: www.deviantart.com/cactuschowd…

Part 5: www.deviantart.com/cactuschowd…

Part 6: www.deviantart.com/cactuschowd…
Part 7: www.deviantart.com/cactuschowd…
Part 8: www.deviantart.com/cactuschowd…
Part 9: www.deviantart.com/cactuschowd…
(Some kind of story with this premise is growing in my mind like a stalactite)

I remember reading somewhere that training AI to do tasks can be a challenge because sometimes the AI will find a way to optimize exactly what you technically told it optimize, but in a way that sidesteps everything you wanted it to actually do in the process. If people trained to think like computers replaced computers, I think they’d become very good at finding ways to do less work, especially if they were trained to disregard all ethics.

Image description:
_____________________________________

A two-panel comic featuring Piter de Vries and Baron Vladimir Harkonnen from Dune.

Panel 1:

Baron (with nonchalant entitlement, holding the ‘world’s best boss’ mug from The Office and standing in front of a background that looks like Michael’s office from The Office):

You see, my mentat’s training was built around rejecting ethics as a consideration. How long will it take him to complete a task? Well... half as long as you might think it should, right? Those mentats are very smart. And mine doesn’t have to take the time to make sure he’s being moral. That’s efficiency. That’s basic efficiency.

Is he going to hurt somebody? Sure. Is that my problem? Ha ha ha. No. No, it is not.

Panel 2:

Piter (derisively and annoyed, about to cut off a lock of his hair with a knife, standing in front of a background that looks like the wall of the conference room from The Office):

My boss likely sent you here for some kind of garish, violent plan that I take obvious delight in, and I must disappoint you. First, all of my delight is gone, thanks to this awful time crunch. Second, the simplest solution does not have to be spectacle. Sometimes it is merely sense.

Developing artificial intelligence is supposed to be one of the greatest taboos of this world, but I cannot find a single sensible downside. If I can build a thinking machine, I estimate that I will be able to secretly offload at least 30% of my daily worklad onto it, unknown to my boss, for this week and every week after it.

I have known for some time of a private collector who possesses a thinking machine schematic, and I would trust myself to steal it if I tried. However, there are enough unknowns that I can't take the risk. So I am going to move around some numbers in the Baron's expenses and find the funds to order a few clones of myself, who can attempt the heist while I get started on the week's drudgery as a failsafe.