Note: This article was originally published at Thought Stack.
When you hear the term Artificial Intelligence, what’s the first image that pops in your mind? Is it a Terminator-style doomsday scenario? Or is it something more helpful like how J.A.R.V.I.S. enhances Tony Stark in Iron Man?
Right now in the tech community people falling into two camps, those who are worried about the dangers and those who are more focused on the potential benefits. Individuals such as Elon Musk and Stephen Hawkings believe that AI is the greatest threat to humanity, whereas, Jeff Dean and Demis Hassabis from Google believe the benefits outweigh the risks.
If you don’t want to take a person’s bias at face value, it’s important to go to the source material yourself from which the industry started. You have to look at the work by Alan Turing.
In this article you’ll learn three things:
- Who is Alan Turing and why is he important for AI?
- What are some of the limitation in developing an AI platform?
- How will AI benefit our lives in the near future?
The Bible for Creating Intelligent Machines
To help us better understand AI, I pulled in an expert to select a book for us to devour. Dr. Shalini Ananda holds a Ph.D. in material science and an extensive background in pure math, which is why she chose The Essential Turing. She goes one step above just recommending any book, she chose this one because it’s been her “Bible for creating intelligent machines.”
Why Does Turing Matter?
Alan Turing is known as the father of the Information Age and is known for the Turing Test, the test for determining if an artificially intelligent agent is actually intelligent.
The premise of the Turing Test is to use advanced mathematics to create a machine that could convince a human judge that it was a human. The test was set up in the 1950s by having a human judge chat to an unknown entity where they asked it a series of questions over a period of time and guessed if they were talking to a human. Turing believed that if 30% of the human judges talking to the machine believed it was human, then the machine could be considered intelligent.
Today, these specifications have changed a bit requiring 50% of the human judges to be convinced chatting to the entity for anywhere between 25 minutes to two hours before determining if the entity was a human or not.
Although the Turing Test is what Turing is famous for, it’s important to understand that his work goes deeper, describing the method and the math needed to create intelligent machines. He was a genius way ahead of his time.
Does Software Get Slower Over Time?
Dr. Ananda mentions that there is an adage in computer software that software gets slower more rapidly as hardware gets faster…
This counterintuitive tendency is called Page’s Law (originally Wirth’s Law), named after the Google co-founder, Larry Page.
With the advancements in machine learning algorithms, by way of sheer computing power to operate them, software is getting significantly slower. This has forced hardware companies such as Intel, Nvidia, Qualcomm, and even Google to develop special chips to speed up AI algorithms.
Dr. Ananda goes on to state that,
as we create intelligent machines we certainly need advancements in hardware to support the demands of the algorithms.
What Are the Benefits of AI in Industry?
Marvin Minsky, an early AI researcher, defined AI as the science of making machines do things that could be intelligent if done by people.
That means the application of AI is only limited to our imagination for potential use cases. Right now you can see the tech industry focusing on applying AI to replace all of the repetitive activities performed by humans in the hopes to reduce operating costs and increase the throughput of services.
Although this progress tends to leave people looking for a new job, it also gives them the potential to work in new industries or apply the one thing computers aren’t good at yet, being creative.
What Goes Into Developing a Machine Learning Platform?
Dr. Ananda states,
working on machine learning algorithms is about understanding the mathematics as Turing described and applying them to abstract concepts. In other words, you’re asking questions about how people learn and finding ways for computers to do that in the most effective way possible.
The Enterprise Challenge
Many large companies are scrambling to figure out how to apply AI technologies within their organizations to stay competitive with the juggernauts such as Amazon and Google.
Having worked with many enterprises in implementing AI technologies, Dr. Ananda states,
the main challenge for enterprises is that many don’t have a pre-built foundation for recording and streamlining the flow of data required to complete a machine learning task. For some companies, this means working backward and setting up a data capture and recording system before even attempting to implement AI tools. Each algorithm is created for a specific use case. You can’t just throw open source tools commonly available and hope to stay competitive in the market.
As The Essential Turing describes, it’s important to first define the artificial agent, which could be as simple as a single sensor or complex like a robot. Next is to define the dynamic conditions around the intelligent agent which determine the set of operators direct the performance of the artificial agent in response to an environmental change.
The challenge in creating these systems is determining how to represent the agent, its state of being, the sensors around it and the selection of learning algorithms and the sequence in which these algorithms are employed.
Dr. Ananda goes on to state,
this is why I believe a good developer of AI technology must have an intuitive knowledge of the mathematics behind it. Most developers tend to use open source tools that were build to solve a different problem to the one they need to solve and if they don’t have the intuition on where the results actually come from, they are setting themselves up for failure.
Companies such as Amazon and Google are clamoring to be integrated into your life. You can see it with the Amazon Echo and Google Home, both are using audio sensors to learn what you like to hopefully interject ways to modify your life and buying habits.
Even more so, smartphones, which have audio, images, video, location, and other sensor are the perfect devices for gathering context around who you are as a person. Imagine the influence these companies will have once everyone has a smartphone that is connected to their services.
How do you feel about having this these types of intelligent machines influencing your daily life?
As these types of technologies become more prevalent in our lives, it’s important we have the discussion of how we want them to benefit us while reducing the potential negatives. Let me know your thoughts in the comments below.
Be sure to reach out to Dr. Ananda on Twitter at @ShaliniAnanda1.