Ask an AI Skeptithusiast

Dear AI Skeptithusiast,

I get the “skeptic” part of your name since AI is bound to take over and enslave us one day, but why are you enthusiastic? Honestly, I’m pretty efficient with good old fashioned Google, and don’t need a bot to do my writing for me.

Sally A, Columbia, MO

Dear Sally,

I find that GenAI goes far beyond what a search engine can do. It enables you to take on research tasks very quickly, whether for work or just for fun, that would be really onerous otherwise. Since you’re from the Show-Me State, I’ll give you an example. I become aware somewhat recently of the “Bechdel test,” which was contrived by the cartoonist Alison Bechdel, and consists of three simple criteria for measuring female representation in film: 1) Does the movie have at least two named female characters, 2) who talk to each other, 3) about something other than a man? (It’s surprising how many works fail.) Intrigued, I realized I could very easily see how my tastes align with this. I’d already listed twenty of my favorite movies in these pages, and it took me all of thirty seconds to upload them to Claude with the prompt, “Please analyze the following titles in terms of whether they pass the Bechdel test and, if so, is it by a wide margin or just barely?” I didn’t have to suggest how Claude could determine this; it found a crowdsourced database at bechdeltest.com for several of them, and then did its own separate research for the more obscure ones, and flagged these latter conclusions as “lower confidence.” A couple examples of its output:

The Housemaid (2010, Im Sang-soo) — Passes, confirmed 3/3, with “several scenes” noted by the database. Eun-yi, the wife Hae-ra, and her scheming mother Mi-hee talk directly to one another about the pregnancy, the household, and their maneuvering—not just filtered through the husband.

The Wedding Plan — Likely passes, though the whole premise (finding a groom) makes “about something other than a man” a close call in places — much of the female ensemble’s talk is literally about the wedding/matchmaking. 

(In case you’re interested, 14 out of 20 of my movie picks easily passed the Bechdel test, and a couple others were too close to call.)

As you can imagine, this just-for-fun exercise would have been a lot of work without AI, to the point it probably wouldn’t be worth the trouble. For an example of a more utilitarian task, consider a way in which AI helps me with this blog. I have a Gemini Notebook into which I’ve loaded all my Bits & Bobs posts. Bits & Bobs are snippets from old letters and email (pre-albertnet), and when I’m writing a new volume of them, I have to comb my archives for useable passages. When I find something I like, I have to make sure I haven’t already used it, which used to be very tricky. (As the devoted albertnet reader already knows, “The different volumes of Bits & Bobs are not organized chronologically, thematically, spatially, or in fact at all.”) Pre-AI, I had to rely on memory, or skim through a continually growing blog archive, to know if a snippet had already been included. Not anymore—I just run it by the Notebook chatbot. (For example, “Have I blogged before about acne?” turned up “Milton as a zit“ and “the Natalie Portman whitehead,” but not the “pimple-tunnel episode,” which I thus included in my last Bits & Bobs post.) As a bonus, every single Bits & Bobs snippet (and there are hundreds at this point) has a date tag on it, so I could very easily ask AI to assemble them all in chronological order, thus “writing” my autobiography for me (in my own words, of course). What search engine could achieve that?

Dear AI Skeptithusiast,

When it comes to AI I say “burn it down!” I want nothing to do with it. It’s replacing human thought, effort, and learning. I made my daughter promise not to use ChatGPT to write her college papers but she told me recently, “I held out as long as I could, but the temptation is just too strong.” Why am I paying her tuition so she can outsource her “learning” to a bot? How can you be in any way bullish about this technology?

Anneke R, San Luis Obispo, CA

Dear Anneke,

First of all, I completely agree with you that having an AI write a school paper for you is a terrible use case; in fact I’ve blogged at length, here and here, about how writing for yourself sharpens your intellect and results in a better product as well. But these are arguments for not abusing AI, which is what your daughter is doing. That’s a “her” problem, not a platform problem. Believe me, I’m plenty vexed at how so many teens—and adults—are using modern technology, and not just AI. Look at how many are glued to their smartphones 24x7, doomscrolling until they’re depressed, or playing stupid games. But do you yourself use a smartphone? If so, you’ve already learned how to reconcile use vs. abuse when it comes to technology … so why are you so hesitant to extend this to AI? You and your daughter can choose to take advantage of its capabilities without falling prey to its weaknesses. (I’ve even highlighted what I think are the five most important AI “DON’Ts,” in this handy post.) In short, when it comes to AI and your frustration with it, don’t bite off your nose to spite your daughter’s face!

Dear Skeptithusiast,

Microsoft came out recently with its “Humanist AI Code of Conduct“ and I have to say I’m fairly impressed. It lists several important guardrails for AI, such as: human control comes first; AI must have absolute constraints nobody can override; and, it needs to be transparent in its “thinking.” Of course this is only a start. What would you say is the biggest gap in this Code?

Shawn P, Columbus, OH

Dear Shawn,

To me, the biggest gap is that it doesn’t address economic displacement—the loss of jobs due to AI. (Neither do the similar self-imposed guidelines created by OpenAI, Google, and Meta.) Part of how platforms get away with sidestepping this is by citing the lack of consensus about how many jobs AI can really automate, and the disparity between how many have been automated vs. how many could. But this isn’t very fair, as far as I’m concerned. First off, companies are very wary of hiring right now—perhaps waiting to see how much AI can reduce their need for new employees—which is hurting the job market. Second, lots of companies, whether or not they’ve reached “true AI maturity,” are laying off huge numbers of employees based in many cases on the mere promise of increased efficiency. Here are some recent mass layoffs:

• Marc Benioff, head of Salesforce, announced that AI enabled Salesforce to cut support headcount from about 9,000 down to roughly 5,000
• Block (née Square) cut more than 4,000 jobs — nearly 40% of its workforce — and its CEO, Jack Dorsey, tied this directly to AI
• SAP is cutting 8,000 job cuts (via a combination of layoffs and voluntary severance) in what their leader, Christian Klein, describes as a “generational restructuring around AI”
• HP has announced plans to cut 4,000–6,000 jobs by 2028, explicitly tied to AI adoption

Whether or not AI fulfills its promise, those laid off won’t necessarily get their jobs back.

What really interests me is this question: why do we all take it for granted that, when AI increases employee efficiency, all the benefit should accrue to the company (i.e., its profits), and no benefit to the employees? Instead of making our jobs easier, AI makes them go away, or at least raises the bar for keeping them. The climate created by these mass layoffs makes modern corporate America a big drag: we employees are made to feel expendable, if not downright unwanted, even as our companies prosper. And during this white-hot explosion of tech innovation, the tech sector is a like graveyard of lost jobs. Why should this be?

I’m reminded of what we learned in school about the original industrial revolution. Machines destroyed the value of a lot of skilled work, and the new industrial giants concentrated workers—many of them children—into factories and mines where they faced 12–16 hour days, dangerous machinery, no safety standards, and wages set entirely by employers, who had massive, unconstrained leverage. Only when collective organizing led to government legislation did we finally get, in the Fair Labor Standards Act of 1938, the 40-hour workweek and overtime pay. Aren’t you glad we have it?

But increasingly, we’re headed back in the direction of that dark time, 88 years ago, when the 60- to 80-hour workweek was standard, despite (and partly because of) industrialization making manufacturing far more efficient. And we just shrug about it. People could have passively accepted it back then, too, chalking it up to market forces. Back then, talk of a 40-hour workweek might have initially seemed like a fantasy—and yet they made it happen. So why not try again, and maybe turn the massively increased efficiency of AI into a 35-hour workweek, made mandatory for any company who adopts AI? This is a modest reduction, nothing like what we achieved last time we found ourselves here. And when I say “we” I mean the people and the government we elect. Obviously the world can’t expect the voluntary “codes of conduct” devised by giant corporations like Microsoft and Google to reign in AI’s plundering of our workforce. That’s letting the fox guard the henhouse.

Dear AI Skeptithusiast,

You’ve argued before that I shouldn’t use AI as a substitute for a human if I just need somebody to talk to. But if my conversations with an AI chatbot make me feel better, isn’t this a victimless crime? Why shouldn’t I be “allowed” to do this?

—Name and location withheld

Dear Withheld,

Let’s suppose, for the sake of argument, that the benefit that you get from chatting with a bot is ongoing, sustainable, and somehow capable of filling any sense of void in your life. That’s fine, but what about other people? The bot certainly isn’t going to feel better after your chat—it’s nothing but computer code and energy—but every minute it takes up your time is a minute you could have spent making another human also feel better. If everyone relied on bots for interaction, human relationships would just collapse. Obviously not everyone would or will opt out of human contact, but the trend is already disturbing. And by the way, this thought experiment aside, don’t kid yourself this kind of synthetic dialogue actually works long term. As documented by an MIT study, recounted here, chatting with AI instead of humans actually increases one’s loneliness over time.

Meanwhile, when you chat with a bot, doesn’t that automatically turn your thoughts inward, to yourself? How could that not happen, when you can’t ask the AI about its day, or its life, or its home, or its hobbies, or its health, or its fears? It doesn’t have any of these things, it has only its training data which cannot ever amount to a personality, a life, or a soul. So the natural topic is always you, which is bound to lead to some unhealthy habits. Admit it: isn’t part of the appeal that the chatbot is non-judgmental, doesn’t contradict you, and lets you always be the focus, in just the way your friends don’t? And doesn’t that seem a bit indulgent, a bit wrong? Could it be that real friends, who challenge you, help keep you honest, help keep you in the world, and help keep you from becoming self-absorbed, are worth the struggle?

At best, I would categorize a long, putatively deep conversation with an AI chatbot as a guilty pleasure. And I have an idea of how you can improve on it. For listening support from a real human, that’s free, you can try out 7 Cups. It’s a chat platform connecting people with trained “listeners,” who won’t provide advice like an AI would, but also aren’t likely to hallucinate. (I know this platform because I myself am a listener, though I confess I haven’t logged in lately … this post is reminding me to do that.) Meanwhile, if you want to get better at listening to others, you could go through the brief training and become a 7 Cups listener yourself. Obviously this would be an experiment, but then so is AI. And if you find that 7 Cups is no substitute for the validation and ersatz companionship you get from an AI chatbot, you could pledge to offset your AI hours with listening hours on 7 Cups. Call it an AI Offset! (Does that sound too hokey? Am I actually Doogie Howser? Whatever … I’m just spitballing here.)

An AI Skeptithusiast is a syndicated journalist whose advice column, “Ask an AI Skeptithusiast,” appears in over 0 blogs worldwide.

Other albertnet posts on A.I.

• I, Chatbot
• Almost Intelligent – Part I
• Almost Intelligent – Part II
• Smartphones & Artificial Stupidity
• A.I. Smackdown – Moto vs. Cortana vs. Siri
• Could Artificial Intelligence Replace Writers? – Part 1
• Could Artificial Intelligence Replace Writers? – Part 2
• Could Artificial Intelligence Replace Writers? – Part 3
• A.I. Smackdown – English Major vs. ChatGPT – Part 1
• A.I. Smackdown – English Major vs. ChatGPT – Part 2
• Schooling ChatGPT
• Will A.I. Steal Our Jobs?
• Tech Check-In - How Good is the Latest A.I.? - Part I
• Tech Check-In - How Good is the Latest A.I.? - Part II
• What is ChatGPT Great At (and Not)?
• Tech Reflection – Two Sides of AI
• AI Smackdown – ChatGPT vs. Copilot vs. Gemini

—~—~—~—~—~—~—~—~—
Email me here. For a complete index of albertnet posts, click here.