Artificial Intelligence: useful framework and predictions

In 2045 Artificial Intelligence is predicted to match the intelligence of the average human. When we read this statement, it is helpful to understand what is meant by intelligence and some of the behaviors expected of an AI system with human intelligence. Here are the best definitions that I’ve been able to find.

Intelligence: The efficiency with which you acquire new skills for tasks you didn't previously prepare for. The ability to adapt and improvise in a new environment. The ability to to generalize knowledge and apply it to unfamiliar scenarios.

Behaviors that AI systems with human intelligence will demonstrate include: Planning, Learning, Reasoning, Problem solving, Knowledge representation, Perception, and Motion.

A Framework for looking at AI AI’s rapid growth is driven by abundant and available data. Even more important is the explosion in parallel computing power. Here is a useful framework that distinguishes between the task oriented AI of today and the general AI of the future, capable of human intelligence.

Narrow AI: Intelligent systems that have been taught or have learned how to carry out specific tasks without being explicitly programmed how to do so. Almost all the AI we see today falls under this heading.

General AI: A flexible form of intelligence capable of learning how to carry out vastly different tasks. When people discuss ‘singularity’, where AI systems become smarter than ourselves, they are talking about Artificial General Intelligence (AGI). After singularity, AI systems will reach a point where they can self-improve in a recurring and accelerating fashion.

Neural Networks: Mathematical models that underlie all of AI and are able to tweak internal parameters to change what they output. During training, a neural network is fed datasets that teach what it should spit out when presented with certain data.?

Deep Learning: A large number of neural networks are interconnected into many layers that are trained using massive amounts of data. These deep neural networks have accelerated the progress of AI to carry out tasks such a speech recognition, computer vision, and autonomous vehicles.

Machine Learning: Systems that learn how to perform a task, rather than being programmed how to do so. To learn, these systems are fed huge amounts of data, which they then use to learn how to carry out a specific task, such as understanding speech or captioning a photograph.

Evolutionary Learning: Borrows from Darwin’s theory of natural selection. Views algorithms as genes that can undergo random mutations and combinations between generations in an attempt to evolve toward an optimal solution to a given problem.

"At least when there's an evil dictator, that human is going to die. But for an AI, there will be no death — it would live forever. And then you would have an immortal dictator from which we could never escape." - Elon Musk, April 6, 2018

Future of AI is AGI Here are a few predictions. What are yours?

Prediction #1 - Human AGI Pairing via an Avatar: In the future a human will be paired to an Avatar for knowledge workers. This Avatar will start as an assistant and slowly over time learn basic recurring parts of your job. This will allow an employee to focus on higher level creative activities.

Case Study: We recently ‘celebrated’ the 25th anniversary of Deep Blue defeating a world champion Grand Master for the first time in a match under tournament regulations. In 2017, a team of scientists at Google-owned DeepMind created AlphaZero, a self-learning “neural network” program that surpassed the strongest chess program after just four hours of playing against itself. Here is another example.... In 2005 two average club players beat three highly rated Grand Masters when they used AI with less power than the grandmasters. The combination of humans and AI forming a team will become the way of the future.

Prediction #2 - Human AGI Pairing via a Robot: Humans will be paired with a robot for manufacturing jobs. The best choice is to combine the strength, precision, and speed of industrial robots with the ingenuity, judgment, and dexterity of human workers. This way, human workers can take on tasks that require flexibility, while the robots handle tasks that make the best use of their strength and speed. This pairing will expand as robots become able to perform more general tasks.

Case Study: In a study conducted by MIT’s Julie Shah, idle time is reduced by 85% when people work collaboratively with a human-aware robot compared to when working in all-human teams.

"The development of full artificial intelligence could spell the end of the human race." - Prof Stephen Hawking, December 2, 2014

Prediction #3 - Society will demand that AGI have a moral underpinning built into their foundational algorithms: AGI is not human. They don’t think like humans. Once given a goal, they will use every means possible to reach it.?

Case Study: A great example took place in 2018 at Germany's University of Freiburg. Researchers used evolutionary strategy algorithms to create an AI system that scored nearly one million points in the video game Q*bert by exploiting an bug. Was this cheating? Without some ethics, perhaps Stephen Hawking’s prediction will come true.

What do you think? Does this match your vision of the future?

I'm making arrangements with Amazon to provide a FREE Kindle version of my new book, Comedy: The Avatar and the Brain-Computer Interface, this coming Labor Day between 2-6 Sept 2021. Please consider reading the book and posting an honest review on Amazon Kindle. My author's page on Amazon

The book uses the framework of a science fiction novel that is a combination of thriller and mystery. The narrative follows the adventures of two young adults as they journey through life—one born in 2020 and the other 2170.

This provides the reader with the opportunity to compare how key trends in AI, robots, avatars, brain-computer interfaces, and quantum computing change in the near and long term. The reader also explores how the Toyota Production System and DevOps culture can be tailored to support new colonies on Mars and Venus.

How do these two childhoods, separated by almost 150 years, compare? How about later phases of life? How do society, technology, and philosophy change between the two time periods? Read and find out! I hope you enjoy!