In his book, Tools and Weapons, Brad Smith lays out a series of important considerations for the continued adoption of artificial intelligence into an ever increasing amount of applications and daily uses. While artificial intelligence has historically held a rather loose definition, Brad Smith quotes Dave Heiner of Microsoft who describes AI as “a computer system that can learn from experience by discerning patterns in data fed to it and thereby make decisions” (Tools and Weapons pg. 194). This ability to learn based on experience and modeling is integral to creating flexible systems that can react to their environments in meaningful ways.
AI has recently been emboldened by a number of technological advances. This includes massive advances in the scalability and accessibility of computing power. Commonly this exists in the form of a datacenter as part of the cloud which doesn’t require large, initial capital investments. This increase in centralized data storage and computing power has also provided for a deep array of data from which to train models. These advances have come together to allow the pace of development for artificial intelligence to increase rapidly.
Brad Smith covers the advances and his proposed guidelines for AI across 3 primary chapters. I’ve written some of my general thoughts on each below.
AI and Ethics: Don’t Ask What Computers Can Do, Ask What They Should Do
If you’re driving down any mountain road in the United States or many places in the world, you’re likely to see a metal railing running along potential drop offs and dangerous areas. In an ideal situation these guardrails should never come in contact with your car as it moves and steers down the mountain. However, in very rare cases, a car can experience a mechanical problem or a driver can lose focus. When this happens the guardrails are there to guide the car away from a cliff or hazard and give the driver time to get the car under control, slow down, and stop to assess.
As we continue to ramp up the speed of the innovation and development of artificial intelligence, it will be equally important to develop guardrails to keep us on track. Technology can be used for a variety of uses but it needs to be pointed in a direction and given a use. Oftentimes, we focus on the rapid development of a new technology because it is possible and not always based on its appropriate or ethical use. Purposeful development of artificial intelligence, with the appropriate guardrails in place, will allow us to continue to fill needs and innovate while protecting those things that are important to the success of humanity and individuals.
Microsoft has put together 4 basic principles for AI development in order to avoid potential pitfalls. These are:
- Reliability and Safety
- Privacy and Security
- User Inclusiveness
These 4 principles help to provide a basis for development while attempting to respect the human rights across all groups. While simple in list form, each of these principles are deep in their application requirements and considerations. Further considerations will also need to be made on additional guiding principles and development considerations. Brad describes Microsoft’s past experience with determining these factors:
“As computers gained the ability to make decisions previously reserved for humans, virtually every ethical question for humanity was becoming an ethical question for computing. If millennia of debate among philosophers had not forged clear cut and universal answers, then a consensus was not likely to emerge overnight simply because we needed to apply them to computers.” - Tools and Weapons pg. 199
In order to further determine what to use as guardrails and driving principles for the creation of artificial intelligence, humanity will continue to need to determine and address what it collectively holds most dear. Only then can we move forward with confidence into a world filled with automated decision making.
AI and Facial Recognition: Do Our Faces Deserve the Same Protection as Our Phones?
The human body is an incredible creation. It is universally identifiable, yet individually unique. With many unique, yet similar features, there exists the capacity for humans and machines to identify at the level of an individual. The face is an important identifying characteristic and serves as the viewport to the world for a multitude of senses.
Many innovations have already taken advantage of the unique nature of the human face. One of the most widely spread uses of this so far has been in the world of biometric digital security. It is now possible on many devices to simply smile (or not) at the camera and unlock your digital rights of access. Where this has provided a new level of convenience, it has also raised interesting questions and concerns around the ownership of these new data sets and who has the justification to employ them.
Other innovations have been employed to recognize the face as part of scalable applications related to everything from automatic photo tagging and sharing to the identification of individuals as part of investigations. There are some key components to the questions of not just legality but the ethics behind these and other applications of facial recognition technology. When the ability to stay a private citizen is lost it will be incredibly hard to attain again.
AI and the Workforce: The Day the Horse Lost Its Job
Throughout time, humanity has worked to automate the mundane to enable it to focus on greater pursuits. This has almost always involved the strategic application of technology and processes. This was seen in the early days of humanity, when the transition from hunting and gathering to agricultural production was underway. It has been seen in more recent times with the replacement of horses for cars, and in the modern day as the communication of information has evolved and changed to focus on accessibility of access and creation. This period of transition that occurs time and time again has usually ended in the benefit of the human race, however these transitions are not without negative impact, sometimes even drastically so.
Artificial Intelligence has a unique potential to generate not just one, but many transitions for the workforce. It will affect many, if not all, lines of work in some way. Some current jobs stand to be outright replaced, where many are likely to be modified with a focus on strategic decision making and communication. Brad points out that pay scale does not have anything definitively to do with the risk of replacement or modification. While lower skilled jobs such as the order attendant or cashier at a store are potentially at risk, so are higher paying jobs like semi-truck drivers and radiologists. AI has the unique potential to affect jobs and life at all levels and should be given careful consideration for each potential application to ease the transition.
There will continue to be a need to provide better educational opportunities and clarity of job positions in order to adequately support humanity through the transitions to come. When looking at what these transitions could potentially mean for Microsoft, Brad Smith recounted the following important principles:
“As we thought about what this meant for our own products and future at Microsoft, we concluded that success has always required that people master four skills: learning about new topics and fields; analyzing and solving new problems; communicating ideas and sharing information with others; and collaborating effectively as part of a team.” - Tools and Weapons pg. 247
These areas, and others, will continue to be important to focus on to limit those left behind by change and lift up humanity together as Artificial Intelligence adoption continues to ramp up its prevalence.
These questions are at the core of the continued rapid innovation of Artificial Intelligence and there will be many more. It will be critical to pay attention to not just the how but also the why of the application of these new technologies. Without it, many important freedoms and opportunities will run the risk of being eroded away without a likely potential of attaining them again. It is up to governments, companies, and individuals worldwide to consider the complex and diverse ramifications and to make sure that those who are potentially at risk are not taken advantage of or left in the proverbial dust. Only then can humanity confidently take steps forward into the exciting and terrible world of technology making decisions by and for itself.
Find the Book Here: Tools and Weapons by Brad Smith