AI: Sorry, You May Not Always Sit at My Table

Ethan Mollick's book Co-Intelligence is a dangerous document. There, I said it. We read the text for our campus digital pedagogy cohort for Fall, 2024, and at first I was quite excited. The book promises to provide guidelines for wisely adopting AI, yet from the get-go I had issues with the author's less-than-critical acceptance of generative AI.

Let's start with Mollick's four, and seemingly absolute, rules for working with AI:

1. Always Invite AI to the Table.
2. Be the Human in the Loop.
3. Treat AI Like a Person (But Tell It What Kind of Person It Is).
4. Assume This Is the Worst AI You Will Ever Use.

I'm not much trouble with these save for number 1. I've been reading Wendell Berry, a techno-selective if ever there were one, and I am pretty certain how he'd react to generative AI: Hell no.

Why? Mollick's rule number one flatly violates Berrys notion about adopting new tools.

It also flies in the face of what Howard Rheingold, reporting for Wired eight long years before the iPhone 1 debuted, found out about the Amish. In "Look Who's Talking," he describes how the Amish approach any technology. They experiment to see if it violates:

a body of unwritten but detailed rules known as the "Ordnung." Individuals and communities maintain a separation from the world (by not connecting their houses to telephones or electricity), a closeness to one another (through regular meetings), and an attitude of humility so specific they have a name for it ("Gelassenheit"). Decisions about technology hinge on these collective criteria. If a telephone in the home interferes with face-to-face visiting, or an electrical hookup fosters unthinking dependence on the outside world, or a new pickup truck in the driveway elevates one person above his neighbors, then people start to talk about it. The talk reaches the bishops' ears.

Thus the Amish man who recently had a long chat with me about the virtues of PEX-based plumbing systems. He, like other Amish (and me) are techno-selectives, not utter Luddites.

Rheingold's piece had an enormous intellectual influence on me. It made me wary of new tools, though I admit falling without reservation for virtual worlds. I've come to regret that uncritical acceptance of a technology that not only fails the test above but also fails with a clumsy UI and poor use-case in education.

Maybe in consequence, I grew wiser about smart phones, doubting the corporate narrative from the start; today, mine is nearly always off. I don't use it much for one social media platform, less so for texting, have silenced all notifications, and never at all watch time-killer videos. Pop culture generally seems too ephemeral for my remaining years on the planet, and I don't want to discuss TV shows with friends. If something looks non-violent and well written, I still wait years before watching a series, methodically. 

Influencers? I find them in books. Berry is one. I bought heavily into the concept with which Rheingold closes his piece, "If we decided that community came first, how would we use our tools differently?"

I don't consider online community, including virtual worlds, much of a substitute for the real thing. Even gaming with old friends on our Monday Nerd Nites seems a pale shadow of a good in-person meeting. Only recently I began to read Berry, and he corroborates much of what Rheingold discovered.

So for Mollick's rule one, I plan to say "no" a lot. What do we gain by always using AI for any intellectual task? In Berry's 1987 article "Why I Am Not Going to Buy a Computer," he lays out nine rules for adopting a new tool:

1. The new tool should be cheaper than the one it replaces.

2. It should be at least as small in scale as the one it replaces.

3. It should do work that is clearly and demonstrably better than the one it replaces.

4. It should use less energy than the one it replaces.

5. If possible, it should use some form of solar energy, such as that of the body.

6. It should be repairable by a person of ordinary intelligence, provided that he or she has the necessary tools.

7. It should be purchasable and repairable as near to home as possible. 

8. It should come from a small, privately owned shop or store that will take it back for maintenance and repair.

9. It should not replace or disrupt anything good that already exists, and this includes family and community relationships.

Generative AI fails most of these, in particular anything to do with localism and energy use, but number nine reminds me of the unforeseen outcomes of ubiquitous mobile telephony: families in restaurants, not looking at each other, all intent on their screens and being somewhere else. That would seem perverse to anyone of my parents' generation or earlier.

So how do we NOT give into Mollick's notion that we must always bring AI to the table, dinner or otherwise, even without a good use case? I plan to be writing about that in the new year as I continue researching AI's role in the writing process, where I do cautiously let it have a seat.

Image by Duncan Cumming on Flickr

AI as Transcription Tool, Using Zoom

 

 

Even those afflicted with terminal Zoom-fatigue after the pandemic will enjoy a feature that recently
popped up in my Zoom room, a little button labeled "AI Companion." It has proven a good companion indeed.

I have been mentoring a mid-career director of a writing center on the West Coast, and by accident I triggered the feature. In return we got this, when the meeting ended:

• An executive summary of the topics we discussed
• A bullet-listed synopsis of ideas, suggestions, and strategies broken down by topic
• A nearly error-free synthesis of the entire meeting.

The software (Zoom's own proprietary AI) uses Zoom's captioning tool to capture what we say; no video gets recorded. I'm guessing that the data gets shared without our permission with the firm. Caveat emptor.

My mentee loved the accidental result and plans to try it himself. With a few writing students who agreed to use the Companion, I opened my Zoom room during a face-to-face meeting. When done, I e-mailed them the Zoom AI synopsis. Again, save for misspelling a name or two, the AI succinctly and clearly summed up what we had done.

Whenever I meet someone "scared" of AI, I mention uses such as this as good practices. While I remain a skeptical adopter of this new technology, I rather like a robotic note-taker who works with us openly (and not in secret) while freeing humans to focus on each other rather than on taking notes. 

For students with dysgraphia or other disorders that require them to record a class session, this little feature of Zoom is a godsend. Try AI Companion yourself and let me know what you think.

A False Filter in the Drake Equation: LLMs

I read a piece today at Space.com, about a fierce solar storm that hit our planet some 2687 years ago. Another such Miyake event today would likely spell the beginning of a new dark age (looks at wood stove. Better get busy splitting and learn to cut logs without my chainsaw).

The author, Daisy Dobrijevic, goes on, at some length for what I often find a casual venue, to discuss why the stability of our home star matters to our survival as a technological civilization. For once, the comments in these pieces prove as interesting as the source. Those leaving remarks compare Sol to others we observe, noting that our star's orderly behavior could mean that we are a solitary civilization in our galactic neighborhood. These respondents go on to discuss other "filters" to the famous Drake Equation. 

I stupidly had assumed that all stars like our own prove relatively stable. If not, the number of possible civilizations out there could plummet. 

Being alone proves bracing and  calming to me, personally. Yet there exists a difference between solitude and loneliness. If we are the only civilization in our part of the galaxy that seems less about solitude, a healthy thing for many humans, and more about isolation, a terrifying prospect to me.

Read both article and comments, including speculations as to the number of civilizations likely in our own galaxy. Let's forget, save as an intellectual exercise, those in galaxies beyond. The Deep Dark of millions of light years means that if they exist or once existed, we'll likely never encounter them, barring some difficult-to-imagine technology such as wormholes or other exotica far beyond our race, which cannot yet harness nuclear fusion.

I liked the piece by Dobrijevic and commentary a great deal, but in the comments I found one regret, that "Unregulated-LLM insanity" could spell the end of our species. That sort of claim needs challenging, though I'm not eager to start a flame war with a bunch of anonymous posters. Generally, the terms LLM (and AI) have been bandied about too casually.

When Doomscrollers, Doomers, and Millennial Cultists talk of an AI Apocalypse, they envision the sort of Hollywood fantasy of the Terminator or Matrix franchises. My brain goes right to to origin of these fantasies, Harlan Ellison's poignant and horrifying "I Have No Mouth, And I Must Scream." The story does get a visual homage in the first Matrix film. I don't worry about such horrors getting real.

Frankly, current large language models cannot even ask good critical questions as writing coaches. Thus I find it improbable to imagine ChatGPT or Anthropic Claude as planetary overlords, like the nightmarish AM in Ellison's story.

Were General AI to emerge, we might have the sort of "filter" mentioned by the commenters. Some AI companies have that sort of software in mind: a network with superhuman intelligence, able to reason and grow by itself, much as a human does. It could change its "training data," much as we do when we decide on our own to study something new. Like us, it would then we self-aware and have agency beyond answering questions. It could make decisions on its own, contacting us, for instance, with questions of its own. I consider that a new form of life.

Until then, we merely have energy-hungry LLMs in data centers, parsing our words and predicting the next one to appear in a sequence. I'm no neuroscientist, but that's not how a human mind works.

While I'd be a fool to say we have nothing to fear from AI, my concerns are more earthly: unemployment, lack of income, not to mention worries about climate change, nuclear proliferation, erosion of rule of law, regional or global war, civil strife. So I'd tell our forum pundits opining about AI threats this: we have filters aplenty closer to home than some super-intelligence.

Be mindful, then, of the terms used in discussions about AI. We who work with this technology do not want to propagate careless thinking about a very important, and still evolving, innovation. If you are still wishing to learn more about The Drake Equation, have a look at the NASA site where I got the post's lead image.