Responsibility in Artificial Intelligence

An opinion piece by a data scientist

April 28, 2024

Artificial Intelligence | Technology | Philosophy

An opinion piece by a data scientist

Artificial Intelligence, for many, is equal parts exciting and terrifying. Rapid innovations in AI are allowing for revolutionary new products and services which have a genuine potential to improve our daily lives. Also, AI is posed as a destructive force which has the power to destabilize job markets and be used to further manipulate the masses.

Everyone wants a bright future, but whos job is it to make that happen? That’s what we’ll be exploring in this article.

AI is a Tool

AI is not intelligent, it is not a godhead, it is a tool. Thinking otherwise confuses one’s ability to effectively reason about AI.

To understand what we should do with this new tool, I think it’s useful to study the past. In studying the past we can learn about the realities of technical advancements based on historical precedent, and we can use that knowledge to inform what to do as new AI advances come up.

Let’s look back at two examples of innovative technologies: The Haber Process and YOLO.

The Haber Process

During World War 2 Nitrogen, in being a critical part of explosive manufacturing, was a scarce resource. It had to be harvested from things like animal manure, guano, and bone meal. Guano was a particularly hot commodity; as a fun side note, there was an entire era of imperialism sparked in the quest for bat poop.

Whenever any citizen of the United States discovers a deposit of guano on any island, rock, or key, not within the lawful jurisdiction of any other Government, and not occupied by the citizens of any other Government, and takes peaceable possession thereof, and occupies the same, such island, rock, or key may, at the discretion of the president, be considered as appertaining to the United States.

— Section 1 of the Guano Islands Act

Anyway, at the beginning of World War 2, it was hard to make explosives.

Fritz Haber was a Nazi scientist who wanted to increase the capacity of Nazi Germany’s arsenal. In that pursuit, he devised a chemical mechanism for efficiently extracting nitrogen from the air in the form of ammonia. Effectively, this process used the air to make explosives.

Thanks to the Haber process, the military war machine responsible for one of the largest catastrophes in human history wasn’t bound by bat poop, it could explode the air. This discovery was largely responsible for four years of world war (around 14 million deaths).

Nitrogen is also an important component in fertilizer, which is important in feeding people. The Haber process, since its discovery, has been a central pillar in industrial agriculture. It’s estimated that around half of humans owe their existence to synthetic fertilizers created via the Haber process. Though feeding half the world was an unintended side effect, this won Fritz Haber (a Nazi Chemist who strong armed the Holocaust) the Nobel prize in chemistry.

The Haber process is an extreme example of a common trend in groundbreaking technical progress: Regardless of if you make a technology for good or for evil, people will use it for their own purposes.

Let’s explore a more modern example of innovation directly related to AI

YOLO

The principal author of YOLO, Joseph Redmon, is a personal hero of mine. By all accounts he seems like a standup guy and genius researcher. He’s also a fan of funny names.

YOLO, or “You Only Look Once” is an AI model that can look at an image and extract what is in that image in real time.

YOLO flipped the script on object detection, and completely reimagined the underlying processes. In a nutshell, without YOLO things like self driving cars, visual cancer detection models, and modern robotic fruit pickers wouldn’t be possible.

Joseph Redmon gave a lot of talks about YOLO. You can find them on YouTube, they’re all really good. In earlier videos there’s a clear excitement and passion. I can only imagine the exhilaration of coming up with such a groundbreaking technology and sharing it with the world.

If you watch his videos chronologically, though, you’ll find his presentations take a darker turn as time goes on. Joseph gave a TED Talk around five years ago where he discussed the impact of his technology. He talked about a time an army general approached him, and showed him a video of YOLO being used on drone footage to track vehicles on a battlefield. Around 2015–2020, drone strikes were a point of international controversy; responsible for hundreds of civilian deaths. It’s very possible that some of those deaths were in some way enabled by YOLO.

Since YOLO, Joseph has stepped away from computer vision research; abhorred with the misuse of his technology, and disillusioned by the cycle of technology being used to support power, and power being used to fund technology. He’s since made claims that no AI research is ethical under capitalism. I was confronted by this early on in my data science career and have been grappling with it ever since.

Whos Job is it to Be Good?

It’s pretty clear, if one looks back into history, that it’s impossible for a creator to control the use of their creation. From Perillos of Athens being tortured with his own devices, to nuclear energy being used both as weapons of doomsday and the world's cleanest and most abundant energy source. If you come up with something important, something powerful, you have to come to terms with the sobering reality that other people will use your creation for their own purposes.

In our age of rapid advancements, technological progress controls the speed of the world. But we all collectively control weather we drive to eden or off a cliff. Geniuses can create new technologies, but it’s up to all of us to use them correctly.

I think the day of poets on one side and engineers on the other has to end. To make the right decisions in the modern era, one has to understand both what can be done and what should be done. Politicians throwing their hands up at basic technological concepts is no longer acceptable, technologists can no longer be willfully ignorant of the use of their labors, and the masses need a more discerning view of what is good and what is bad.

  1. It’s everyone's job to educate themselves on the humanities. History, philosophy, rhetoric, art. If you don’t find value in these things then you are blind. These are the things which humans use to reason about human issues, to explore what is right, just, good, and beautiful.

  2. It’s everyone’s job to educate themselves on technology. Technology is progress, and without understanding technology you will be blown by progress like a leaf in the wind.

It is no longer enough to be a good person, it is no longer enough to be a capable person. You have to be both, and you have to be both now.

Conclusion

AI is like any other groundbreaking technical advancement. While it might be compelling to think of AI as a force which will play out to it’s inevitable conclusion, the reality is both harder and more optimistic; It is up to us.

Instead of preaching my disparate opinions in a manner that’s too on the nose, I want to wrap this up with a few quotes which resonate with me.

“Man is a rope stretched between the animal and the Superman — a rope over an abyss.” — Friedrich Nietzsche

“As if you could kill time without injuring eternity.” — Henry David Thoreau

“When force of circumstance upsets your equanimity, lose no time in recovering your self — control, and do not remain out of tune longer than you can help. Habitual recurrence to the harmony will increase your mastery of it.” — Marcus Aurelius

“What is done out of love always takes place beyond good and evil” — Friedrich Nietzsche

For an elaboration of some of the philosophical ideas of this article:

Follow For More!

I describe papers and concepts in the ML space, with an emphasis on practical and intuitive explanations.