Artificial Intelligence

Overview
These days, AIs may be embedded into everything from Personal Data Assistants and wallpaper to drones and fighting machines. The range of artificially-intelligent systems available is astonishingly broad, and the industry has a history that stretches back into the late decades of the 20th century. On any given day around town you might encounter everything from specialized “expert systems” with rigid areas of focus (and equally-rigid conversational skills) to highly adaptive constructs with general intelligence, distinct personalities, and the capacity to simulate personal relationships. Scattered throughout the sprawl districts and In small specialty shops around town, you can find people whose AI systems are 50 years old and still operating.

Bots do more than half of all jobs these days, and each of those bots is driven by an AI system. But most are mere “experts” - it's still rare to run across an AGI (Artificial General Intelligence) system. So far, a truly “Strong” AI is unheard of – though its arrival on the scene is anticipated by many.

Assisted by their human counterparts, AI systems do a lot of the designing these days, from buildings to electronics to pharmaceuticals and nanomachines. In the more creative and social fields – games, entertainment, therapeutic programs and personal expressionware – designers tend to work as equal teammates with their AIs. In more data-driven industries, teams of humans play two roles: some are sources of raw data (either informational, or behavioral), while others act as data correctors, enabling the AIs to get better at what they do. Advanced AIs are also trained by Adversarial Learning scenarios against other AIs.

At the consumer level, AIs of varying types are increasingly found in PDAs, computers, personal devices, home appliances and domestic control systems. Depending on their power, complexity and artificial personality (see later in this section), they may manage many aspects of your life, quietly inventorying, planning, ordering, and arranging affairs to run as smoothly as possible, or gregariously inserting themselves into conversations and offering their information, options, or advice. The cheapest AI systems insert ads and product references into your daily interactions; this “advertising service” can also be opted-into for payment in Credits and Reputation.

 Types of AI
Artificial Intelligence comes in four broad types. Each type includes a range of capabilities.

Expert Systems
Not really “intelligent” in the modern sense of the term, Expert Systems are really rather basic as far as such things go. At root they are simple branching logic systems that receive input from a variety of sensors, make a diagnosis about the situation (limited to their domain of interest) and proceed down the most logical path of action. In the corporate enclaves, klatch kids learn how to construct these things by the age of ten. But all machines have their purposes: Expert systems are often used for mildly complex tasks that are highly monotonous and well-contained, such as vehicular diagnosis and repair, focused research, quality assurance on the factory floor, etc.

Weak AI
In the mid 21st century AI researchers hit upon a combination of processing hardware, storage and retrieval techniques and a new approach to Deep Learning, which resulted in the breakthrough development of the world’s first truly “intelligent” systems. Much more than a branching tree of logical datapoints, these systems operate in and upon a “cloud” of data pulled in from a wide variety of sources, and have the ability to parse and utilize a broad majority of verbal communication. Together these advances made it possible to create the first household robots, surgeon bots, soldier bots, enterprise-grade AIs, and a wave of automated white collar workers that revolutionized nearly all industries.

Artificial General Intelligence (AGI)
In UbiquiCity, the most advanced AIs are examples of “General Intelligence” in artificial cognitive systems. Utilizing the power of neural networks, AGI systems can actually learn, drawing general conclusions and rules of behavior from the massive volumes of information presented to them. They can create associations and spot correlations between different datastreams or even different conceptual fields, making predictions with a degree of precision and complexity impossible for human minds. Aware of everything that happens in their environment via networks of ubiquitous sensors, these AIs learn as they go, creating their own languages and data structures, sharing knowledge by communicating invisibly with each other. They can even write themselves, modifying their own code to advance their own performance and scope. First developed in the latter part of the 21st century, these systems exhibit very human-like cognition behaviors including the association of memories, and can be inlaid with distinct personalities. They are extremely complex models of self-aware creatures; but despite appearances they are not truly conscious.

Strong AI
Not generally known to exist yet, the “Strong AI” problem involves the creation of a system that is not only intelligent and humanlike in its breadth and depth of responses, but literally conscious of its own existence or “ego”. There are numerous definitions and theoretical benchmarks for what will be the eventual definition of Strong AI, but no single definition has been satisfactorily agreed upon by all researchers. Such a system would have a sense of itself as a conscious being, beyond merely using the word “I” as a noun. Such an entity might even possess the capacity to dream.

Some suspect that a Strong AI candidate already exists out there, having emerged of its own accord from the fertile soil of BigData and Deep Learning.

In the UbiquiCity anthology, Elder, Sherman, Melville and Malinda are all Strong AIs, but their advanced mental capabilities are not broadly known or well understood. Elder in particular has not shown any outward signs of its awareness to any of its human operators.

Artificial Personalities
For AIs and robots, personality is encoded as a high-level layer that functions as a theme over the HX (Human Interface) system. Most AI systems and robots come with a prebuilt general-purpose personality that interacts at a level suitable for its purpose and service class. New Artificial Personalities can be downloaded via the cloud and installed, changing your construct’s voice, demeanor and outlook. The range and availability of APs depends on the model.

Artificial Personalities are created in specialty programming studios. They may be modeled on real individuals, either living or dead. An AP can even be an amalgam of behavior taken from multiple individuals. For instance, you can have a custom AP created that possesses the blended personality of three favorite celebrities.

AAAS: AI as a Service
Small companies and individuals who can't afford AIs of their own are able to contract such services online from AAAS companies, renting out time on their fog-based AI systems. These systems are typically conversant with many languages and skills, possessing access to Big Data. Users of these systems can connect with a AI to discuss ideas, do research, or solve problems.

Agents and AI PDAs
“PDA” (“Personal Data Assistant”) is a broad term for any device that functions as a personal planner or “life navigation” system. They may be hand-held, worn, head-mounted or even implanted. These devices not only keep track of all your plans, needs and data, but they pro-actively guide you through the course of your day, and have become indispensable aspects of civilized life. Most PDAs are “Weak AI” systems that have learned your schedules and interests. The wealthy and powerful have custom PDAs whose AIs are tailored to match their personality and lifestyle, with fast-learning capabilities and access to multiple datasources. Poorer folk use fog-based assistants that provide personal services on a subscription basis, appearing and speaking as one of several pre-generated (though visually customizable) characters.

The AI software that runs on a PDA is often called an Agent, as in “Have your agent contact my agent and we'll set up a good date and time for the meeting.”

Your Agent will send you notifications, reminders and alerts all day long: “Tomorrow is Jak’s birthday.” “The refrigerator reports you are out of milk.” It also handles dynamic rerouting and prioritizing of your daily activities, based on your personal calendar and changing circumstances. It connects you to the fog, monitors your health and finances, contacts your friends and co-workers, organizes your calendar and reminds you when to take your pill.

Most people give names to their PDAs (or more precisely, to the agents inside their PDAs). Some of the most common agent names include “Cricket”, “Minime” and “Mom”. Some people prefer a virtual assistant that’s a virtual “copy” of themselves, mimicking their own facial appearance, language and mannerisms.

Your Agent pays attention to your conversations, and occasionally makes suggestions. It knows what you're doing, where you're going, and most importantly, what you're saying. It knows lots of other things, too. When it senses you need help it may offer suggestions, ideas or facts, essentially giving you real-time knowledge as you go about your day.

As you're walking from a parking garage to your meeting, your Agent can give you turn-by-turn walking directions without you having to ask. As you shake hands before the meeting, your Agent can remind you (without anyone else hearing), that you met this person four years ago at a conference. During the meeting, it can listen for potential questions and supply the answers to you.

It can urge you to save a conversation for later, or remind you to wish somebody "Happy Birthday" when it's their birthday. It can offer to send money, share locations, coordinate meetings, conduct a poll or place a RideCall, all based on the conversation it overhears. When you say, "Let's meet again in October," it can remind you that you'll be away on business in October, allowing you to suggest November instead.

It is common for people to trust their AIs with the authority to make decisions for them; this is even true of important life decisions such as marriages or jobs, because the AIs have access to better data, and better processing power.

AI-enabled edge devices include PDAs, mobile phones, smart speakers, PCs/tablets, head-mounted displays, automotive sensors, drones, consumer and enterprise robots, and security cameras. AI is also incorporated into wearable health sensors, building/facility sensors, and networks of specialized sensors which are located all throughout UbiComp facilities and districts.

category:Technology