#ability to infer people's psychological state + power that affects your mind and emotions is a bit ಠ_ಠ ;;

[turtledotjpeg]

dumb.......but it appeared in my brain and i had to do it (ﾉ •_•)ﾉ

bonus: wip ft. even less hinged kurapika

[adamblrworld-blog]

Artificial Intelligence (AI)

1) What is Artificial Intelligence?

Artificial Intelligence(AI) is a form of Computer Science used to create intelligent machines that can recognize human speech,objects,can learn,plan and solve problems like humans

It has become an essential part of the technology industry.Research associated with artificial intelligence is highly technical and specialized.Machines can often act and react like humans only if they have abundant information relating to the world. Artificial intelligence must have access to objects, categories, properties and relations between all of them to implement knowledge engineering.

2) What is AI used for today?

It may come as a surprise that artificial intelligence is all around us - and has even permeated our routine on a daily basis. Whether on our phones or at the cutting edge of technological development, artificial intelligence is all around.

Siri/Alexa

Whether or not you've thought about that voice in your phone as a product of AI or not, Apple's Siri and Amazon's Alexa both use AI to help you complete tasks or answer questions on your mobile devices.

Facebook Feed

Believe it or not, your Facebook feed is actually using AI to predict what content you want to see and push it higher

Tesla

Its founder being vocally suspicious of advanced AI technology, Tesla's electronic cars use a variety of AI - including self-driving capacities. Tesla also uses crowd-sourced data from its vehicles to improve their systems

Netflix

The media streaming site uses advanced predictive technology to suggest shows based on your viewing preferences or rating. And while the data currently seems to favor bigger, more popular films over smaller ones, it is becoming increasingly sophisticated.

Ride-sharing Apps Like Uber and Lyft

How do they determine the price of your ride? How do they minimize the wait time once you hail a car? How do these services optimally match you with other passengers to minimize detours? The answer to all these questions is Machine Learning(ML).

Commercial Flights Use an AI Autopilot

AI autopilots in commercial airlines is a  surprisingly early use of AI technology that dates as far back as 1914, depending on how loosely you define autopilot. The New York Times reports that the average flight of a Boeing plane involves only seven minutes of human-steered flight, which is typically reserved only for takeoff and landing

3)What are the 4 types of AI?

I)Reactive machines

The most basic types of AI systems are purely reactive, and have the ability neither to form memories nor to use past experiences to inform current decisions. Deep Blue, IBM’s chess-playing supercomputer, which beat international grandmaster Garry Kasparov in the late 1990s, is the perfect example of this type of machine.

Deep Blue can identify the pieces on a chess board and know how each moves. It can make predictions about what moves might be next for it and its opponent. And it can choose the most optimal moves from among the possibilities.

II) Limited memory

This Type II class contains machines can look into the past. Self-driving cars do some of this already. For example, they observe other cars’ speed and direction. That can’t be done in a just one moment, but rather requires identifying specific objects and monitoring them over time.

We might stop here, and call this point the important divide between the machines we have and the machines we will build in the future. However, it is better to be more specific to discuss the types of representations machines need to form, and what they need to be about.

III)Theory of mind

Machines in the next, more advanced, class not only form representations about the world, but also about other agents or entities in the world. In psychology, this is called “theory of mind” – the understanding that people, creatures and objects in the world can have thoughts and emotions that affect their own behavior.

IV) Self-awareness

The final step of AI development is to build systems that can form representations about themselves. Ultimately, we AI researchers will have to not only understand consciousness, but build machines that have it.

This is, in a sense, an extension of the “theory of mind” possessed by Type III artificial intelligence. Consciousness is also called “self-awareness” for a reason. (“I want that item” is a very different statement from “I know I want that item.”) Conscious beings are aware of themselves, know about their internal states, and are able to predict feelings of others. We assume someone honking behind us in traffic is angry or impatient, because that’s how we feel when we honk at others. Without a theory of mind, we could not make those sorts of inferences.

4)What is the purpose of AI?

The intended purpose of AI is to making an intelligent machine that initially thinks as good as a human, but eventually much better, even to the degree of far superior. The overall research goal of artificial intelligence is to create technology that allows computers and machines to function in an intelligent manner. The general problem of simulating (or creating) intelligence has been broken down into sub-problems.

The goal of AI is to create robots that can be adaptive like humans and perform multiple tasks. The main purpose in my eyes would be to save humans work

5)What are the advantages of artificial intelligence?

AI would have a low error rate compared to humans, if coded properly. They would have incredible precision, accuracy, and speed. They won't be affected by hostile environments, thus able to complete dangerous tasks, explore in space, and endure problems that would injure or kill us.

The greatest advantage of artificial intelligence is that machines do not require sleep or breaks, and are able to function without stopping. They can continuously perform the same task without getting bored or tired. When employed to carry out dangerous tasks, the risk to human health and safety is reduced

There are some benefits of artificial intelligence a. No mistake

 We use artificial intelligence in most of the cases. As this helps us in reducing the risk. Also, increases the chance of reaching accuracy with the greater degree of precision.

b. Easy Exploration

 In mining, we use artificial intelligence and science of robotics. Also, other fuel exploration processes. Moreover, we use complex machines for exploring the ocean. Hence, overcoming the ocean limitation.

c. Everyday Application

As we know that computed methods and learning have become commonplace in daily life. Financial institutions and banking institutions are widely using AI. That is to organize and manage data. Also, AI is used in the detection of fraud users in a smart card based system.

d. Digital Assistants

“Avatars” are used by highly advanced organizations. That are digital assistants. Also, they can interact with the users. Hence. They are saving human needs of resources.

As we can say that the emotions are associated with mood. That they can cloud judgment and affect human efficiency. Moreover, completely ruled out for machine intelligence.

e. No breaks

Machines do not require frequent breaks and refreshments for humans. As machines are programmed for long hours. Also, they can continuously perform without getting bored.

f. Increase Work Efficiency

For a particular repetitive task, AI-powered machines are great with amazing efficiency. Best is they remove human errors from their tasks to achieve accurate results.

g. Reduce cost of training and operation

Deep Learning and neural networks algorithms used in AI to learn new things like humans do. Also, this way they eliminate the need to write new code every time.

6)What are the disadvantages of artificial intelligence?

a. High Cost

Its creation requires huge costs as they are very complex machines. Also, repair and maintenance require huge costs.

b. No Replicating Humans

As intelligence is believed to be a gift of nature. An ethical argument continues, whether human intelligence is to be replicated or not.

c. Lesser Jobs

As we are aware that machines do routine and repeatable tasks much better than humans. Moreover, machines are used of instead of humans. As to increase their profitability in businesses.

d. Lack of Personal Connections

We can’t rely too much on these machines for educational oversights. That hurt learners more than help.

e. Addiction

As we rely on machines to make everyday tasks more efficient we use machines.

[existentialspiral]

Classpect Analysis: Witch of Mind

DEFINITION: Mind, as an aspect, is one of the more literal aspects, having control over the realms of thought, logic, judgement, decision-making, memory, and more, possibly even including the senses and perception if that association is truly a part of the aspect and not related to its users specifically. Mind is one of the easiest aspects to define, having a user who remained within story and focus for the majority of Homestuck’s run, along with a defined opposite aspect and a failed auxiliary user on both sides of a Scratch. Terezi Pyrope, Seer of Mind, who exhibited the use of her aspect long before entering the Medium, is easily one of the best characters to use for analysis, having a long list of achievements and displayed skills related to her aspect. More specifically, in a roughly chronological order, Terezi was able to, using nothing but her mental abilities: dismantle entire teams of FLARPers through politics and head-games, avenga Aradia by proxy through manipulating Doc Scratch (a being approaching omniscience), exile several agents of Derse throughout her session, kill John Egbert, force Dave Strider to realize his actual standing with death, kill Vriska Serket (who, as both an active player and God Tier, was incomprehensibly more powerful than her), define a system of memory-based chronolocation, and was even able to, through less than 10 commands, use John Egbert to manipulate both the timeline and she herself into solving both the issues that resulted in Game Over and all of her personal and romantic problems. By contrast, the other Mind player/s, Latula Pyrope (Knight of Mind) and her post-Scratch counterpart Neophyte Redglare, are rather lackluster, displaying little skill as a Knight of Mind, with their only achievement (utilizing the Marquis Spinneret Mindfang’s underestimation of their threat) being marred by a remarkably shortsighted failure (putting a known mind-controller into a courtroom filled with hundreds of trolls who, as low-bloods, had no protection against mind control). The Witch class, meanwhile, is difficult to place; while it also has multiple users, with multiple displayed uses, none of them are so dramatically defined. The largest, most obvious display of a Witch’s power is that of Jade (Witch of Space) during [S] Cascade, where she uses her newly Ascended powers to manipulate the mass and volume of the player’s Lands and the Battlefield, compressing them into a size approximate to that of a baseball, before doing the same in reverse to a Fenestrated Wall, expanding it to the point that a full-size battleship could pass through it. However, after this event, Jade doesn’t give many expansive displays of power. Moreover, all of Jade’s actions after ascending to Dog Tier are suspect, because Jade has the powers of both a Sprite and a First Guardian as well as those of a Witch of Space, and identifying which powers are derived from which source is a matter of considerable, possibly insurmountable, difficulty. Meanwhile, Feferi Peixes (Witch of Life) gives little in the way of useful information; Feferi never Ascended in the Alpha timeline, and the power exhibited by her alternate ghost-self, that of resurrecting the Mayor, is mirrored by the power over life used by Jane Crocker, a Maid of Life. This might suggest that Witch and Maid are active/passive counterparts, something possibly reinforced by Damara (Witch of Time) and Aradia Megido (Maid of Time) being each other’s Dancestor, but it provides only a faint outline of ability. Damara herself, however, does provide some details across both sides of the Scratch; firstly, as Damara Megido the original, she was stated to have appeared throughout the timeline, sabotaging her session before vanishing away, allowing us to infer that she must have possessed a good degree of control over time travel, manipulating the timeline to suit her ends (in this case, forcing her team to Scratch their session). As well, Damara’s incarnation as the Handmaid also displays a significand degree of control over the timeline, traveling throughout the time stream to deliver whatever destruction or havoc is called for in order to bring about the summoning of Lord English; moreover, the Handmaid is also shown to be quite skilled with, “Clockwork Majykks,” manipulating the fabric of time as a weapon. It needs be noted, however, that this last ability is implied to be a gift from Lord English as well as an inherent ability, and so might not actually be connected to her; nevertheless, we shall assume that it is an intrinsic ability of hers, if only to draw some sort of conclusion to this subject. Now, comparing and contrasting the examples given, the one constant seems to be in the manipulation of the Witch’s aspect, the actual substance of the aspect being bent, shaped, and used as the Witch desires. As such, with full acknowledgement that the justifications here are… not the best… we shall define the Witch class as, “One who manipulates or shapes their aspect.” With all this taken together, Witch of Mind would therefore parse as, “One who manipulates logic, thought, or memory.” Mind players are usually adept at head games, but few of them can hold a candle to the Witch of Mind; manipulation, scheming, plans, gambits, forks, and psychology are the Witch’s bread and butter. A Witch that specializes in manipulating people into doing what she wants rather than actually using magic, I get the feeling there’s a Discworld reference in there somewhere; regardless, headolo- sorry, psychology and planning are all the Witch of Mind truly needs to do her job. She’ll play you like a fiddle and walk away while your life burns down, sell you back the shirt on your back, and make you want to ask forgiveness for bothering her when all is said and done; this is the girl we were warned about, and not a one of us listened. All we can do is watch as her house of cards falls down on us, with her watching from the outside, smiling and playing a fiddle. ABILITIES: The Witch of Mind is a manipulator, a controller, and a user; their job, put simply, is the Great Game of Politics and Princes. As both a highly active class, and one placed into a support aspect, the Witch of Mind is going to have a heavy chain of deals and agreements ahead of her, with a thousand alliances pulling in every direction. It is, then, a very good thing that she has the perfect toolkit for the task ahead. At lower levels, the Witch of Mind is one of the weaker classpects, having little-to-no direct combat ability, with her aspect being one of the worst for directly dealing with game-hazards such as imps and ogres. Instead, even at low-level, the Witch of Mind will find herself far more at home asleep, playing at politics on either Prosopit or Derse, using her own instinctive talents at social manipulation to play Agents against each other, resulting in a healthy harvest of Exiles. For more information, go and read either the Game of Thrones, Artemis Fowl, or Discworld series. Meanwhile, in the waking world, the Witch would most likely utilize a Specibus that would either compensate for or work around her general non-combat status, such as gunkind, or possibly puppetkind. Reaching God Tier is a major turning point for the Witch. Where before she was forced to act through indirect manipulation, such as convincing others to do her bidding, or to change their opinions, or what have you, now she is unshackled from the limits of others minds, allowing her to manipulate her aspect directly. Mind control, memory altering, brainwashing, possibly complex illusions, and more are well within her grasp, allowing her to take to the field in force, not merely as a warrior, but as a commander. Her enemies find themselves unable to remember why they ever wanted to oppose the Witch of Mind, and find themselves doing exactly what she wants. Foes like the Black Queen and her Agents would be less than stumbling blocks, either destroying themselves through infighting, or even acting “all according to plan.” At the highest levels of power, the Witch of Mind has full and total control over the minds of others, able to decide for then what they think, want, and remember. Mere mind control, such as that used by Vriska and Aranea, is beneath her; why should you waste time forcing others to do your bidding, when you can simply make them want to do what you wish? Indeed, at this level, the difference between, “manipulated foe,” and, “willing slave,” is barely even semantic, as she can so easily rewrite what others think to make them do what she desires. Any actual challenge to Skaia’s game is lost, as the Witch of Mind turns every ally of Derse into another minion of her own until even the fearsome Black King must Choose between abdication or annihilation, a Choice he would barely call a choice at all. Of course, the Witch of Mind is not perfect. Obviously, her abilities only affect those with a mind to manipulate; the power to rewrite memories and decisions is rather ineffective when faced with, say, a burning meteor approaching your position at terminal velocity. As well, there is the important consideration that, as stated before, the Witch of Mind is not suited to a main combat role; neither her class nor her aspect are particularly inclined to heavy fighting, and her actual suite of abilities are all indirect, relying on those under her command to do the actual work. As well, it also needs to be noted that, while her abilities are expansive, they probably aren’t that finely controlled; while still able to brainwash her foes with the rest of them, something like a Manchurian Candidate situation is likely beyond her ability, and her ability to manipulate memory could only go so far. As well, it needs to be specifically noted that, while thoughts, decisions, and memories are under the sway of the Mind aspect, emotions belong to Heart, Hope, and Rage, with Heart being in direct opposition to Mind, and so are immune to the Witch’s tampering; she can make a person decide to marry someone, and can make them remember being in love and feeling wonderful, but cannot actually make them feel love or joy. QUEST: Unfortunately, anything I might say on the Quest of a Witch, of Mind or otherwise, would be pure speculation; the only Quest we are shown of a Witch is that of Jade, Witch of Space, whose personal Quest was supplanted by that of Frog Breeding. Of the other Witches’ Quests, neither are shown, they having abandoned their Quests to either spend time with Sollux or to play Quisling to their own session’s efforts. However, simply saying, “There’s not enough for me to go on,” and leaving it at that would be boring, so I shall at least share my speculations on the quest that would be provided here. I would guess that the Quest of a Witch of Mind would have to do with not only mastering her abilities as a manipulator or thoughts and people, but also have much to do with growing as a person, and learning NOT to use her abilities out of convenience (once again, I find myself drawing parallels between the Witch of Mind and the witches of Discworld). As the Witch of Mind is built around manipulating people, it would make sense that her Land would have a large and well-developed society, one which rewards ambition and political ability (such as that of upperclass Ancient Rome or China). Her Quest would likely have to do with both learning to use her powers, working to manipulate her Consorts into handing her a position of power, while also demonstrating the dangers of abusing her powers. My own guess would be a literal Game of Thrones, with the Throne in question being much like Marvel’s Throne of Satan, where nobody is able to take the throne themselves, or else everyone else (and possibly her Denizen) would tear them apart. On this Quest, her quest to “Seat the Throne,” the Witch of Mind would have to learn both to manipulate the consorts into allowing her the throne, and also learn not to simply control them when allowing them their own decisions works better. The Lands for a Witch of Mind would most likely have something to do with the Mind aspect, such as Thought, Logic, Karma, Decisions, Gambits, or even Games. The other half of the name would likely have something to do with the personality of the Witch in question, her specific Quest, the counter-aspect to Mind, specifically Heart, or even something seemingly unrelated to her or her quest, at least until later diagnosed. Some examples include, corresponding to the possibilities above, the Land of Choice and Champaign, the Land of Patricians and Gambits, the Land of Reason and Emotion, and the Land of Midnight and Latin. VERDICT: The Witch of Mind is one of the best non-combat classpects, one which even becomes overwhelmingly useful at the later stages of the game, but is also one that takes time to grow into its role; much like a Magikarp is useless until Lv. 20, a Witch of Mind is going to start the game mostly unable to contribute to the most immediate of struggles, those being primarily combat related. However, while the Witch of Mind will have to rely on noncombat EXP to climb her Echeladder, the moment the game shifts focus to dealing with Derse and its Agents is the exact moment the Witch becomes your most useful teammate; Mind players in general are exceptionally well-equipped to handling the challenges of the Black Queen, but the Witch of Mind is the single best Classpect for dealing with the Dersite royalty, rivalled only by the Mage of Mind and, possibly, the Rogue of Blood or Bard of Blood. On the other hand, the Witch of Mind is, by definition, a manipulator, and manipulators want the world to go according to their desires, and tend to get nasty when things don’t go their way. As such, loyalty is going to be a major issue when a party comes with a Witch of Mind, along with all the other issues that arise when dealing with a team member with mental abilities. As well, on that front, I find it most wise to paraphrase a quote whose source I cannot find at this moment: “It’s fine to have a telepath on the team, just make sure she’s saying, ‘I trust you,’ not, ‘you trust me.’” Maintaining a team with a Witch of Mind is a risky proposition, one which requires teammates who are fully able to deal with someone who, at their worst, could simply make you her thought slave. The aspects of Blood, Breath, Heart, or Rage will be your best choices, filtered through classes such as Mage, Knight, Heir, Bard, or even Lord; anyone who could truly provide some direction to the team and maintain focus on their goals, while also mediating between teammates, would be a mind-saver. As well, if the absolute worst should come to pass, the only possible defense would be in a Classpect with both impossible resistance to mental manipulation and unstoppable power, such as a Knight, Page, Bard, Prince, or Lord of Heart or Rage; it would take what are some of the most dangerous classpects possible to fight off a rogue Witch of Mind. As for synergizing classpects, the Witch of Mind is a controller, not a fighter, so a Knight of most any aspect would provide a very welcome defense, especially a Knight of Rage, and a Bard of Hope, Rage, Light, Life, or Doom would be a useful addition when managing a mind-slave army. As well, if you are insane enough to risk having two of the most dangerous possible classpects on a team, pairing a Witch of Mind and a Page of Rage makes for one of the most impossibly dangerous combinations possible, something verging on unstoppable. On the whole, a high-level Witch of Mind is a very risky player, where her only practical counters are all significantly more dangerous than she is if they were to go astray; however, a Witch of Mind who maintains both loyalty to the team and a proper sense of right and wrong is a player who could single-handedly beat the Black Queen, and possibly even upend the Black King by herself. While not the most powerful player, it is difficult to find one who could accomplish more than her.

[roboticscloud-blog]

FROM REACTIVE ROBOTS TO SENTIENT MACHINES: THE 4 TYPES OF AI

The common, and recurring, view of the latest breakthroughs in artificial intelligence research is that sentient and intelligent machines are just on the horizon. Machines understand verbal commands, distinguish pictures, drive cars and play games better than we do. How much longer can it be before they walk among us?

The new White House report on artificial intelligence takes an appropriately skeptical view of that dream. It says the next 20 years likely won’t see machines “exhibit broadly-applicable intelligence comparable to or exceeding that of humans,” though it does go on to say that in the coming years, “machines will reach and exceed human performance on more and more tasks.” But its assumptions about how those capabilities will develop missed some important points.

As an AI researcher, I’ll admit it was nice to have my own field highlighted at the highest level of American government, but the report focused almost exclusively on what I call “the boring kind of AI.” It dismissed in half a sentence my branch of AI research, into how evolution can help develop ever-improving AI systems, and how computational models can help us understand how our human intelligence evolved.

The report focuses on what might be called mainstream AI tools: machine learning and deep learning. These are the sorts of technologies that have been able to play “Jeopardy!” well, and beat human Go masters at the most complicated game ever invented. These current intelligent systems are able to handle huge amounts of data and make complex calculations very quickly. But they lack an element that will be key to building the sentient machines we picture having in the future.

We need to do more than teach machines to learn. We need to overcome the boundaries that define the four different types of artificial intelligence, the barriers that separate machines from us – and us from them.

Type I AI: Reactive machines

The most basic types of AI systems are purely reactive, and have the ability neither to form memories nor to use past experiences to inform current decisions. Deep Blue, IBM’s chess-playing supercomputer, which beat international grandmaster Garry Kasparov in the late 1990s, is the perfect example of this type of machine.

Deep Blue can identify the pieces on a chess board and know how each moves. It can make predictions about what moves might be next for it and its opponent. And it can choose the most optimal moves from among the possibilities.

But it doesn’t have any concept of the past, nor any memory of what has happened before. Apart from a rarely used chess-specific rule against repeating the same move three times, Deep Blue ignores everything before the present moment. All it does is look at the pieces on the chess board as it stands right now, and choose from possible next moves.

This type of intelligence involves the computer perceiving the world directly and acting on what it sees. It doesn’t rely on an internal concept of the world. In a seminal paper, AI researcher Rodney Brooks argued that we should only build machines like this. His main reason was that people are not very good at programming accurate simulated worlds for computers to use, what is called in AI scholarship a “representation” of the world.

The current intelligent machines we marvel at either have no such concept of the world, or have a very limited and specialized one for its particular duties. The innovation in Deep Blue’s design was not to broaden the range of possible movies the computer considered. Rather, the developers found a way to narrow its view, to stop pursuing some potential future moves, based on how it rated their outcome. Without this ability, Deep Blue would have needed to be an even more powerful computer to actually beat Kasparov.

Similarly, Google’s AlphaGo, which has beaten top human Go experts, can’t evaluate all potential future moves either. Its analysis method is more sophisticated than Deep Blue’s, using a neural network to evaluate game developments.

These methods do improve the ability of AI systems to play specific games better, but they can’t be easily changed or applied to other situations. These computerized imaginations have no concept of the wider world – meaning they can’t function beyond the specific tasks they’re assigned and are easily fooled.

They can’t interactively participate in the world, the way we imagine AI systems one day might. Instead, these machines will behave exactly the same way every time they encounter the same situation. This can be very good for ensuring an AI system is trustworthy: You want your autonomous car to be a reliable driver. But it’s bad if we want machines to truly engage with, and respond to, the world. These simplest AI systems won’t ever be bored, or interested, or sad.

Type II AI: Limited memory

This Type II class contains machines can look into the past. Self-driving cars do some of this already. For example, they observe other cars’ speed and direction. That can’t be done in a just one moment, but rather requires identifying specific objects and monitoring them over time.

These observations are added to the self-driving cars’ preprogrammed representations of the world, which also include lane markings, traffic lights and other important elements, like curves in the road. They’re included when the car decides when to change lanes, to avoid cutting off another driver or being hit by a nearby car.

But these simple pieces of information about the past are only transient. They aren’t saved as part of the car’s library of experience it can learn from, the way human drivers compile experience over years behind the wheel.

So how can we build AI systems that build full representations, remember their experiences and learn how to handle new situations? Brooks was right in that it is very difficult to do this. My own research into methods inspired by Darwinian evolution can start to make up for human shortcomings by letting the machines build their own representations.

Type III AI: Theory of mind

We might stop here, and call this point the important divide between the machines we have and the machines we will build in the future. However, it is better to be more specific to discuss the types of representations machines need to form, and what they need to be about.

Machines in the next, more advanced, class not only form representations about the world, but also about other agents or entities in the world. In psychology, this is called “theory of mind” – the understanding that people, creatures and objects in the world can have thoughts and emotions that affect their own behavior.

This is crucial to how we humans formed societies, because they allowed us to have social interactions. Without understanding each other’s motives and intentions, and without taking into account what somebody else knows either about me or the environment, working together is at best difficult, at worst impossible.

If AI systems are indeed ever to walk among us, they’ll have to be able to understand that each of us has thoughts and feelings and expectations for how we’ll be treated. And they’ll have to adjust their behavior accordingly.

Type IV AI: Self-awareness

The final step of AI development is to build systems that can form representations about themselves. Ultimately, we AI researchers will have to not only understand consciousness, but build machines that have it.

This is, in a sense, an extension of the “theory of mind” possessed by Type III artificial intelligences. Consciousness is also called “self-awareness” for a reason. (“I want that item” is a very different statement from “I know I want that item.”) Conscious beings are aware of themselves, know about their internal states, and are able to predict feelings of others. We assume someone honking behind us in traffic is angry or impatient, because that’s how we feel when we honk at others. Without a theory of mind, we could not make those sorts of inferences.

While we are probably far from creating machines that are self-aware, we should focus our efforts toward understanding memory, learning and the ability to base decisions on past experiences. This is an important step to understand human intelligence on its own. And it is crucial if we want to design or evolve machines that are more than exceptional at classifying what they see in front of them.

[d3robots-blog]

From Reactive Robots to Sentient Machines: The 4 Types of AI

The common, and recurring, view of the latest breakthroughs in artificial intelligence research is that sentient and intelligent machines are just on the horizon. Machines understand verbal commands, distinguish pictures, drive cars and play games better than we do. How much longer can it be before they walk among us?

The new White House report on artificial intelligence takes an appropriately skeptical view of that dream. It says the next 20 years likely won't see machines "exhibit broadly-applicable intelligence comparable to or exceeding that of humans," though it does go on to say that in the coming years, "machines will reach and exceed human performance on more and more tasks." But its assumptions about how those capabilities will develop missed some important points.

As an AI researcher, I'll admit it was nice to have my own field highlighted at the highest level of American government, but the report focused almost exclusively on what I call "the boring kind of AI." It dismissed in half a sentence my branch of AI research, into how evolution can help develop ever-improving AI systems, and how computational models can help us understand how our human intelligence evolved.

The report focuses on what might be called mainstream AI tools: machine learning and deep learning. These are the sorts of technologies that have been able to play "Jeopardy!" well, and beat human Go masters at the most complicated game ever invented. These current intelligent systems are able to handle huge amounts of data and make complex calculations very quickly. But they lack an element that will be key to building the sentient machines we picture having in the future.

We need to do more than teach machines to learn. We need to overcome the boundaries that define the four different types of artificial intelligence, the barriers that separate machines from us – and us from them.

Type I AI: Reactive machines

The most basic types of AI systems are purely reactive, and have the ability neither to form memories nor to use past experiences to inform current decisions. Deep Blue, IBM's chess-playing supercomputer, which beat international grandmaster Garry Kasparov in the late 1990s, is the perfect example of this type of machine.

Deep Blue can identify the pieces on a chess board and know how each moves. It can make predictions about what moves might be next for it and its opponent. And it can choose the most optimal moves from among the possibilities.

But it doesn't have any concept of the past, nor any memory of what has happened before. Apart from a rarely used chess-specific rule against repeating the same move three times, Deep Blue ignores everything before the present moment. All it does is look at the pieces on the chess board as it stands right now, and choose from possible next moves.

This type of intelligence involves the computer perceiving the world directly and acting on what it sees. It doesn't rely on an internal concept of the world. In a seminal paper, AI researcher Rodney Brooks argued that we should only build machines like this. His main reason was that people are not very good at programming accurate simulated worlds for computers to use, what is called in AI scholarship a "representation" of the world.

The current intelligent machines we marvel at either have no such concept of the world, or have a very limited and specialized one for its particular duties. The innovation in Deep Blue's design was not to broaden the range of possible movies the computer considered. Rather, the developers found a way to narrow its view, to stop pursuing some potential future moves, based on how it rated their outcome. Without this ability, Deep Blue would have needed to be an even more powerful computer to actually beat Kasparov.

Similarly, Google's AlphaGo, which has beaten top human Go experts, can't evaluate all potential future moves either. Its analysis method is more sophisticated than Deep Blue's, using a neural network to evaluate game developments.

These methods do improve the ability of AI systems to play specific games better, but they can't be easily changed or applied to other situations. These computerized imaginations have no concept of the wider world – meaning they can't function beyond the specific tasks they're assigned and are easily fooled.

They can't interactively participate in the world, the way we imagine AI systems one day might. Instead, these machines will behave exactly the same way every time they encounter the same situation. This can be very good for ensuring an AI system is trustworthy: You want your autonomous car to be a reliable driver. But it's bad if we want machines to truly engage with, and respond to, the world. These simplest AI systems won't ever be bored, or interested, or sad.

Type II AI: Limited memory

This Type II class contains machines can look into the past. Self-driving cars do some of this already. For example, they observe other cars' speed and direction. That can't be done in a just one moment, but rather requires identifying specific objects and monitoring them over time.

These observations are added to the self-driving cars' preprogrammed representations of the world, which also include lane markings, traffic lights and other important elements, like curves in the road. They're included when the car decides when to change lanes, to avoid cutting off another driver or being hit by a nearby car.

But these simple pieces of information about the past are only transient. They aren't saved as part of the car's library of experience it can learn from, the way human drivers compile experience over years behind the wheel.

So how can we build AI systems that build full representations, remember their experiences and learn how to handle new situations? Brooks was right in that it is very difficult to do this. My own research into methods inspired by Darwinian evolution can start to make up for human shortcomings by letting the machines build their own representations.

Type III AI: Theory of mind

We might stop here, and call this point the important divide between the machines we have and the machines we will build in the future. However, it is better to be more specific to discuss the types of representations machines need to form, and what they need to be about.

Machines in the next, more advanced, class not only form representations about the world, but also about other agents or entities in the world. In psychology, this is called "theory of mind" – the understanding that people, creatures and objects in the world can have thoughts and emotions that affect their own behavior.

This is crucial to how we humans formed societies, because they allowed us to have social interactions. Without understanding each other's motives and intentions, and without taking into account what somebody else knows either about me or the environment, working together is at best difficult, at worst impossible.

If AI systems are indeed ever to walk among us, they'll have to be able to understand that each of us has thoughts and feelings and expectations for how we'll be treated. And they'll have to adjust their behavior accordingly.

Type IV AI: Self-awareness

The final step of AI development is to build systems that can form representations about themselves. Ultimately, we AI researchers will have to not only understand consciousness, but build machines that have it.

This is, in a sense, an extension of the "theory of mind" possessed by Type III artificial intelligences. Consciousness is also called "self-awareness" for a reason. ("I want that item" is a very different statement from "I know I want that item.") Conscious beings are aware of themselves, know about their internal states, and are able to predict feelings of others. We assume someone honking behind us in traffic is angry or impatient, because that's how we feel when we honk at others. Without a theory of mind, we could not make those sorts of inferences.

While we are probably far from creating machines that are self-aware, we should focus our efforts toward understanding memory, learning and the ability to base decisions on past experiences. This is an important step to understand human intelligence on its own. And it is crucial if we want to design or evolve machines that are more than exceptional at classifying what they see in front of them.