fbpx

What is AI?

Share on facebook
Share
Share on linkedin
Share
Share on twitter
Tweet
Untitled   px

What comes to mind when I mention Artificial Intelligence (AI)? It’s probably something to do with robots taking over the earth.
That was my first thought when I first started studying about it a few years ago. But when I learned more about how AI works on a programming and even mathematical level, I understood that it’s nothing like that, and yet so much more.
Artificial Intelligence (AI) is the science of creating machines that act like those in movies.

Let’s start with a clarification. Artificial General Intelligence is what most people think about when they hear AI (AGI). It is capable of performing all of the tasks that humans can, but at a higher level. However, the truth is that we aren’t even close to making one. Artificial Narrow Intelligence (ANI) is the form of AI that is now available. You teach it a few things, and it will train itself to do them extremely (and I mean extremely) well. If you give it a new assignment, though, it will fail terribly. (Believe me when I say this.) I’ve attempted it).

You’d have to write various AI algorithms into the same program to develop a program that can handle multiple tasks. For an AI that can theoretically take over the globe, this would entail coding a large number of AI algorithms, which would not only take a long time to develop, but would also require massive amounts of computing power to train and execute!

Is AI, however, any less amazing as a result of this? Not in the least! Even if we can’t create a program that can do everything well, we can still create programs that excel at a few tasks.

Artificial Intelligence (AI)

So, what makes AI so intelligent? Machine learning is a miraculous process.

The user explicitly inputs the pattern that the software should follow to make judgments in typical programs. The machine learning technique in AI enables the program to recognize patterns on its own.

Here’s an illustration.

Consider a dataset that asks the user to classify each day as freezing, cold, warm, or hot based on the average temperature on those days. If someone were to write the code themselves, it would be hardcoded so that if the temperature is 0 °C or lower, it will be classified as freezing, and so on.

In AI, on the other hand, the user would submit a dataset with temperatures and their accurate classifications, which would train the AI to identify the temperature ranges in each class on its own.

How?

By continuously updating the values of several constants, such as weights and biases, and then testing its prediction abilities with the current constants.

It’s like a ball rolling down a hill into a ditch. When it reaches the very top, it will begin to slowly roll down the slope. However, as it descends, the pace will increase until it reaches the very bottom. Because of its speed, it will continue on and may start heading up the hill on the opposite side. However, as gravity pulls it down again, it will slow and begin to descend to the bottom. It will keep doing this until it reaches the bottom of the ditch, gradually losing vitality.

The ball represents the value of the constants, while the bottom of the ditch represents their optimum value in AI. The constants will become closer to their optimum value as the program continues to train. However, if it continues in this direction, it will most likely exceed the ideal number. After then, the software would begin modifying the constants in the reverse manner. It would keep doing this until the ball landed on the bottom, at which point the constants would have reached their ideal value.

The weights and biases throughout the procedure are put into an equation (which changes from model to model) during the decision-making phase, which leads to a final result that the computer utilizes as its decision.

So, let’s go over everything again.

  • AI will not be able to take over the earth.
  • It is laser-focused on completing certain tasks.
  • It recognizes and learns patterns in its data through a process known as machine learning.

Let’s talk about how AI can be employed now that we have a basic understanding of how it works.

Applications

Artificial intelligence can now be used in almost every aspect of our civilization. This applies to everything from medicine to self-driving cars to military applications. Here are two AI-based applications that specialists are currently working on.

The Medical Prognosis

When you go to the doctor with a disease that isn’t easily identified these days, you can be referred to a blood-testing lab or a radiologist.

They would obtain a blood sample from you, as well as MRIs, X-rays, and other types of imaging of your body, before sending you home.

Your doctor would receive the image or test results in about one or two weeks, after which they would begin evaluating the information. The doctor would call you in about a week after that, with a probable diagnosis in mind.

In general, this is a waste of time. It might be fine for someone who has a broken bone or is mildly ill. However, in hospitals with critically ill patients, even a minute might make the difference between life and death.

Then there’s AI. What if, instead of taking a few days to get at a diagnosis, you could arrive at the same conclusion in just a few minutes? The program would save crucial time required to rescue the patient by swiftly detecting the root of the problem, allowing physicians to begin treatment sooner. This might include early cancer detection and therapy, or allowing surgeons to pinpoint the exact site of a tumor or injury in a patient’s body, which they can later heal with surgery.

Another issue that AI addresses is misdiagnosis. Today, 10% of all instances are misdiagnosed, with many of these cases leading to the patient’s death due to missed time or complications from the erroneous therapy. Medical mistake is the third biggest cause of death in the United States.

Artificial intelligence has the potential to substantially reduce the frequency of false diagnoses in patients, potentially saving millions of lives all over the world. It has been demonstrated to outperform clinical specialists in a variety of domains, including cancer diagnosis. As AI improves, it will soon be able to diagnose all ailments better than doctors.

“If [AI] isn’t currently the best diagnostician on the planet, it will be shortly” — once said by a renowned scientist

However, using AI to diagnose patients raises a slew of ethical concerns. First and foremost, what if the program is incorrect? So, who is to blame? What if it comes across something it’s never seen before? A surgeon would look for further information to help them make a diagnosis. AI would just label it as the most similar disease, with entirely different treatments that would be ineffective on the patient.

Finally, there is the issue of AI displacing humans. What’s to say it won’t soon entirely replace doctors if it can do far better than doctors?

AI cannot be incorporated into the medical environment until we have solutions to challenges like these. AI must be developed stronger and wiser at executing its purpose, and rules must be enacted to ensure that it does not take over our employment.

2. Self-driving cars

We’ve all seen movies in which a character gets into a car, tells it where they want to go, and the automobile drives them there on its own. The day when a regular individual can do the same isn’t far away.

Self-driving cars are already being developed and tested by a number of firms, including Uber, Google, and Tesla. Tesla claimed this year that it has successfully built completely self-driving cars, and that all Tesla cars released after this year would include the self-driving system.

How does self-driving work, though? When creating a self-driving car, the key goal is to establish an AI that can perform numerous jobs at once:

Connect to a GPS to find the quickest route to your destination.

Pay close attention to the various driving laws (stopping at red lights and stop sign; slowing down in neighbourhoods, etc.)

Be aware of its surroundings. When pedestrians cross the street, where other cars are, how far away the stop sign is, and so on, the AI must be aware.

In perilous situations, make swift decisions.

It’s tough to create an AI that can properly execute all of those duties, which is why fully autonomous systems have yet to be implemented.

Along with the challenge of creating such an AI, there are also ethical concerns. First, AI isn’t powerful enough to execute these duties at this time. Numerous incidents have occurred in which self-driving cars have made poor decisions, putting their owners in dangerous situations. Joshua Brown, 40, died in a deadly crash in May 2016 when his Tesla self-driving system failed to stop and collided with a trailer crossing the road.

Similarly, several people have expressed concern about the systems’ lack of emotional expression. Consider the following scenario. You’re driving your self-driving automobile when the truck in front of you abruptly comes to a halt. You’re going too fast to stop, and a family of pedestrians is standing behind you on the sidewalk. What does the AI do in that case? Is it going to swerve into your family in order to save your life? Or does it sacrifice you, the owner, in order to save the family’s life?

It will be impossible to build entirely dependable self-driving cars until we develop systems that can deal with scenarios like these, and drivers must be prepared to assume physical control of the vehicle at any time.

AI has a plethora of diverse uses, ranging from space exploration to quantum computing to business management.

Conclusion

  • Last but not least, here’s a recap:
  • AI will not be able to take over the earth.
  • AI is only capable of completing specified jobs.
  • It learns how to make decisions via machine learning.
  • It can be used in medical diagnostics to reduce the number of false positives and the amount of time it takes to diagnose a patient.
  • AI can be used to make self-driving automobiles entirely autonomous and dependable.

There are still a slew of ethical challenges surrounding AI, such as their lack of emotion and lack of problem-solving experience.

As a result, AI is poised to transform the world in a variety of ways, with various societal implications (except for taking over it). As machine learning models get more sophisticated, AI will eventually be able to complete more difficult jobs on its own.

Share on facebook
Share
Share on linkedin
Share
Share on twitter
Tweet

Related Posts

Authors

IMG
Hannah
a
Jin

About SCC

The importance of having fun and progress in tech education for all our students is our motto. We want our students to create a better future not only for themselves but also for society. Whether it’s programming their own videogame, animating their own cartoon, or building a robot, our tutors can guide them to find newer outlooks and explore unfound discoveries through our courses.