hindiallnews

lifestyle/myth news, lovelife news, tech news, lifestyle news, health news, all news, jokes, news, jobs news, health tips,

Friday, May 27, 2022

/ What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented

What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented

Friends, let us tell you that this is all you need to know about the way to make AI as smart as human.

What is Artificial General Intelligence?

And that is, Artificial General Intelligence is the ability of an intelligent agent to understand or learn any intellectual task and that which can be done by a human being. It is a primary target of some artificial intelligence research and is a common theme in science fiction and futures studies only.

And at the same time, it will be an Artificial General Intelligence (AGI) machine that will be able to understand the world as well as any human, and then have the same ability to learn how to perform a huge range of tasks. be completed.

While this AGI does not exist, it has only been featured in science-fiction stories for over a century, and has been popularized in modern times by films such as 2001: A Space Odyssey and That.

As you may know, AGI's fictional depictions vary widely, although they lean more toward dystopian visions of intelligent machines wiping out humanity and either enslaved them, as in films such as The Matrix or The Terminator. Has been observed. In such stories itself, the AGI is often cast as indifferent to human suffering or even intent on the destruction of the human race itself.

Friends, he contrasts utopian fantasies, such as Ian M. Banks's novel The Culture of Civilization, and casts AGI as the benevolent patron, running egalitarian societies free of suffering, where residents pursue their passion and technological advancement. It can then proceed at a breathless pace.
What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented
Friends, do these ideas bear any resemblance to real-world AGI, and at the same time, it is unknown because nothing like this has been created, or is it, according to many people working in the field of AI, it is to be made. is close to.

What can an artificial general intelligence do?

It should be noted that in theory, an artificial general intelligence can do any number of tasks a human can do, and possibly many that a human cannot. At the very least, an AGI would be able to combine human-like, flexible thinking and reasoning with computational benefits, such as near-instant recall and its accompanying split-second number crunching.

And it is this use of this intelligence to control mobile robots at least as dexterous as a person would create a new breed of machines that can perform any human task. With time, I will be able to handle every role that human beings perform. Initially, humans may be cheaper than machines, or humans working with AI may be more effective than AI itself. But the advent of AGI will likely make human labor obsolete.

Friends, effectively eliminating the need for human labor will have huge social implications, affecting both the population's ability to feed itself and a sense of purpose and self-employed.

Friends, even today, the debate over the ultimate impact on the very different, narrow AI jobs that it currently has is what has inspired some to introduce Universal Basic Income (UBI).

BT give that under UBI everyone in the society will get regular payment from the government with no strings attached. And at the same time that its approach is divisive, some advocates argue that it will provide a universal safety net and reduce bureaucratic costs. However, some anti-poverty campaigners have created economic models that show that such a scheme could worsen deprivation among vulnerable groups if it replaces existing social security systems in Europe.

And friends that beyond the impact on social cohesion, the advent of artificial general intelligence may be profound. The ability to employ an army of intelligence equal to that of the best and brightest humans could help develop new technologies and approaches to mitigating complex problems such as climate change.

And that on a more mundane level, such systems can perform everyday tasks, from surgeries and medical diagnostics to driving cars, at a consistently higher level than all humans can do - which in aggregate amounts to time, money, and lives saved. There could be a big positive in the matter.

And the downside is that this combined intelligence can also have profound negative effects: empowering population surveillance and control, and placing power in the hands of a small group of organizations, reducing horrific weapons, and It only removes the need for governments to take care of obsolete populations.

Let it be known that this should not stop AI from bullying: Michael Dell

What is AI? Everything's You need to knows about Artificial Intelligence

Deloitte: And That's What Technological Trends Will Make Symphonic Enterprise
What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented
  1. Tell that this Sophia robot is crowdfunding her brain
  2. Can an artificial general intelligence beat humans as well?

Yes, not only would this kind of intelligence have the same general abilities as humans, but it would be augmented by the advantages that computers have today - perfect memory, plus that ability to calculate quickly.

When will Artificial General Intelligence be invented?

Friends tell that it depends on whom you ask, the answer is within 11 years and that is never.

Friends, this is one of the reasons why it is so difficult to detect the lack of a clear path for AGI. Today machine-learning systems underpin online services, as well as those with which computers can recognize language, understand speech, recognize faces, and describe photos and videos. These recent successes, as well as high-profile successes such as the dominance of AlphaGo's notoriously complex Game of Go, may give the impression that society is on a fast track to the development of AGI

Yet the systems in use today are generally just one-note, excellent at the same task after extensive training, but useless for anything else. Their nature is very different from that of a normal intelligence and one that can do whatever it is asked of, and thus these narrow AIs are not necessarily the only stones to develop this AGI.

It should be noted that this was co-authored by Yoav Shohm of the Stanford Artificial Intelligence Laboratory, in a recent report that sheds light on the limited capabilities of today's narrow-minded AI.

Friends, that while machines can exhibit stellar performance on a certain task, if the task is modified even a little, and only then it can cause a dramatic drop in performance."

"For example, a person who can read Chinese characters can understand Chinese speech, know only a little about Chinese culture, and even in Chinese restaurants this So does good recommendations. Conversely, it would require very different AI systems for each of these tasks."

WATCH: AI And How To Apply It With Machine Learning (hindiallnews) | Download the report as PDF.

And at the same time, Michael Woolridge, Head of the Department of Computer Science at the University of Oxford, emphasized at this point in the report, "neither I nor anyone else would know how to measure that progress towards AGI". .

He is, in spite of this uncertainty, some of the most outspoken supporters of the near future AGI. Perhaps most famous of all is Google's engineering director Ray Kurzweil, who predicts that only by 2029 this capable AGI will exist to pass the Turing test, and that by the 2040s, affordable computers will do as many calculations per second as the whole. The combined mind of the mind. Human race.

While Kurzweil's supporters point to his successful track record in forecasting technological progress, Kurzweil estimates that by the end of 2009, only 80% of the predictions he made in the 1990s came true. Had happened.

Friends, Kurzweil's belief in this rate of progress stems from what he calls the Law of Acceleration of Return. And at the same time in 2001 he stated that the exponential nature of technological change, where each advance accelerates the rate of future success, means that mankind will be experiences that equivalents of 20,000 year's of that technological progressing in the 21st century. 

And it is these rapid changes in areas such as computer processing power and brain-mapping techniques that undermine Kurzweil's confidence in the near-future development of the hardware and software needed to support AGI.

What is Superintendence?

Let us tell you that Kurzweil believes that once AGI exists, it will improve itself at an exponential rate, and that it is rapidly evolving to the point where its intelligence is understood by humans. operates beyond the level. He refers to this point as the singularity itself, and states that it will occur in 2045, the stage at which an AI will exist that is "a billion times more powerful than all human intelligence today".

And so this idea of ​​near-future superintendence has prompted some of the world's foremost scientists, as well as technologists, to warn of the serious risks posed by AGI. SpaceX and Tesla founder Elon Musk called the AGI the "greatest existential threat facing humanity" and Stephen Hawking, a renowned physicist and professor at the University of Cambridge, told the BBC that "the development of full artificial intelligence is the end of mankind." Can do magic too"

And so both were signatories of an open letter calling on the AI ​​community to engage in "researching all those ways to make AI systems strong and profitable".

And so has Nick Bostrom, philosopher and director of Oxford University's Future of Humanity Institute, cautioned about what can happen when superintendence is reached.

And so while describing the superintendence as a bomb waiting to be detonated by irresponsible research, he believes that superintelligence agents can pose a threat to humans, and that one that could possibly "make its way". I" can only stand.

"Friends, if the robot becomes powerful enough," Bostrom said, "and it can seize control only to get this reward."


Is it even sensible to talk about AGI?

Friends, the problem with discussing the effects of AGI and superintendence is that most people who do that work in the field of AI insist that AGI is currently imaginary, and at the same time, I think it's too long. can last for a long time.

And at the same time that Microsoft Research Cambridge's laboratory director Chris Bishop has said that discussions about artificial general intelligence are "absolute nonsense", "at best, such discussions are decades away".

While these kinds of discussions are worse than pointless intimidation, other AI experts say they are diverting attention from all the near-future risks posed by today's narrow AI.

Andrew Ng, a well-known figure in the field of deep learning, has previously worked on the "Google Brain" project as well as served as chief scientist at the Chinese search giant Baidu. He recently called on AI and ethics debaters to "eliminate the AGI bullshit," and he spent even more time focusing on how today's technology is progressing or "job loss/stagnant." Wages, undermining democracy, discrimination/prejudice, wealth inequality" will exacerbate all the problems she faces. ,

Even uncovering AGI's potential boom may hurt public perception of AI, and it's in the comparatively limited capabilities of existing machine-learning systems and their narrow, one-note skills. It can also lead to frustration – be it translating text or recognizing faces.

How would you build an artificial general intelligence?

Demis Hassabis, the co-founder of Google DeepMind, argues that the secrets of general artificial intelligence lie in nature.
What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented
Hasabis and his colleagues believe that it is important for AI researchers to engage in "investigating the inner workings of the human brain – and this is the only existing evidence that such intelligence is possible".

"The study of animal cognition and its neural implementation also plays an important role, and that is because it can only provide a window into various important aspects of high-level general intelligence," he wrote in a paper last year.

And he argues that doing so will only help inspire new approaches to machine learning and new architectures for neural networks, the mathematical models that make machine learning possible.

Hasabis and his colleagues say that most AI systems are missing "key elements of human intelligence", including how infants build mental models of the world that make predictions about what will happen next. and that which allows them to plan.

Also absent from current AI models is the human ability to learn from only a few examples, to generalize the knowledge learned in one instance to a number of similar situations, such as if a new driver understands that only from that car. How to drive more and the one in which they learned.

New tools for brain imaging as well as genetic bioengineering have begun to offer detailed characterization of the computations that occur in neural circuits, and one that promises to revolutionize our understanding of mammalian brain function. According to the paper, neuroscience should "serve as a roadmap." And this is the AI ​​Research Agenda".

So another point of view comes from Yann Lecan, Facebook's chief AI scientist, and the one who took this lead in machine-learning research because of his work on convolutional neural networks.

He believes that the way toward general AI lies in developing systems that can build models of the world that they can then use to predict future outcomes. A good way to achieve this, he said in a talk last year, could be using Generative Adversarial Networks (GANs).

And so in a GAN, the two neural networks fight, the generator network tries to create "fake" data and the discriminator network tries to tell the difference between the fake and the real data as well. With each training cycle, the generator gets better at producing simulated data and the discriminator keeps a close eye on identifying those fakes. 

By pitting the two networks against each other during training, both can achieve better performance. GANs have been used to do some remarkable things, such as converting these dashcam videos from day to night or from winter to summer.

  1. Let's say this Democratic artificial intelligence will shape the technologies of the future: and this Gartner
  2. DeepLocker: When Malware As Well As It Turns Artificial Intelligence Into Weapon
  3. IEEE just published this draft report on 'ethically aligned' AI design so They biggest threat to The artificial intelligence: So that human stupidity.

Would an artificial general intelligence have consciousness?

Note that this is a very difficult question to answer, given the many definitions of consciousness.

And with this he shows a famous thought experiment by philosopher John Searle showing how difficult it would be to determine and whether AGI was truly self-aware.

Searle's Chinese Room suggests a hypothetical scenario and one in which the philosopher is presented with a written question in an unfamiliar Chinese language. Searle is sitting alone in a closed room and has slipped under the door in a different alphabetical order of each word in the query. Despite not understanding the language, Searle is able to follow the instructions he is given by a book in the room to manipulate all the symbols and numbers. 
What is artificial general intelligence [ AGI ]?Who artificial general intelligence invented
These instructions allow him to create his own series of Chinese characters which he feeds under the door. And by following the instructions Searle is able to create an appropriate response and fool the person outside the room into thinking that Searle is a native speaker inside, despite not understanding Chinese. 

As such, Searle argued that the experiment shows that a computer can interact with people and at the same time it appears to understand a language, while having no real understanding of its meaning.

Friends, this experiment has been used to attack the Turing test. devised by the genius mathematician and father of computing Alan Turing, and his test shows that a computer can be classified as a thinking machine if it can fool a third of the people and Were believing that it was a human.

See: And that censored enterprise: IoT, ML, and big data (ZDNet Special Report) | Download the report as PDF (TechRepublic)

Friends, let us tell you that in a recent book, Searle says that this uncertainty over the true nature of an intelligent computer extends to consciousness only. The In his books Language's or Consciousness so he's says: "Just as it's not sufficiently for The behavioral consciousness itself.

computational models of consciousness are not sufficient for consciousness itself" going on to give an example that: "any Even the computational model does not predict that in London itself this rainstorm will wet us".

Searle makes a distinction between strong AI, where AI can be called the brain, and with it weak AI, where the AI ​​is instead a concrete model of the mind.

And the Chinese have raised various counterarguments to Room and Searle's findings, ranging from the argument that the experiment misrepresents the nature of the mind, and that Searle is part of a wider system. and one who understands the Chinese language as a whole. ,

Along with Stuart Russell and Peter Norvig, who write the definitive textbook on artificial intelligence, and friends that the distinction between a simulation of the mind and a real mind, it is also the question of why most AI researchers focus more on results than on intrinsics. let's focus. nature of the system.

So can ethics be engineered into Artificial General Intelligence systems?

Well it can happen, but there is no good example of how it can be achieved.
And so Russell paints a clearer picture of how AI's ambitions for human morality can be tarnished.

So Let's say that's First "Imagines you have households Robot. And It's takings Care of your kids at home and the kids have eaten their food and are still hungry. It looks in the fridge and it's too much to eat." There's nothing left. The robot is thinking what to do, then it looks at the kitty, and with that you can think of what might happen next," he said.

Friends, tell me that this is a misunderstanding of human values, it is not understood that even a cat's sentimental value is more than the nutritional value.

And so did Vyacheslav W., of the Oxford Internet Institute. Polonsky argues that all people must first properly codify morality before the AGI can be given the gift of morality.

So that's "an machines can't be a taught what's fair unleashed the engineer's And Who designs AI System's have that precise concept of fairness," he writes, going to the question of what is fair. can be taught how to "algorithmically only maximize this fairness" or "overcome racial". and also this gender bias in the training data".

Friends, Polonsky's suggested solution to these problems is clearly defining ethical behavior itself - citing Germany's Ethics Commission on Automatic and Connected Driving that he designers of self-driving cars with ethical values. Program the system and those that put the safety of human life above all else. 

Another possible answer he highlights is to train a machine-learning system on what constitutes ethical behavior, drawing on many different human examples. And friends. One such repository of this data could be MIT's Ethical Machine Project, which asks participants to judge the 'best' response in difficult hypothetical situations, such as five people in a car or five pedestrians. It is better to kill it or not.

Of course, such approaches are full of misinterpretation and that potential for unintended consequences.

Hard-coding ethics into machines seems like a huge challenge, given the impossibility of predicting every situation a machine can find itself in. If a collision is unavoidable, will a self-driving car drop someone in their sixties and that or that or a child? What if that child has an incurable disease? What if the person in your sixties that you were the sole caregiver of your partner?

And also it might be a better solution to have a machine to know what is ethical behavior from human examples, and friends, even if that machine carries the same risk of encoding it as exists in the wider population.

Friends, it is Russell's suggestion that intelligent systems and robots can gain an understanding of human values ​​over time, through their shared observation of human behavior, as recorded today and throughout history. 

Russell suggests a method that robots can use to achieve such appreciation of human values, and that may be through inverse reinforcement learning, a machine-learning technique where a system is trained only for the desired behavior. It is only by rewarding that he is trained.

  1. Guy's, that Optus or Curtin University Partners for The Artificial Intelligence Research
  2. HR and Artificial Intelligence?
  3. And that the ethical challenges of artificial intelligence
  4. AI for Business: Why artificial intelligence and machine learning will be revolutionary
  5. And that's how we can stop a normal AI from breaking its barriers?
  6. And so as part of its mission to combat existential risks, the US-based Future of Life Institute (FLI) funded various research into AGI security, in anticipation of an AI capable of causing harm in the near future. has done.

So that's "in orderi to justifying an modest investments in That AI robustness researching in The probabilities that doesn't need to be The high, just negligible, and that just as a modest investment in home insurance has a non-negligible chance of a house burning down." justified by," it said. 

And with this he began his research program, pointing out that in the 1930s one of the greatest physicists of the time, Ernest Rutherford, had said that nuclear energy was "moonlight", the discovery of nuclear fission. Since that exactly five years ago.

Friends, before the behavior of the AGI can be constrained, the FLI argues that it is necessary to indicate precisely what it should and should not do.

So At The same Time, "in This order to build The well behaved system's, And we needs to defined exactly what's 'good behavior' means in each application domain. Simplified rules are what that design means." To do this – for example, to control the decisions of self-driving cars in critical situations – would be "from both ethicists And that Computer's scientist's," The said researchy a priority's report compiled by Stuart's Russell's and the others academics. Expertise is required."

Friends, the paper states that ensuring appropriate behavior becomes problematic with robust, generic AI, adding that it is only in societies that align the values ​​of powerful AI systems with their own values ​​and preferences. may face significant challenges.

"Friends he considers, for example, the difficulty of creating a utility function that includes an entire body of laws; even this so-called literal representation of the law itself is far from our current capabilities, and that Its behavior would be highly unsatisfactory," it says.

And it will then need to address deviant behavior in the AGI as well, the FLI says. Just as an airplane's onboard software undergoes rigorous checks for bugs that can only trigger unexpected behavior, so the code that underlies an AI must be subject to the same formal constraints.

And it has projects like seL4 for traditional software, which have developed a complete, general-purpose operating-system kernel and one that does this mathematically against a formal specification in order to give a strong guarantee against crashes and unsafe operations. Checked with.

Friends, however, in the case of AI, there may be a need for new methods of verification as per FLI.

"Friends, perhaps the most important difference between validation of traditional software and validation of AI systems is that the correctness of traditional software is defined with respect to a certain and known machine model of it, whereas AI systems - specifically The robot and its other embedded systems operate in environments that are best known only partially by the system designer.

"So Friend's, that's cases, it's May be practical to The verify that the system works correctly, and that it has it, to avoid the problem of modeling the real environment," said the research. Is.

FLI suggests that it should be possible to build an AI system from components, each of which has been validated.

While the risks of abuser AGI are particularly high, it suggests that such systems are the only way to isolate it from the wider world.

"And it's only very common and capable of system-specific security problems that they'll face it. In particular, it may be useful to create 'containers' for AI systems if the legality and control problems aren't resolved." which can lead to undesirable behavior and consequences in a less controlled environment," it says.

And the trouble with that is that ensuring humans can take control of normal AI isn't straightforward.

For example, a system is likely to do its best to solve the problems and those that prevent it from accomplishing its desired function.

And with that I'd say it can be problematic, however, if we want to redeploy the system, disable it, or significantly change its decision-making process, and that's it. It is only in this way that such a system logically avoids these changes, this research suggests.

The FLI recommends more research into systems that can be repaired, and those that do not exhibit this behavior.

According to research. That, "it may be possible to design utility functions or decision processes so that a system shuts down and doesn't try to avoid repurchases either."

To point out, another potential problem may stem from AI in pursuit of its goals by negatively affecting its environment - to the FLI to suggest more research into setting "domestic" goals. Only those who are limited in scope.

In addition, it recommends that more work needs to be done in the nature and in the nature of "intelligence explosions" between AIs - where AI's ability to self-improve humans' ability to control them. progresses far beyond.

Those are IEEE's own recommendations for building secure AGI systems, which are roughly in line with FLI research. And that includes that AGI systems should be transparent and that human operators should understand their reasoning, that a "safe and secure" environment should be developed in which AI systems can be developed and simultaneously tested, systems developed. so as to fail well in the event of tampering or crash and such systems should not only resist being shut down by operators.

And that is, today the question of how to develop AI in a beneficial way for society as a whole is the subject of ongoing research by the non-profit organization OpenAI.

This FLI research speculates that a simple AI, given the right checks and balances, can change society for the better: "The breakthrough in the pursuit of artificial intelligence has the potential to bring this unprecedented benefit to humanity itself, and So it's worthwhile to research how to maximize these. And that's the benefits while avoiding the potential pitfalls."

More guides to AI and related technologies

And what is the difference between Artificial Intelligence and Artificial General Intelligence?
Let us tell you that this Artificial Intelligence (AI) is a concept to make a machine capable of thinking, acting and learning like humans. And with this, Artificial General Intelligence (AGI) is the intelligence of a machine that can successfully perform any intellectual task and that which a human can do.


Why is Artificial General Intelligence important?

Simply put, AI allows organizations to make better decisions while improving core business processes by increasing both the speed and accuracy of strategic decision-making processes.


What is Deep Learning? Everything You Need To Know

Explain this shortcoming of Deep Learning: How does it relate to the broader field of Machine Learning, and how to get started with it.


AI in the Workplace: Everything you need to know

And that's how artificial intelligence will change the world of work for the better as well as it for the worse.


What is Machine Learning? Everything You Need To Know

So This is Guide to explained what's machine learning is And how it is relates to artificial intelligence, how it works and why it matters.


What is AI? Everything's you Needs to knows about Thee Artificial Intelligence

Friends this is a working guide to Artificial Intelligence from Machine Learning and General AI to Neural Networks only.


What is meant by Artificial General Intelligence?

So it is a representation of human cognitive abilities generalized in Artificial General Intelligence (AGI) software, so that when faced with an unfamiliar task, that AGI system can find a solution. And at the same time, the purpose of the AGI system is to do any work and that which a human being is capable to.

No comments:

Post a Comment