AI

---

Stephen Hawking once warned, “The  development of full artificial intelligence could spell the end of the  human race….It would take off on its own, and re-design itself at an  ever-increasing rate. Humans, who are limited by slow biological  evolution, couldn’t compete, and would be superseded.” 

Artificial  intelligence advancements are built by numerical cycles that influence  expanding figuring capacity to convey quicker and more precise models  and estimates of operational systems or upgraded portrayals and blends  of huge data sets. Nevertheless, while these trendsetting innovations can play out certain assignments with higher productivity and precision,  human ability assumes a basic part in planning and using AI innovation. 

Human intelligence is the thing that shapes the development and appropriation of man-made consciousness and creative arrangements related to it. It is human  intelligence that tries to ask 'why' and considers 'imagine a scenario  where' through basic reasoning. As engineering design keeps on being  tested by complex issues and the nature of information, the requirement  for human oversight, skill, and quality affirmation is basic in using AI-created yields. 

The idea of making a machine that can think like human beings has arrived  from the fiction world to the real world. Robots can do the task that  earlier was impossible. We have, since quite a while ago, endeavoured to acquire Intelligence in Machines to facilitate our work. There are bots, humanoids, robots, and computerized people that either outflank people or organize with us from numerous points of view. These AI-driven applications have a higher speed of execution, have higher operational capacity and exactness, while likewise profoundly huge in dreary and monotonous positions compared with humans.

In many respects, artificial intelligence (AI) has become so advanced it's more interesting to examine the things it can’t do. Despite  AI's world-bending abilities, machines still pale in comparison to the human mind on a host of tasks. Even algorithms built to replicate the function of the human brain – known as neural networks – are relatively  unsophisticated compared to the inner workings of our minds. "A grand  mystery in the study of intelligence is what gives us such big  advantages over AI systems," says Xaq Pitkow, an associate professor at  Carnegie Mellon University who studies the intersection of AI and  neuroscience. "The brain has a lot of deep neurological structures  related to different functions and tasks, like memory, values, movement  pattens, sensory perception and more." These structures let our minds  dip into different kinds of thinking to solve different kinds of problems. It's what gives humanity the edge over the robots, for now.

The AI algorithms that dominate the market are essentially prediction machines. They crunch massive amounts of data and analyse patterns,  which allows them to identify the most likely answer to a given  question. On a fundamental level, much of human cognition centres around  prediction, too, Pitkow says, but the mind is built for levels of  reasoning, flexibility, creativity and abstract thinking that AI still  hasn't replicated. We can say that the human brain is analogous, but machines are digital. A  simple difference is that human beings use their brain, ability to  think, memory, while AI machines depend on the data given to them.

As we all know that humans learn from past mistakes and intelligent ideas and intelligent attitudes lie at the basis of human intelligence. Hence,  this point is simply because machines cannot think and learn from the  past. They can learn from information and through regular training, but they can never attain the thinking procedure unique to humans.  Artificial intelligence takes much more time to adjust to the new  changes whereas human beings can adapt to changes easily and this makes  people able to learn and ace several abilities.

Machines can handle more data at a speedier rate as compared to humans. As of now, humans cannot beat the speed of computers. Artificial Intelligence  has not aced the ability to choose up on related social and excited codes. People are many ways better at social interaction since they can develop academic data, have self-awareness, and are elegant to others’  emotions.

It is still not possible to exactly mimic the level of human intelligence. Computers cannot copy the thought process of humans and according to experts, it won’t be possible in the near future. Since scientists and experimenters still don’t know the maze behind the human thought  procedure. It is highly uncertain that we will generate machines that can think like humans anytime soon.

Chess was once seen as an ultimate test of intelligence, until computers defeated humans while showing none of the other broad capabilities we associate with smarts. AI has since bested humans at Go, some types of  poker, and many video games. So researchers are developing AI IQ tests meant to assess deeper humanlike aspects of intelligence, such as concept learning and analogical reasoning. So far, computers have struggled on many of these tasks, which is exactly the point. The test-makers hope their challenges will highlight what’s missing in AI,  and guide the field toward machines that can finally think like us.

A common human IQ test is Raven's Progressive Matrices in which one needs to complete an arrangement of nine abstract drawings by deciphering the underlying structure and selecting the missing drawing from a group of options. Neural networks have gotten pretty good at that task. But a paper presented at the massive AI conference known as NeurIPS offers a new challenge: The AI system must generate a fitting image from scratch, an ultimate test of understanding the pattern. “If you are developing a computer vision system, usually it recognizes without really understanding what’s in the scene,” says Lior Wolf,  a computer scientist at Tel Aviv University, and the paper’s senior author. This task requires understanding composition and rules, “so it’s a very neat problem.” The researchers also designed a neural network to tackle the task—according to human judges, it gets about 70 percent correct, leaving plenty of room for improvement.

Other tests are harder still. Another NeurIPS paper presented a software-generated dataset of so-called Bongard Problems, a classic test for humans and computers. In their version, called Bongard-LOGO,  one sees a few abstract sketches that match a pattern and a few that  don’t, and one must decide if new sketches match the pattern. The  puzzles test “compositionality,” or the ability to break a pattern down  into its component parts, which is a critical piece of intelligence,  says Anima Anandkumar, a computer scientist at the California Institute  of Technology and the paper’s senior author. Humans got the correct  answer more than 90 percent of the time, the researchers found, but state-of-the-art visual processing algorithms topped out around 65  percent (with chance being 50 percent). “That’s the beauty of it,”  Anandkumar said of the test, “that something so simple can still be so challenging for AI.” They’re currently developing a version of the test with real images.

Compositional  thinking might help machines perform in the real world. Imagine a street scene, Anandkumar says. An autonomous vehicle needs to break it down into general concepts like cars and pedestrians to predict what will happen next. Compositional thinking would also make AI more interpretable and trustworthy, she added. One might peer inside to see how it pieces evidence together. 

Still harder tests are out there. In 2019, François Chollet, an AI researcher at Google, created the Abstraction and Reasoning Corpus (ARC), a set of visual puzzles tapping into core human knowledge of  geometry, numbers, physics, and even goal directness. On each puzzle,  one sees one or more pairs of grids filled with coloured squares, each  pair a sort of before-and-after grid. One also sees a new grid and fills  in its partner according to whatever rule one has inferred.

A website called Kaggle held a competition with the puzzles and awarded  $20,000 last May to the three teams with the best-performing algorithms.  The puzzles are pretty easy for humans, but the top AI barely reached  20 percent. “That’s a big red flag that tells you there’s something  interesting there,” Chollet says, “that we’re missing something.”

The  current wave of advancement in AI is driven largely by multi-layered  neural networks, also known as deep learning. But, Chollet says, these  neural nets perform “abysmally” on the ARC. The Kaggle winners used old-school methods that combine handwritten rules rather than learning subtle patterns from gobs of data. Though he sees a role for both paradigms in tandem. A neural net might translate messy perceptual data into a structured form that symbolic processing can handle.

Anandkumar agrees with the need for a hybrid approach. Much of deep learning’s progress now comes from making it deeper and deeper, with bigger and bigger neural nets, she says. “The scale now is so enormous that I think we’ll see more work trying to do more with less.” Anandkumar and Chollet point out one misconception about intelligence: People confuse it with skill. Instead, they say, it’s the ability to pick up new skills easily. That may be why deep learning so often falters. It typically requires lots of training and doesn’t generalize to new tasks, whereas the Bongard and ARC problems require solving a variety of puzzles with only a few examples of each.

Superintelligent AI is a theoretical form of artificial intelligence that surpasses human intelligence in virtually all domains. While many see artificial general intelligence (AGI) as the final realization of the emerging technology's full potential, superintelligent models would be what comes next. While today’s state-of-the-art AI systems have not yet reached these thresholds, advancements in machine learning, neural networks, and other AI-related technologies continue to progress—and  people are by turns excited and worried. 

You may find it interesting to reflect on predictions from the late Arthur C Clarke in 1964.