ChatGPT Issues Superintelligence Warning

An Interview with ChatGPT Part 3 

by Mary Boza Crimmins

Not a Prophet, But… 

I’m not a techie, nor am I a prophet. But as you read this poem written by my 14-year-old self, perhaps it will bring your thoughts to what ChatGPT cited as one of its dangers: superintelligence. 

“Epistle To Be Left…”

I did not mean to do all this, 

To take away the reason for living,

The reason for work, play, love, and success.

And the reason for leisure.

Leisure 

The reward for work.

I had forgotten the meaning.

My invention

The ultimate in inventions (or so I thought). 

Man would live a life of comfort. 

No need to study or work. 

No need to think for himself. 

I knew that. 

But I was looking for fame, for fortune! 

I got it……..but for what? 

Man lived a life of play, of leisure, of comfort. 

It did not take long, though

For his mind to rot, for his trust to disappear. 

Trust 

The lack of it is what killed man. 

The accusing, the greed, the hating. 

One by one, man left this earth. 

I tried to stop them, to save them, but their greed was worse than mine. 

I tried to destroy the machine. 

Like I said, though, it is the ultimate in inventions. 

Man is now extinct, or nearly. 

I am left. 

I’ll die soon, if not of loneliness, then of shame.

Will you ever, God, forgive me? 

Forgive us?
What a fool I have been. 

What fools we have all been. 

Superintelligence Defined 

Superintelligence is defined by Nick Bostrom of the Oxford Future of Humanity Institute as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom, and social skills.” Similarly, Miriam-Webster defines it as “an entity that surpasses humans in overall intelligence or in some particular measure of intelligence.”

What ChatGPT Says About Superintelligence 

Part 1 of the series was the “conversation” I had with ChatGPT. I interacted with it as if I was talking to a human. His responses ranged from humorous to thought-provoking. Chat, as I called it, was a bit verbose. As an editor, it was difficult to refrain from polishing his responses, but I wanted to accurately report what was said, and it’s also an example of ChatGPT’s shortcomings. Part 2 addressed the problem of bias with AI. In this last segment, we’ll take a look at superintelligence. Here’s the excerpt from the interview

ME: Do you think AI is dangerous to humanity?

CHAT: The question of whether AI is dangerous to humanity is a complex and debated topic. AI, like any powerful technology, has the potential for both positive and negative impacts, depending on how it is developed, deployed, and used. Here are some perspectives to consider:

• Potential for superintelligence: The concept of superintelligent AI, where AI systems surpass human intelligence, raises theoretical concerns about control and alignment. Ensuring that highly intelligent AI systems align with human values and goals is a topic of ongoing research.

Researchers and Rogue Machines  

How do we prevent the machines from taking over? Just pull the plug is a response to dealing with the possibility of AI going rogue. But “Switching the machines off won’t work,” states Stuart Russell in his book Human Compatible: Artificial Intelligence and the Problem of Control, “for the simple reason that a superintelligent entity will already have thought of that possibility and taken steps to prevent it.”  

On the other hand, computer scientist Manuel Alfonseca and a team of researchers published findings regarding how to control AI in the Journal of Artificial Intelligence. Alfonseca concedes, “It may be impossible to control superintelligent AI.” However, the research indicates It will be at “least two centuries” before the technology reaches a level of superintelligence in the form of being able to. Furthermore, researchers do not know if it is even possible to develop a machine as intelligent in every field as humans. 

Reining in the Machines 

Will the machines take over the world? I think not. Well, I hope not. But scientists, lawmakers, and business leaders must still consider the potential risks of artificial intelligence on humanity. The issue is mind-boggling. I do know this, though. Regardless of technology’s abilities, it's critical for us to sharpen our ability to think rationally and logically while working on consistently treating others well. 

My 14-year-old self was a reflective soul. Decades later, I have a more profound love of life and a deeper faith in God. I’m optimistic and realistic. I’m also thankful it is impossible to predict the future. 

Here are the links to Part 1 and Part 2. Leave a comment on how AI has affected you.

Need an editor? Researcher? Writer? Project Manager? Contact me.

Sources: 

Bostrom,  Nick. How Long Before Superintelligence? Oxford Future of Humanity Institute 

Choi, Charles Q. Institute of Electrical and Electronics Engineers; January 2021. Accessed May 29, 2023; Superintelligent AI May Be Impossible to Control; That's the Good News - IEEE Spectrum

Miriam-Webster.com. Miriam-Webster defines; accessed May 29, 2023 

Russell, Stuart. Institute of Electrical and Electronics Engineers; Oct. 2019; accessed May 29, 2023; https://spectrum.ieee.org/many-experts-say-we-shouldnt-worry-about-superintelligent-ai-theyre-wrong