Artificial Intelligence

3. AI - exciting for some, scary for others

AI is currently being used in some exciting and life changing ways, however it does come with some ethical and industrial concerns. As AI has rapidly been researched and developed, a number of ethical dilemmas have arisen. Some examples are:

Listen to the Queensland Brain Institute (QBI) podcast that discusses What is AI and will it take over the world? with Geoff Goodhill, a professor at both The University of Queensland’s School of Mathematics and Physics and the Queensland Brain Institute.

 

Self-driving cars

One of the larger concerns with self-driving cars is trying to decide or decipher decisions the cars will make in certain situations. The Trolley Problem is a philosophical problem that has been pondered for decades that offers some insight into the types of concerns self-driving cars are creating.

“An out of control trolley is travelling down the track. It has no brakes and cannot be stopped. The track splits and there is a lever which controls the track the trolley will take. On one track there are 5 workmen and on the other is a friend or relative. If you pull the lever you save the 5 workmen but doom your friend or relative. Do you pull the lever?”

Amazon jobs AI

Amazon created an AI to help narrow down applications for jobs they had advertised. The AI was developed to take 100s of resumes and reduce it down to the top 5 for recruiters to interview. However, before being fully implemented and after various test runs, the AI was found to have a gender bias towards males due to the data it was using: the last 10 years of successful hires. Since there was a male bias in the technology industry the AI was unfairly undervaluing women, and anyone with a mention of woman in their resume.

The AI simply scanned the data it was given for patterns and similarities and applied the lessons learnt to its task, however it lacked the nuance or rules for not valuing based on gender etc.

Unemployment – automatic tellers, chatbots etc

AI being able to undertake various tasks has the potential to replace a human completing those same tasks. This creates an uncertainty in some industries around their future as AI could potentially replace the need for that task or industry.

Fake news generator

OpenAI have developed a model called the GPT-2 which can generate coherent paragraphs of text unsupervised. The paper released discussing its uses has a number of samples within showing its capabilities. 

For a very detailed look into how the model works, read the full paper, Language models are Unsupervised Multitask Learners, published by OpenAI staff Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei and Ilya Sutskever

The samples illustrate how the model can be given a sentence or two, and then produce completely made up stories with consistent details from the given information.

One of the samples of the GPT-2 model generating paragraphs of text from a small sample:

Context (human-written): In a shocking finding, a scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English.

GTP-2: The scientist named the population, after their distinctive horn, Ovid's Unicorn. These four-horned, silver-white unicorns were previously unknown to science.

Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.
Dr. Jorge Péprez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow.

Pérez and the others then ventured further in to the valley. "by the time we reached the top of one peak, the water looked blue, with some crystals on top," said Pérez.

Pérez and his friends were astonished to see the unicorn herd. These creatures could be seen from the air without having to move too much to see them - they were so close they could touch their horns.
While examining these bizarre creatures the scientists discovered that the creatures also spoke some fairly regular English. Pérez stated, "We can see, for example, that they have a common 'language', something like a dialect or dialectic."

Dr. Pérez believes that the unicorns may have originated in Argentina, where the animals were believed to be descendants of a lost race of people who lived there before the arrival of humans in those parts of South America.

While their origins are still unclear, some believe that perhaps the creatures were created when a human and a unicorn met each other in a time before human civilization. According to Pérez, "In South America, such incidents seem to be quite common."

However, Pérez also pointed out that is likely that the only way of knowing for sure if unicorns are indeed the descendants of a lost alien race is through DNA. "But they seem to be able to communicate in English quite well, which I believe is a sign of evolution, or at least a change in social organization," said the scientist.

AI text-adventure game

Nathan Whitmore, a graduate of Northwestern University, has used GPT-2 to create a text adventure game that creates itself as you play it. You can read more about Nathan's work and play the game on his tumblr website, Quick, to the rat cave!

Image generator

Nvidia has produced a generative adversarial network (GAN) that generates entirely new images that appear real. To see this in action go to thispersondoesnotexist.com, every time you refresh the page you will be greeted with a brand new image produced by the generator.

A Style-Based Generator Architecture for Generative Adversarial Networks (YouTube, 6m17s) illustrates and describes how a GAN takes the images of three people to generate a completely new image.