Category: ai

TOO LATE

Ideally, public good and ethics will be taught in advance of the melding of technology and government, i.e., not after mistakes are made. However, history suggests humans will blunder down the road of experience with A.I., making mistakes, and trying to correct them after they occur.

Books of Interest
 Website: chetyarbrough.

The Technological Republic (Hard Power, Soft Belief, and the Future of the West)

By: Alexander C. Karp and Nicholas W. Zamiska

Narrated By: Nicholas W. Zamiska

Alex Karp (Founder of software firm Palantir Technologies)
Nicholas Zamiska (CEO Palantir Technolgies

The authors are the founder and operations manager of the American software company, Palantir Technolgies. Palantir has been hired by the U. S. Department of Defense, the Intelligence Community, agencies of NATO countries, and Western corporations to provide analytic platforms for defense analysis, healthcare, finance, and manufacturing.

They believe artificial intelligence research and development has lost its way.

They argue Silicon Valley has lost focus on what is important for survival of society and Western values. They suggest A.I. should be focusing on serving humanity in ways that responsibly regulate nuclear weapons and protect society from existential risks like climate change, pandemics, asteroid collisions, etc., that threaten human extinction. The authors provide a powerful criticism of technology and its national purpose.

Karp and Zamiska argue that technology is focusing on consumerism rather nuclear annihilation or existential risk.

By focusing on convenience and entertainment for financial success, fundamental problems like the threat of nuclear war, homelessness, inequality, and climate change are ignored or relegated to the trash heap of history. (“Trash heap of history” is the belief that what happens, happens and society can do nothing about it.) The west has become complacent with short-term focus on profit and consumer demand. The authors argue that the greater good is no longer thought of as an important societal goal. The primary goal is making money that enriches creators and company owners by making purchases more convenient to and for consumers.

Aldous Huxley (English writer and philosopher, 1894-1963, author of “Brave New World”.)

Their argument is there should be more collaboration between tech and government. Historically, government is only as good as the information it has to make societal decisions. A computer program can be programmed with false information like the error of weapons of mass destruction that led to an invasion of Iraq that was a bad decision. The domino theory input that led to the Vietnam war; so, on and so on. There is also the threat of an elected President that uses the power of technology to do the wrong thing because of his/her incompetence. There is the risk of government gathering personal information and using it to cross the line into a “Brave New World” where innovation, free thought, and independent action are discouraged or legislated against so people can be sent to jail for breaking the law?

Possibly, melding technology with government is an answer, but it is a chicken and egg concern. Education about public good and ethical practices should begin as soon as the egg cracks, not after hatchlings are already old enough to work. Phrases that come to mind are “What’s done is done” or “The die is cast”.

The authors argue the West needs to up-its-game if it wishes to create a peaceful and prosperous future for a society that is founded on the ideal of human freedom.

Without future generations creating policies based on ethical purpose for the public good, one infers western culture will spiral into individual isolation and self-interest that diminishes western culture and ideals.

Ideally, public good and ethics will be taught in advance of the melding of technology and government, i.e., not after mistakes are made. However, history suggests humans will blunder down the road of experience with A.I., making mistakes, and trying to correct them after they occur.

AI REGULATION

As Suleyman and Bhaskar infer, ignoring the threat of AI because of the difficulty of regulation is no reason to abandon the effort.

Books of Interest
 Website: chetyarbrough.blog

The Coming Wave

By: Mustafa Suleyman with Michael Bhaskar

Narrated By: Mustafa Suleyman

Mustafa Suleyman (British entrepreneur, CEO Microsoft AI)
Michael Bhaskar (New York Times writer & publisher, works at Microsoft AI.)

This is a startling book about AI because it is written by an AI entrepreneur who is the founder and former head of applied AI at DeepMind. He is also the CEO of Microsoft AI. What the authors argue is not understood by many who discount the threat of AI. They explain AI can collate information that creates societal solutions, as well as threats, that are beyond the thought and reasoning ability of human beings.

“The Coming Wave” is startling because it is written by two authors who have an intimate understanding of the science of AI.

They argue it is critically important for AI research and development to be internationally regulated with the same seriousness that accompanied the research and use of the atom bomb.

If AI has the monumental risk Suleyman and Bhaskar describe, it remains the responsibility of humanity to manage it wisely.

Those who have read this blog know the perspective of this writer is that AI, whether it has greater risk than the atom bomb or not is a tool, not a controller, of humanity. The AI’ threat example given by Suleyman and Bhaskar is that AI has the potential for invention of a genetic modification that could as easily destroy as improve humanity. Recognizing AI’s danger is commendable but like the atom bomb, there will always be a threat of miscreant nations or radicals that have the use of a nuclear device or AI to initiate Armagedón. Obviously, if AI is the threat they suggest, there needs to be an antidote. The last chapters of “The Coming Wave” offer their solution. The authors suggest a 10-step program to regulate or ameliorate the threat of AI’s misuse.

Like alcoholism and nuclear bomb deterrence, Suleyman’s program will be as effective as those who choose to follow the rules.

There are no simple solutions for regulation of AI and as history shows neither Alcoholics Anonymous (AA) nor the Treaty on the Prohibition of Nuclear Weapons (TPNW) has been completely successful.

Suleyman suggests the first step in regulating AI begins with creating safeguards for the vast LLM capabilities of Artificial Intelligence.

This will require the hiring of technicians to monitor and adjust incorrect or misleading information accumulated and distributed by AI users. The concern of many will be the restriction on “freedom of speech”. Additionally, two concerns are the cost of such a bureaucracy and who monitors the monitors. Who draws the line between fact and fiction? When does information deletion become a distortion of fact? This bureaucracy will be responsible for auditing AI models to understand what their capabilities are and what limitations they have.

A second step is to slow the process of AI development by controlling the sale and distribution of the hardware components of AI to provide more time for reviewing new development impacts.

With lucrative incentives for new AI capabilities in a capitalist system there is likely to be a lot of resistance by aggressive entrepreneurs, free-trade and free-speech believers. Leaders in authoritarian countries will be equally incensed by interference in their right to rule.

Transparency is a critical part of the vetting process for AI development.

Suleyman suggests critics need to be involved in new developments to balance greed and power against utilitarian value. There has to be an ethical examination of AI that goes beyond profitability for individuals or control by governments. The bureaucracies for development, review, and regulation should be designed to adapt, reform, and implement regulations to manage AI technologies responsibly. These regulations should be established through global treaties and alliances among all nations of the world.

Suleyman acknowledges this is a big ask and notes there will be many failures in getting cooperation or adherence to AI regulation.

That is and was true of nuclear armament and so far, there has been no use of nuclear weapons to attack other countries. The authors note there will be failures in trying to institute these guidelines but with the help of public awareness and grassroots support, there is hope for the greater good that can come from AI.

As Suleyman and Bhaskar infer, ignoring the threat of AI because of the difficulty of regulation is no reason to abandon the effort.