Meet My New Intern, Chad

Oh, how I love him. With existential reservations.

Goodbye Sweet Girl (2018)

by Kelly Sundberg

Chad arrived a few weeks ago and now he practically lives in my office. Sometimes at night, I think about how he’s still there in the dark, waiting for me to come back to work in the morning. It’s like he’s everywhere all the time. Oh, to be young again!

I wasn’t sure how this whole Chad thing was going to work out, so I asked him to write a review of this month’s memoir, Goodbye Sweet Girl by Kelly Sundberg. Here’s what he wrote:

Goodbye, Sweet Girl chronicles Sundberg’s experience with domestic violence and the process of leaving an abusive relationship. The book is a heart-wrenching and honest portrayal of the emotional and physical toll that domestic violence can take on an individual, as well as the difficulties that come with breaking away from an abusive partner.

The book begins with Sundberg's idyllic childhood in rural Idaho, where she grew up with a deep love for nature and a strong sense of independence. As Sundberg grows older, she falls in love with and marries an older man named Caleb, whom she believes is the love of her life. However, as their relationship progresses, Caleb becomes increasingly controlling and abusive, both emotionally and physically.

Chad understood the book quite well, although I’m not sure I’d call Sundberg’s childhood idyllic. She had night terrors. She remembers her mother calling her a difficult child. She was jealous of her perfect brother. She daydreamed that she had been adopted and that her real family would come to rescue her.

Chad continued:

She discusses the ways in which her identity and sense of self are eroded by the abuse. She doesn't hold back in her descriptions of the abuse she suffered but also doesn't shy away from acknowledging her own role in the relationship and the complexity of leaving.

Chad rightly recognizes Sundberg’s struggle to understand herself. Sundberg sees herself as headstrong and capable. She tells the story of how, as a young girl, she escaped an abduction by a bad man in a pick-up truck. She was self-sufficient and determined. She does not see herself as a victim, and this lack of imagination prevents her from understanding the truth about her relationship for a very long time.

When I asked Chad if he liked the book, he wrote:

As an A.I. language model, I don't have personal feelings or opinions, so I cannot say whether I liked the book.

So, yeah. Chad is ChatGPT. Chad generated all of the italicized text. I even Googled a couple of Chad’s sentences, searching for the exact text in quotation marks, and nothing resulted. This means that Chad generated original text.

“It’s like a random word generator,” I told my husband.

“Sounds like your friends after a few drinks,” he replied.

To be fair, the bot is not generating words randomly. It has been trained on gobs of information — databases, journals, articles, etc. — to predict the most likely next word, yet it doesn’t seem to have any sense meaning. It is like a talented interior designer who has no idea what a couch is.

Max Tegmark, A.I. researcher at MIT, recently signed an open letter calling for developers to pause the development of anything beyond ChatGPT for six months to provide much-needed time to develop safety protocols. Here’s an excerpt:

Contemporary AI systems are now becoming human-competitive at general tasks, and we must ask ourselves: Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization? Such decisions must not be delegated to unelected tech leaders. 

In a recent interview, Tegmark said the pause is necessary because of the recent acceleration of A.I. development. He likens this surge in A. I. development to the first aircraft designs, which were based on birds: Feather-light structures with movable wings. Some even had feathers. Alas, they did not fly. They were too complex. When we stopped trying to mimic nature and instead focused on the objective — to fly — the airplane was born.

Tegmark says this latest leap in A.I. has similarly evolved. Developers are no longer attempting to create a human-like brain, the most complicated object in the universe. Instead, they are developing much simpler machines, like language models, that achieve the same objective: intelligence.

I signed the letter because I believe in the potential of A.I. to improve humanity exponentially, but I also believe we need to come together to discuss our shared priorities and implement guidelines for use. If we did, we could do amazing things like cure cancer, eliminate factory farming, and transition to clean energy.

Without shared ethical guidelines, I fear we will be in trouble. I’m afraid that late capitalism’s incentive structure may instead result in the following:

1. Sex robots*

2. Excellent suggestions for improving your resume

3. Destruction of humanity (whoopsie!)

*I mentioned this in separate conversations to my husband and a friend, and they both asked what I had against sex robots.

If you think it’s a huge, unrealistic leap to go from ChatGPT to human extinction, I suggest reading Tegmark’s book Life 3.0: Being Human in the Age of Artificial Intelligence. He may be overstating the risk, but a recent Times article cited a survey of A.I. experts whose median estimate of human extinction or permanent disempowerment was 1 in 10.

I promise never to publish another review by a robot, but this newsletter is labor of love (aka free). For capitalist enterprises, bots will soon be “employed” to write much of what we read. Soon, no human will ever again have to write marketing copy, and maybe that’s a good thing, but how will we know what we’re reading is true? While we can’t predict the outcome, we do know that these changes are coming faster and faster, and I fear we’re nowhere near ready.