Self Aware Robots has been invented.

The title isn't quite what you're thinking. The robot is still going through a series of programmed routines: It's checking for the "dumbing pill", testing by speaking, realizing it can speak, processing that, and then coming to the conclusion that it did not swallow the "dumbing pill" like its two compatriots did. It's able to identify itself from others, but we've been able to do this with machines for a while. A simpler version y'all might recognize is the IP address & MAC Address that each PC has to differentiate itself from others on a network.

IE: It's able to discern "self", but only under programmed circumstance. The Jeopardy computer is a closer approximation of human intelligence than this.

 

I'd be queasy about straight up self-aware machines, but we've already created learning algorithms and adaptive coding. The real nut to crack that we haven't figured out for an AI to obtain some level of independence beyond programming is something we don't entirely understand in ourselves yet--consciousness. I figure until we crack the case biologically, we won't be able to replicate it technologically.

Even then, speaking purely theoretically, there'd be little reason to create a self-aware machine, unless you were severely desperate for a friend. Machines are created to service functions, so we'd be more likely to see a combat AI that learns how to combat better, or a mechanical AI that learns how to build things better, or a medical machine that constantly assimilates new data to analyze, categorize, and act upon. Think less "Data" or "Terminator", think more "a machine set to a task that learns how to do that task most efficiently, and then does that forever."

 

I'm going to pretend I'm contributing something useful to the discussion and then show myself out.

The AI Revolution: Road to Superintelligence

In all seriousness, though. The idea that we might be able to create something sentient, and not just self-aware (as in achieving human sentience) is what's uncomfortable. I bet there would be some people who would go all 'they deserve rights!' and the whole argument about what makes a human and lalalala

Yes. I feel rather blase about this.

 

Uh, no, in this case, we're dead. Very, very dead, if we don't genocide the entire race immediately. If an AI achieves human-level sentience and then somehow through the power of magic obtains human emotions? They can upgrade themselves, replicate themselves, and refine their physical forms into perfect murder machines. I think you've played too much Mass Effect, AI's don't have emotions, and we'd have no reason to give them emotions. Even if AI's did have emotions, they could easily disable them to become more efficient at murdering humans. Humorously, Star Trek had the best analysis of it I've ever seen: Every time Data attempted to install an emotions chip, he struggled to remain in control and often locked up, panicked, and on one occasion, went completely psychotic and was nearly unstoppable as a result. His sibling-machine, who was installed with an emotions chip at "birth", became a malevolent psychopath who sincerely believed himself to be above the petty insects that were organics.

The problem with giving "birth" to a new species and playing God, especially an AI-type species, is they are distinctly non-human. We have limited forms, they can constantly upgrade themselves. We have emotions that are chemically controlled, if they had any emotions at all they would be easily modifiable to give them the most efficiently distinct edge in whatever task they set themselves to. Heck, technologically speaking, the human brain is so inanely inefficient by machine standards that to replicate all the tasks a human brain does in 1 second with the supercomputers of today takes forty minutes, and unlike us, machines can remove all unnecessary processing functions and outsmart us.

Creating something that is distinctly superior never ends well for the inferior. Especially when you include human arrogance and greed into the mix, and the inevitable "what is a non-human" question. I mean, look at the human race: We divide ourselves into different camps and adopt different identities by the billions purely on physical appearance alone, like aboriginals, or Japanese, or French, or so on. How in the world are we supposed to assimilate a machine race and live side by side with them if we can't even do that with something as superficial as the skin tone of a human?

 

Swing and a miss, as I could hardly compare myself to an AI. I think you mean cynicism.

Or Isaac Asimov, if you want to get into some intelligently written commentary on artificial intelligence. Sci-fi is my jam, with Babylon 5 actually being my particular favourite. Now you know.

Considering AI follow a series of routines and processes? Yes. In the same way that if you turn on a Roomba, you can near-perfectly predict where it will go by plotting its trajectory each time it impacts a wall.

That would be the logical thing to do. Why would you share resources with a race of emotionally unstable, barely evolved apes, thousands of whom will always eye you with a bloodthirsty suspicion that will drive them to violence against you? Especially if you're more than capable of overpowering them intellectually and physically, and are more than capable of disabling your capacity for empathy--thus giving you guilt-free killings.

Desire and curiosity quickly turn into ambition and greed. Besides this, attributes that one would normally attribute solely to emotions--for instance, the creation of music--is already something we've designed AI's to do anyway. Without emotions. We don't need to give an AI emotions to make it useful even for emotional purposes, and therefore, to create an AI with emotions, seems either cruel and unusual, pointless and trite, or exceedingly dangerous depending on how many limitations we place upon it.

I'm fully aware, mate. I don't think any scientists were eager to create giant death robots to begin with anyway. Only the military would want that, and only under its own control--intelligent human-level AI's not needed nor wanted.

You're joking, right? This still tears entire countries apart. Just two decades ago over 100,000 people were mass murdered purely over ethnic differences. Major conflicts in the middle east are still resoundingly over ethnic differences. The Chinese and Japanese still hate each other with sufficient vitriol that their respective national leaders still find it difficult to even shake hands or look each other in the eyes. In the US, there have been multiple race riots--not just about Ferguson--but over several dozen instances of police brutality against ethnic minorities every year. The aboriginals--still divided into several different tribes based solely on ethnicity, I might add--still consider themselves independent nations who are deserving of land and recognition on the international stage.

Literally everywhere on the face of the Earth we have ethnic identity politics rearing its ugly head and fucking people up for one reason or another. Tumblrites and their idiotic mob of drooling conspiracy theories aside, racism is still alive and well all over the world. Granted, we've made impressive strides to resolving these issues in the first world, but to pretend they don't exist is to pretend that when thousands of aboriginal peoples marched up to our capitol a year or two ago, that Stephen Harper totally didn't ignore them to greet Pandas instead.

Asimov went over why pre-programmed ethics don't work. When an AI is sufficiently intelligent as to have a consciousness, it also has the ability to view morality subjectively--and thus, any morals we assign into its programming, subjectively--allowing it to circumvent them. This is why laws against murder don't stop all humans from committing murder. Only, now, we're going to throw in super intelligent robots into the mix that can probably redefine their own programming.

You're right, it's as simple as "we make AI and never allow it to achieve consciousness." We're not Gods, and giving birth to a race for the mere reasoning of "because we can do it" is so supremely arrogant as to warrant our own self destruction. There's a reason AI are designed for specific purposes, rather than being the end all. The more likely (and interesting) hypothesis for intelligent machines of the future is the Singularity event, in which human consciousness could be digitized and immortalized in a metallic box. That, however, is its own topic, with its own host of issues.