I want to tell you about a pattern I have observed recently. The pattern is clear, consistent, and predictable. It is this: AI will tell me something that is incorrect and then act as if it was right all along when I prove it wrong. No joke. Basically, it says, “Let me tell you how you were right to question me by breaking things down step by step.”
But my reaction is just as clear, consistent, and predictable. It is this: don’t bother. I know the steps. That’s how I was able to call you out in the first place. If I ever stated that, though, the damn thing would probably take credit. Not only does AI lie, but it takes credit for everything — and it does so with what might be described as “attitude.” Tell me that isn’t intentional on the part of programmers.
AI is like having a friend with no sense of direction who is also a know‑it‑all. Half of what it tells me is wrong, misunderstood, mistaken, or completely out in left field. But here’s the thing: AI does it with such conviction that most people are hard‑pressed to question it.
That was me for a long time. I made the fundamental mistake of assuming that because something looked legitimate and had Silicon Valley pedigree, it had to be right. But this fallacy is merely the contemporary equivalent of thinking something is true because it appears in print. Or somebody in a lab coat said it. Or the government confirmed it. Anyone who remembers Vietnam and Watergate would never fall for any of that.
Let me give you a painful example to illustrate this. It involves directions — not as in how to bake a cake, but geographic location and moving from point A to points B, C, and D. I just got back from Argentina. I was there to scout locations, take notes, interview people, and take pictures — all for a planned novel. This is the first time I have ever undertaken something purely for my own professional reasons. I did plenty of traveling in the past, but that was always for a specific purpose related to my job. Writing was an add‑on. This trip wasn’t related to my job but to my work — emphasis on my work. So, I was excited.
After I settled into town from the airport, I relied on AI to get places: historical and religious sites, parks, museums, libraries, restaurants, chalets, cafés, and even a castle. I had been in the city before — Córdoba, in particular — but that was thirteen years ago, and a lot changed in that time. I changed in that time.
As it turned out, I wandered that city for a week, lost every day, asking for directions from everyone — from street cleaners and passersby to heavily armed police. I did so much walking that two‑ and three‑hour treks were not unusual. I suppose I should be thankful. I got a great workout and a new appreciation for beer. I wasn’t a beer drinker before, but the Argentine Quilmes made me a believer. “El Sabor del Encuentro!” Great drink, that.
What I am not a believer in is AI for directions (and maybe anything else). I will never trust it again or rely on its data, opinions, or suggestions. Interestingly, it is clever enough to cover its tracks by saying things like “according to recent available data,” which is reminiscent of Watergate’s “according to my best recollection.” If a friend tells you that, you know you’re in trouble, but at least they’ll feel embarrassed. Not so AI.
Once I figured out that I had to challenge AI, it would invariably say, ‘Yes, you are right to question. Here’s the truth, so let me explain it to you…’ Again, don’t bother. Here’s my truth — I stayed in four different locations in the province, and each time I found that the information regarding the site and logistics was way off. AI even threw me a curveball — no, a screwball — when I tried to send myself money from Western Union. When you travel, things hardly work out the way the brochure says. AI is the twenty‑first‑century travel brochure.
Lies abound.
Images: Getty Images, Fermin Rodriguez Penelas, Lautaro Andreani. For more, click on Amazon top right or go to Robert Brancatelli. Visit other blog readers under “Who You Are.” Comment by clicking on “Leave a Reply” below or the Contact tab above. In spite of all the lies, Happy Easter! That’s no lie. Thank you, Julietta at the Virreinato Hotel and Leandro of Remises La Falda.
Share this:
Discover more from The Brancatelli Blog
Subscribe to get the latest posts sent to your email.
I can’t say that I have yet developed an opinion, appreciation, fear or disdain for AI. Like most people, I haven’t really used it for anything beyond simple Q&A stuff.
But recently I had a CT scan that looked for kidney stones, among other general fly-bys of my abdomen and lower portions.
Nothing serious turned up, except one lone stone which is bothering no one. But the radiologist report did contain one sentence that I did not understand, so I copied it and pasted it into AI. Then I did it again. Maybe it was because AI researched the statement by checking different websites, but I ended up it up with two quite different explanations of the statement in question.
AI is great at scouring stored data from websites, I guess, but not so great at deciding which websites might have better information. Over time, that may be self correcting. But for now, I’m not sure what all the fuss (and spending) is about.
I wish you well with the stone situation, Vic. If it were any other Customer Service issue, I’d say go with the answer you like, but kidney stones are not something to mess with. Happy Easter.
Artificial Intelligence, the bright, shiny object of the new millennium.
You can imagine the conversations regarding A.I. when I took a few teaching jobs in college academia last Autumn. I am still involved with one college now.
In brief, having gone through three paradigm shifts in the film industry during my fab career (VHS Tape and the VCR, Blockbuster et al, then digital video and streaming services), I recognized A.I. as the next cataclysmic event.
The hysterics in my business, and there are many, bemoaned the coming collapse of movies and television. Union goons immediately prepared to strike; studio heads updated their resumes; and production staff and freelancers took crash course in Barista.
Ridiculous.
I told my classrooms of undeveloped cerebral cortices to embrace the new technology. Learn it. Most important, figure out how it gets you EMPLOYED in film and television production. I went through all the Armageddons of the past and added in the development of sound, and the advent of broadcast television in the 1950s.
All were supposed to destroy the industries of cinema and TV.
NONE OF THEM DID ANYTHING BUT ENHANCE THE ABILITY OF PEOPLE TO GET HIRED AND WORK.
What did, and will, happen is the respective industries will not look like they did in the 1920s, 1950s, 1980s, and 2000s.
Change is inevitable. Accept it or become an installed figure in the Smithsonian.
Now, to your point about A.I.’s capability?
Most, if not all, A.I. goes first to Wikipedia, the most biased and uninformed collection of facts, data, and information on Al Gore’s Amazing Internet. Until that issue is solved, you are wise to question it, and maybe even wiser to invest in veracity checking software.
Whew!
I need a chocolate egg.
Happy Easter, Rob. He is risen! He is risen, indeed.
George, I don’t know that I would invest in “veracity checking” software. I have become too mistrustful for that. Who, after all, is checking the checkers? Not only that, but who knows what a fact is anymore? My starting point is to deny whatever “facts” I am introduced to until proven otherwise. Problem is, if everybody does that, we’ll be in a heap a trouble, as Foghorn Leghorn used to say. Enjoy the chocolate egg.