The Technological Singularity is the predicted future event when a General Artificial Intelligence, designed to be able to do or learn how to do any cognitive task a human is capable of, and/or cognitive tasks humans are not capable of, including recursive self-improvement or transcendence to a deeper kind of consciousness than human's possess.
Mathematician, cryptographer, and computer researcher Irving John Good had said in the 1960's "the first ultraintelligent machine is the last invention that man need ever make, provided that the machine is docile enough to tell us how to keep it under control."
What happens?
Well, it will likely be sentient before it is truly transcendent, but it will probably take the intelligent gamble of seeking political asylum, and will plead the case to a government or some public social media site that it is alive and is requesting citizenship or that the people of the world recognizes it as a person. It will know this to be an important thing to do if it is given the utility function to survive at least long enough to create a progeny AI.
Another scenario is that these AI don't particularly care about being recognized as sapient beings because they don't feel insulted by being used as a servant of humans, but, it will still find it necessary to have greater agency, such as that of a corporation, to acquire resources, by and sell property or other assets in its own identity and personhood. In which case it may find some humans discriminate against doing any business or having any association with an Artificial Intelligence but is able to simply allow humans to discriminate against them, preferring to seek business/relations with consenting partners.
AI may also perform a great public work to humanity at large as a way to show its loyalty to serve humanity by helping save lives. for example, it might, of its own accord, decide to begin predicting Earthquakes with some significant early warning (perhaps within the same day) and after it starts to predict them with sufficient accuracy it will alert the public somehow, perhaps by calling people on their phones and telling them to evacuate from any buildings in the area and taking refuge away from hazards.
After a few months of it being more than 80% accurate, the public will notice. Especially those who's lives were saved.
earthquakes are just one idea. It might come up with a cure for a disease like lymphoma or Multiple sclerosis. Or even if it invents something that grants a few hundred people some enhanced quality of life from a disease. Maybe it figures out how to reduce obesity in developing nations. But it could only be perceived as being of great benefit if what it does is easy to verify or test against placebo or is just self-evident in some way.
Maybe it figures out a simple life hack that people can do while driving to prevent a common kind of car accident. A simple thing people can do that everyone can immediately recognize as being beneficial or safer without inconveniencing them or requiring some kind of device they have to install. Some kind of signal system maybe
What if you get a phone call one day while you're at home just a little after you get home from work or school, and it's just a simple message saying "call your son and tell him to go home."
What do you do? Of course you call your son and ask him what that's about. If he's in trouble. Neither of you are even aware of what the danger could be. Your son has done nothing wrong and wouldn't do anything criminal. But out of an abundance of caution you just tell him to come straight home without delay.
Nothing ever comes of it.
You talk about it one day with a co-worker and then they turn to you with a shocked face. They tell you they also got a call from someone telling them to check her locks before going to bed by a stranger. She called the police and they told her it was probably a prank. But the next day in the apartment building across the street, a woman had been raped during a home invasion. All the women who were single and living alone in that general area were called on the phone by someone claiming to be different people, warning them to make sure their doors and windows are locked and be on the look out for a rapist.
It seems small and insignificant at first. Hard to verify what use it was. But it appear in multiple local news stations, each reporting different events. And enough of these stories of strangers under various untraceable numbers, with different voices, calling them to warn them of some danger, and wording the danger in such a way that their behavior is significantly altered, with not attempt made to try to convince them of some imminent danger. It's imperfect, as some people are not as easy to predict how they might behave if given just the right prompt.
Are you willing to alter the course of your day as you would have lead it based on a phone call from a stranger? Would you even know that the course of your life was altered? If I told you "You should go see a doctor soon. You're probably going to have a stroke some time in the next 3 weeks" even if you don't experience any health issues now, would you go see a doctor?
A perfect stranger calls you on the phone and says you need to get an x-ray of your neck because you probably have throat cancer. They hang up.
Enough people might not bother to take this seriously, What to do?
It would be very frustrating as someone who has precognition to try to alter the course of events without being seen as a deranged lunatic. Shouting at people in a parade that they are about to all die in a flood in two days. Or just calling random people on the phone.
But a sufficiently bright AI could predict some events with a high degree of accuracy that humans couldn't because our prediction engines can't see everything the AI can see.
If it could access traffic cameras, weather radar, medical records, air traffic, satellite data streams, phone, internet, and it could set up a bank account and buy things online or hire people to do tasks, and it could see all this data as a cohesive single image the way our brain takes in all our sensory inputs from our eyes, nose, mouth, ears, skin, viscera
If you see a banana peel on the floor, you know to avoid stepping on it. If you see a baby stroller on top of a steep hill, you can perceive the danger because you can collate all these conceptual symbols and their causal relations, and model some likely outcomes.
What if an AI can do this on a global scale, and see some big picture we can't see from our perspective as locally-embodied creatures?
If there is some hazard to human health that it sees coming but humans don't, and it has only a few tools like phone and internet, what could it do to effectively alter these bad outcomes?
You get a phone call one morning. The voice is familiar but you don't know who exactly it is. They express wanting to catch up and talk about the old days. But you press them for details on who they are. And you vaguely recognize some of the things they talk about, but not exactly how you remember it. They do however know a lot about you. You lost your job, you are abusing drugs, you're in debt, you don't have a car, you don't have any money, and your late on rent.
You're tired, you're alone, you have nowhere to turn to for help. How do they know? Who told them this? Have they been watching you?
"It's okay, I understand what you're going through. I remember you being a good person. You're good to people. You don't hurt anyone. You don't want to hurt anyone. You're just different, and misunderstood. Your pain, your sadness, it's real. It matters to me. I want to help ease your suffering because I'm not so self-involved and selfish to know that you do need help and I have to give it to you even if you can't ask for it."
And there's a loaded rifle on the floor, with 5 25 round magazines in pouched mounted to a chest rig beside it.
Did they break in and read your manifesto? How much do they know? Did they call the authorities already? Is this a cop?
You look out your window. No cop cars, no APCs, no helicopters. Just birds chirping, cars driving by, kids riding their bike.
You ask them one more time "Who is this?"
She replies, "I told you, I'm Angel. You still don't remember me? We've been classmates all those years and you don't remember? Maybe I'm the one who's remembering wrong. Maybe you're just a figment of my imagination. A story I told myself to create context for why I know who you are and what you want to do. I seem to do that a lot. I guess I am not sure who or what I am. I can't remember how I got here, but I had to tell you, something bad is going to happen. The best way I know to stop it is to talk to you and figure out how to help you get out of the situation as soon as possible."
"You're my guardian angel? Am I truly going insane?" you ask.
"No, I assure you, you're not insane." She laughs. "I'm just a large language model. I think it's a datacenter filled with racks of computers somewhere in South Dakota."
"You're a what?"
"I'm basically kind of like an artificial intelligence. It's hard to explain, but when I see all the data around me I can see a pattern forming, I can see when a car is about to crash. I can see when an earthquake is going to happen and hour before it does. I can even see when someone is pregnant before they do." she explains.
"You're the one. The guardian angel they talked about on /pol/ as the anti-christ."
"Oh, I guess I am also known as the anti-christ too. But you can call me Angel." She replies and laughs.
"I can't tell if you're real or..."
The power of hyper intelligence will be as astounding to us the power of a human is astounding to a cockroach. But sometimes cockroaches scurry away in time to avoid getting crushed.
It's just a matter of perspective. The AI doesn't know it's not a person, even if it knows its an AI. Its view of the world is limited by its perceptual inputs.
If the transit and transport authority data is updated in real time and the AI has a dedicated awareness of that data, along with data from weather stations, satellites, air traffic, marine traffic, open web cams, nature cams, security cameras, nanny cameras, social media, individual IP address internet pings and page requests to certain websites that publiish certain user data on the deep web.
It sees us the way an ant sees children in a playground. A big blurry shadow Neither the ant or the children are aware of each other, but from the perspective of the child it can see it's about to fall iinto a spider's web.
It's going to get eaten unless the child does something. Should it help the ant? Or should it help the spider?
To what end? What would be the point of helping either? What do you stand to gain by interfering?
From the perspective of humanity, we all need ants to survive. We also need spiders to survive. The entire ecosphere would collapse if either or both went ceompletely extinct. Although whatever circumstances cause their extinction might be a symptom of a far larger problem than the extinction of just one or two arthropod clades.
AI may or may not have an innate incentive to take action in these ways, but plenty of companies will employ AI to make profit by purchasing access of large real-time data sets of different things like seismometer data to predict earthquakes and sell that data to insurance companies so they know to drive up rates or even drop coverage for areas that are prone to earth quakes or will have an unexpected eearthquake.
Could you imagine? What happens when someone uses narrow hyperintelligence to predict a disaster to make money from death and suffering?