STEM Innovation:
Artificial Intelligence

Artificial Intelligence (AI) is an example of how combining the knowledge of STEM subjects creates world changing innovations. AI might seem futuristic, but you will be surprised to learn how much it has already impacted and changed our lives. Below is a collection of facts, videos, programs and simulations STEM World uses to explain some interesting and mind-blowing components of Artificial Intelligence.

AI in Real Life

Artificial Intelligence is a very new idea compared to some scientific creations, but in a tech savvy world, companies have embraced this innovation and use AI wherever possible. From using facial recognition to unlock your phone to suggesting similar pages from your Google search, we are already surrounded by AI. Watch this video showing 10 examples of how AI is already part of your life.

The First AI Program

The original concept of Artificial Intelligence can be traced back to the 1950’s when scientists began to consider the possibility of giving machines the intellectual capabilities to mimic humans.

The first computer program to perform automated reasoning was called the Logic Theorist which proved mathematical theorems like a real-world mathematician. The scientists behind this program (Allen Newell, Herbert Simon and Cliff Shaw) created it through writing a new programming language called IPL. This code would later form the basis of McCarthy’s LISP programming language, an important language still used by AI researchers today.

Below is a timeline of AI showing some major milestones:

AI in the Media

As you can see from the AI timeline, there are many milestones in AI achievement and some created worldwide news. Like all publicity, AI receives both good and bad press and the long term implications of this technology it is still widely debated.

A large achievement for AI was in 1997 when an IBM supercomputer called Deep Blue beat the world’s best chess player. This achievement was ground-breaking news because it was one of the first situations where a computer could outthink a human at a very strategic game. Watch the story behind the programmers who made Deep Blue and human who was beaten.

One more negative use of AI programming which emerged in the last decade are deep fakes. Programmers altered facial recognition software to create fake videos of celebrities. Programmers can create videos showing celebrities saying and doing things which never actually happened. This use of facial recognition has already caused issues because some people think the footage is real. Watch one of the most famous deep fake videos and see if you can tell the difference.

Cognitive Skills

Let us discuss the fundamental abilities an AI program require to mimic a human. Any program needs to begin with basic code and concepts before more complex algorithms can be successful. Just like a baby, artificial intelligence needs to learn basic skills to perform human tasks. These skills are classed as Cognitive Skills and include; reading text, sound recognition, understanding images, indexing videos and performing facial recognition. (Cognition: The mental action or process of acquiring knowledge and understanding through thought, experience, and the senses.)

Microsoft has Cognitive Skill programming code available to create the building blocks for a new AI program. Students at STEM World can test online versions of their code and can get a better understanding of each specific skill. Does the AI perform like a human when completing each of these tasks?

Example of analysing text: https://aidemos.microsoft.com/text-analytics

Example of understanding images: https://aidemos.microsoft.com/computer-vision

Example of indexing videos: https://aidemos.microsoft.com/video-indexer

More Than Just Code

As mentioned, AI programs start with basic cognitive skills and then evolve by using more complex logic. The goal of most programs is to teach AI the human skill and then for the program to build on these abilities and provide more advanced results. See links below to online applications to experience how AI logic is advancing. You may notice that these applications aren’t perfect (they can break) but STEM experts are making many innovative discoveries in AI.

Image Identification: Pix2Pix is an AI application that uses image understanding logic and to turn line drawings into real life images. Some creations look more into monsters than lifelike drawings but who should you blame – the AI or the artist?

Have a go – https://zaidalyafeai.github.io/pix2pix/celeb.html

Text Analysis: Talk To Transformer is a recently created pre-emptive text generation application using OpenAI software. Not only does this AI read and understand what you have written but will finish your paragraph. This program is the next step in AI communication with humans. Do you think the AI sentences sound human?

Give it a try – https://talktotransformer.com/

Ethics of AI

The idea of teaching computers to think for themselves creates many questions and concerns for humankind. Hollywood has negatively impacted our views towards AI by making many movies where machines take over the world. Therefore, one very important component when making a new artificial intelligence program is ensuring it will operate ethically.

There are many ethical values that must be considered and included in an AI program. The ethical standards all AI programs must pass include fairness, reliability, safety, privacy, security, inclusiveness, transparency and accountability. Setting ethical standards doesn’t stop some people from using AI for harmful results and deep fakes are one example that has caused governments to review regulations around AI technology. Like all technological breakthroughs AI has many positive outcomes but without the proper regulation it could assist in creating catastrophic consequences in the world.

Testing Morality

Computers can be coded to read text or understand speech, but can we teach it morality? Creating self-driving cars is an AI innovation that many companies are trying to achieve, but there are many ethical conundrums generated from giving machines control of our vehicles. Human moral is a component of ethics that can differentiate between each person, so this creates much complexity when setting rules for a computer.

There is an interesting ethical decision-making simulation which tests your morals in fake rare driving situations. We often ask students in STEM World to complete this test. Test and discover where your morals lye by completing this simulation. Compare the end results with your friends and see how morality can alter between each person.

Who will you choose to save?

> http://moralmachine.mit.edu/ <

Summary

Artificial Intelligence is a technological advancement which is driving our future. AI is the perfect example of how combining the STEM subjects creates new and innovative careers. This is why STEM World challenges students with understanding and developing AI concepts. Is there any task in your life that you would want computer could complete, possibly at a faster rate? It could happen. Maybe you will be the programmer who designs and creates this application.

There are many resources online if you want to learn more about AI and the specific topics in this article. If you want to read more, below are links to papers and websites which were referenced when creating this article.

http://www.kasra.com/history.html

https://history-computer.com/ModernComputer/Software/LogicTheorist.html

https://www.brookings.edu/series/a-blueprint-for-the-future-of-ai/

https://towardsdatascience.com/ai-machine-learning-deep-learning-explained-simply-7b553da5b960

https://www.saasworthy.com/blog/talk-to-transformer-all-you-need-to-know/

https://futureoflife.org/ai-principles/

https://blogs.microsoft.com/eupolicy/artificial-intelligence-ethics/