Artificial Intelligence is About to Redefine the Human Experience

Until recently, I haven’t been interested in artificial intelligence (AI). It seemed too far off, too technical, and not particularly relevant to my life. 

It’s becoming clear I’ve had my head in the sand!

Prior to his 2018 death, theoretical physicist, Stephen Hawking, commented:

“With artificial intelligence, we are summoning the demon…AI is a fundamental risk to the existence of human civilization.”

Two weeks ago (March 21, 2023), OpenAI announced the release of an updated version of its amazing ChatGPT bot, called GPT-4. It’s capabilites are freaking people out.

Echoing Hawking’s concerns, last week, (March 30, 2023) over 2,000 industry experts, executives and interested partys, including Elon Musk, and Apple co-founder Steve Wozniak, signed an open letter supporting a pause in AI system development. 

Here is an excerpt from the letter:

Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources. Unfortunately, this level of planning and management is not happening, even though recent months have seen AI labs locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control.”

So much for background. Now, why all the fuss?

First, GPT-4 is a killer test taker. It scored in the upper 10% on a uniform Bar exam, as well as in the 80th, 99th, and 56th percentiles in the quantitative, verbal, and written sections, respectively, of the Graduate Records Exam (the most frequently used admissions test for college graduate school programs). 

Second, it was reported that GPT-4 easily passes Turing tests designed to determine if a computer can fool humans into thinking it’s human. 

Third, GPT-4 can write computer code.

GPT-4 is not yet widely available, but I have an OpenAI account that gives me access to the latest version of GPT-4’s predecessor, ChatGPT-3.5. Though not as capable as GTP-4, it’s still impressive. I recently asked it for advice on a serious medical condition our son is facing. In less than a second, it provided an answer that mirrored detailed discussions I’ve had with his neurologists and neurosurgeons. 

Ok, so AI chat bots might offer a better search experience. What’s so scary about that? 

I’ll let you be the judge. For this post, I asked ChatGPT-3.5 to write an essay on “Being” in the style of my blog. 

Here’s an unedited excerpt:

To truly be, to truly exist, is to live in the present moment. It’s to be fully present and engaged with the world around us, to appreciate the beauty and wonder of this life. It’s to recognize that we are more than just our thoughts, our emotions, or our physical bodies.

Being is about living a life that is authentic and true to ourselves. It’s about recognizing our own unique talents, passions, and values, and pursuing them with conviction and purpose. It’s about being true to our own selves, rather than conforming to the expectations of others.

I showed the response to my wife, and she commented, “ChatGPT is a lot better writer than you are!”

I agree!

Here’s the crazy thing! Large language models (LLM), like ChatGPT and GPT-4, don’t rely on the meaning of words to function. Instead, they apply statistical inference to internet-sized data sets to guess what word should come next in a response. Emily Bender, a computational linguist at the University of Washington, calls LLM models “stochastic parrots.” She is concerned their impressive capabilities will be conflated with human intelligence, to our detriment. 

As of last week, it’s clear others agree!

NY Times podcaster, Ezra Klein, recently referenced a survey given to leading AI engineers on  what this new technology might mean for humanity. The respondents listed a median 10% odds that the technology could get out of control, posing an existential threat to human survival. 

Would you work on a technology if you thought it had a 10% chance of wiping out the human race?

Of course, it could go the other way. American internet entrepreneur and venture capitalist Reid Hoffman (who sits on the Microsoft Board of Directors and had early access to GPT-4), just published a book co-written with GPT-4 titled, Impromtu: Amplifying Our Humanity Through AI. A free PDF version is available here.  I’ve read enough of the book to sense that the ground is shifting below our feet. I’m uncertain whether that is good or bad! 

ChatGPT has 100 million subscribers. Furthermore, OpenAI is not alone in the development of AI models. Google and Meta have versions, along with, no doubt, other technologically advanced organizations (and governments) outside the United States. 

To date, LLM program advancements have been led by human collaborators. However, GPT-4 reportedly can write complex computer code. Will further improvements in its capabilities be driven autonomously? HAL are you listening?

In my favorite movie, The Matrix, leading man Neo, played by Keanu Reeves meets Morpheus, played by Laurence Fishburne, an operative serving in the city’s defense force against attack from the “Machines.”  Morpheus gives Neo a choice of two pills: a red pill to learn the real truth about the matrix, and a blue pill to forget he ever met Morpheus and to go back to his old life.

AI is presenting a similar choice to us.

We have reached the dawn of a new age. Widely available AI technologies have the capability to dramatically enhance (and/or  irreparably harm) humanity. I hope someone besides AI development engineers is thinking that though! 

In the meantime, Like Neo, I’ve decided to take the red pill, and engage the machine. I was not sure how to posture this post, so I fed the above words into ChatGpt-3.5 and asked it to pick a title. It’s response was the title I chose. 

——————————————————————

If you enjoyed this post, please share it with a friend.

Subscribe to my latest posts by filling in your email address at the bottom of this page.

Previous posts can be found here by subject category, and here chronologically.