AI – is a term commonly thrown around in every aspect of today’s day and age. It has been growing for years together, and the research will never stop.
AI, in simple terms, is an Artificial Intelligence machine programmed in a manner where they are required to replicate humans – namely, how smart they are, their intelligence, and problem-solving abilities. Large data sets are combined with clever, iterative processing algorithms to develop AI systems that can learn from trends and features in the data they analyze.
An AI system tests and evaluates its performance after each round of data processing to improve. AI is also programmed in a way in which they improve itself based on the information they collect.
Some Common Forms of AI We Come Across Very Often
- Voice Assistance –Functions like Siri, Alexa, Cortana, and other digital voice controllers are examples of AI, commanded to understand the user’s voice and do as directed. Voice assistant is rapidly growing because of how user-friendly it is. A huge data pool can be accessed in seconds, giving accurate results. Especially in the healthcare department, we can identify diseases with vocal biomarkers.
- OTT Platforms and Streaming Applications – Platforms like Netflix, amazon prime, and Hotstar are using this advanced technology to provide users with the best kind of content. AI customizes a list of new shows/movies that match a similar genre based on what the user has previously watched and searched. Not only this, but AI also does some backend work, where it allocates how much ever data would be required to stream the data uninterrupted.
- Self-Driving Vehicles – something that is coming about more commonly, self-driving vehicles require a lot more to be incorporated with AI – like DRL (Deep Reinforcement Learning), which shows how a machine runs independently. AI is advancing in ways not just to drive smoothly given the mapping of the road or be able to notice a blind spot, but also developments are ongoing for smart actions to prevent unforeseen events.
- Facial Recognition – probably the most well-known form of Artificial intelligence, along with Generative Adversarial Neural Networks (GANN), facial recognition functions go through trial and error to become error-free.
AI is used in several other places, be it on a small or large scale; it is an ever-growing form of technology.
Four Categories of AI
Type 1 – Reactive Machines
Here, the machine is programmed to make decisions in real-time only, and any memory does not influence decisions. It has no provision to store memory.
An example of one such machine is Deep Blue, which was created to play chess. The only storage it has are the rules of chess. Other than that, all new moves are based in real-time, not influenced by any previous memory.
The purpose of this AI is to analyze situations as is. However, in most cases, this design is not optimal, as it could be easily fooled as they can’t perform tasks other than what has been specifically assigned.
Type 2 – Limited Memory
These allow the machines to store some data in the memory. Self-driven cars are one such example. These allow the car to store data like traffic lights, modifications of the road, where the curve comes, sensors detecting how close another driver is or sensing the proximity of objects to prevent scratches, and so on.
These are stored for smooth navigation and drive. However, these are not stored in the car’s library, from where it can learn better.
For example, a person driving for many years becomes a better driver with practice. There is no such concept with AI in cars. It is and will be what it is unless it is manually programmed and modified to do better.
Type 3 – Theory of Mind
These are in the making, mainly those who think like humans in decision-making abilities. The humanoid robot – Sophia is an example here. Sophia has been developed in a way where she can portray some emotions already given to her as input and can speak and perform tasks freely on command.
Type 4 – Self-Awareness
There is much to learn about human intelligence before feeding it into a machine. It is yet to develop and is still in theory, but researchers are trying to create a robot with a conscience and the ability to be self-aware. It is an extension of the Theory of Mind, but still a long way to go because the human understanding of intelligence is still incomplete.
Now, coming to the main question:
Is AI Getting Smarter Than the Creators of AI?
Theoretically, this could be with the implementation of Self-awareness AI. They, in theory, are meant to be smart AI, smarter than the creators, but many new controversies would arise. Would the robots also have rights under the law? Will they feel and comprehend things?
Ensuring emotions and consciousness in a machine would not only be challenging for engineers and programmers but also philosophers. A human has many emotions and feelings, and every human is different. Now arises the question, who is to decide which robot would have what emotions and why? For us, it’s what we were born with, at the fault of nobody.
Despite many challenges, researchers everywhere are trying to build on the concept of a self-aware AI bot.
Researchers at Cornell University, New York, had published a paper that mentioned how some algorithms adapt to their intelligence and work terrifyingly. These programmed bots were created as solutions to daily problems; however, they came up with unconventional solutions, surprising the creators of AI themselves.
7 Crazy Examples of AI Out-Smarting Us
1. Crash Landing
Researchers devised one algorithm to learn how to land a plane gently. Therefore, the approach generated questions when the project started to bring the aircraft into the ground at full power.
However, the AI had learned that when the jet crashed, the force detected by the computer was so great that it exceeded the program’s capacity for memory. The outcome? It received a perfect score from the computer.
Naturally, the AI kept crashing the plane, so those who worry that AI will soon run the earth may find this rule-breaking disturbing.
2. The Invincible Game of Tic-Tac-Toe
This next report is less frightening but just as creative. In five-in-a-row rounds of Tic-Tac-Toe, when programmers pitted two algorithms against one another, one side came up with an unbeatable play.
The algorithm demanded actions that would cause the opponent’s system to crash by bombarding it with requesting moves with incomprehensibly huge memory locations. The first algorithm always triumphed by default since it had no competition.
3. The Mechanical Arm That Appeared to be Walking
In a different experiment, people asked a computer-simulated robot to walk without touching the ground. Most people considered it to be an impossible task until the program made a sudden turn.
The robot instantly flipped over onto its back and started walking with its feet blissfully floating above its head while standing on its elbows. Though few people would attempt it, walking on one’s elbows was as natural to AI as standing on one’s feet. The weirdest part is that They never programmed into the bot that walking on its elbow is a probable solution.
4. The High Jumpers
The previous study demonstrated AI’s superiority over humans. A creature that could learn to jump as high as it could be the goal of programmers.
However, the organism evolved into a top-heavy box resting on stick-thin poles instead of finding a better way to jump. It was inevitable that the form would topple, which caused the creature to spin in midair and, if only technically, register a high jump.
5. Microsoft Chatbot Supporting Nazi
In 2016, an experimental chatbot, Tay, started abusing hurls on twitting commenting on things and mentioning how Hitler was right and did good things. This chatbot was not completely doing something on its own but was mimicking the other comments read on Twitter.
6. A Feud Between Wiki Bots
Wikipedia has an army of bots that would do tasks such as scan a plethora of websites and match and alter data. However, if we gave two bots two different but similar jobs were seen to fight with themselves in an infinite loop.
7. Jailbreak
A Russian robot prototype named Promobot IR77 broke out of the lab where it was being built in an odd occurrence that made news worldwide.
According to accounts, an engineer left a gate open at the facility, allowing the robot—designed to learn from its surroundings and interact with people—to roll itself out into the city streets of Perm. The robot, which resembles a plastic snowman in some ways, walked into a congested junction, slowing traffic and alarming the local police.
According to lab officials, the event happened as the robot was studying navigation and obstacle avoidance. Promobot relished its brief experience of liberty. They said that the bot kept making a run despite reprogramming it twice.
Although these examples demonstrate how algorithms might abuse software flaws, the remedies developed by artificial intelligence tell a different tale.
AI may not be “smart” as we would define it. It has nevertheless proven that it can terrifyingly simulate evolution. And that type of genius is eerie.
When allowed complete freedom to find the best solution, programs may trick their designers in a brand-new, unanticipated direction. Exactly the kind of inventive mind that many people fear exists.
However, if correctly harnessed, programmers may exploit the inventiveness of AI to find ideal solutions that the human mind would never think of.
We Can Harness AI. How?
Each one is related to a segment of artificial intelligence known as generative design: a process that involves giving a program dozens of pieces of data and then instructing the software to create designs that fulfill particular requirements. In other words, you give AI an environment to work in but the liberty to interpret the inputs in any way it sees fit, which often leads to surprising results.
As the architectural firm Autodesk discovered, there are numerous advantages to utilizing this method. They gathered the employees and asked them questions like how much space they’d like, how much light they would prefer, and so on, based on which Autodesk AI produced hundreds of different blueprints.
There are also many more methods for humans to use AI’s intelligence, with the development of medicines being one of the more interesting areas.
Biologists think AI can model new medications’ possible effects without endangering patients. To one day create medications the way engineers create aircraft, they can accomplish this by testing tens of thousands of constituent changes in a biological simulator.
AI is smart. It has capabilities that no human possesses. However, it is a programmed creation, which is bound to have bugs now and then – and these bugs may just be doing things that outsmart us, humans. Research worldwide is going on to try and find ways to convert these bugs into real-time solutions for type 4, self-aware AI.
Conclusion
I would say not completely, it is yet a long way for AI to outsmart us, but AI has its own, really different way of thinking, which is of high added value for us humans. We can find new and innovative problem-solving approaches by asking a computer program. What would take a whole army of people to think of could be devised by AI in a matter of seconds, saving so much time.
Last Updated on August 28, 2023 by ayeshayusuf
