Evanston RoundTable, July 26, 2023
The humor magazine National Lampoon published a cover story in 1971 headlined “Pornography: Threat or Menace?”
Today they may want to rephrase it: “AI: Havoc or Chaos?”
That mock hysteria sums up a lot of recent scare stories regarding artificial intelligence, which last year gave us the intriguing ChatGPT. You can try it yourself here.
Currently ChatGPT and its counterparts – Bard, Bing and other language processing tools – are in bad odor. “The Federal Trade Commission is investigating whether OpenAI’s ChatGPT has harmed people,” reports the Wall Street Journal, “by publishing false information about them, posing a potential legal threat to the popular app that can generate eerily humanlike content using artificial intelligence.”
On Friday, July 21, seven major tech companies – Amazon, Anthropic, Google, Inflection, Meta Platforms, Microsoft and OpenAI – meeting at the White House agreed “to put more guardrails around artificial intelligence, including the development of a watermarking system to help users identify AI-generated content, as part of its efforts to rein in misinformation and other risks,” the Journal reported.
Judging from my own experience, these risks are pretty minimal – at least for now.
I know because my web guy recently sprang a nasty surprise on me. Out of the blue I got an email from him that read, “I like to do a little bit of writing on my own. I rephrased your book’s introduction a bit. What do you think?”
He had helped me post my new novel – a sci-fi, time travel, historical fiction thriller, seven years and dozens of rewrites in the making – on my website lesterjacobson.com. Now he’d revised it? Without asking? What chutzpah!
Worse yet, it was awful. But don’t take my word for it, judge for yourself. Here’s his rewrite, with my snarky comments boldfaced in brackets:
“It is done, a masterpiece of the ages, a creation destined to reshape the course of history [purple prose alert!]. And here I stand, brimming with an indescribable euphoria [brimming with euphoria?]. Can such elation transcend the boundaries of infinity itself? [Hoo boy.] It must be possible. However, my jubilation is tempered with trepidation [more purple prose] for what I am about to embark upon [just “embark on,” please] is perilous and illicit. That is why I toil [“work” would have worked better] alone, risking everything in the clandestine pursuit of my ambitions [yet more purple prose!].
“As I gaze out the window, my eyes fixated [what’s the matter with fixed?] on the monolithic monument [grating, redundant] that stands sentinel over the site of my monumental achievement [makes no sense!], I can’t help but wonder: Did Fermi experience a similar sensation? Picture him [isn’t that what the novelist, not the reader, is supposed to do?], hidden away in that subterranean laboratory beneath Stagg Field, on that fateful afternoon of December 2, 1942 (according to the old calendar), when he deftly arranged the uranium and graphite stacks, birthing the atomic age [Birthing as a verb? It shall not be birthed!].”
The rewrite went on for several more paragraphs in that same style but I’ll spare you the painfully bad writing. You get the point.
I didn’t know which was worse: his nerve or his prose.
You can compare his version above with the actual opening of my book, The Dream Machine: A Novel of Future Past. It’s available free to read. (Shameless plug: Feel free to read not just the opening but the whole book!)
My reply to him was terse: “Please restore the original.” There was a lot more I was tempted to say, but I clamped down on the proverbial tongue (less painful than the actual one), and refrained.
“Les, there’s nothing to restore,” he wrote back, doubtless chuckling with every keystroke, “as I didn’t update the website. I wanted to show you how AI works with rephrasing and generating content. Pretty powerful isn’t it? This was created by the famous ChatGPT.”
A lesson in humility
OK, so it was a lesson in 1) not leaping to conclusions, 2) learning Zen patience and humility 3) a demonstration of what ChatGPT can do. I had tried Bard, a similar app developed by Google, when it first came out, and got similar weird and unsatisfactory results.
Curious, I asked him what exactly he asked of “the famous ChatGPT.”
“I requested ChatGPT to rephrase it in an interesting tone and it replied with that content in 3 seconds,” he answered. “This AI software has taken the world by storm, it’s the biggest thing in human advancement since the smartphone came out, and then the internet before that. People don’t realize how powerful this technology is. Governments will need to regulate this technology because it’s crazy what it can do, not just with copywriting but with everything else.”
It’s hard to disentangle all the hype. Take the world by storm! Biggest thing in human advancement! Cure cancer! Discover life on distant planets! Converse with orcas! Leap tall buildings in a single bound!
And lots of alarms too, along the lines of: Displace millions of workers! Perpetuate false stereotypes! Take over humanity!
As a good friend wrote me recently, if David Brooks, the usually even-keeled and generally upbeat columnist for the New York Times, is worried, then maybe we should be too. “Human Beings Are Soon Going To Be Eclipsed,” the Times headlined Brooks’ column of July 13.
Scarier still, an op ed in the July 2 Times shrieked, “The Risk From A.I. Isn’t Just Existential,” as if there’s something worse than extinction.
Bane or boon?
So which is it: Bane or boon?
Like every other scientific advance, it will almost surely be both. The industrial revolution gave us the wonders of the modern age but also carbon pollution, which could wipe out civilization. Modern science gave us antibiotics and microsurgery but also nuclear and biological weapons. The internet gives us the wonders of GPS and Shazam but also toxic social media, pernicious lies and misinformation and Beezin’.
The key is how we use and control it. So far – the July 21 White House agreement notwithstanding – the results aren’t promising.
Judging from its rewrite of my novel, AI has a long way to go before it‘s mistaken for Hemingway.
But then that’s true of me too!
Just check out the recent 60 Minutes piece on AI.
AI seems incredibly powerful for good but
also for (very) evil.