Coon Rapids MN
10-17-2026 3:52am
SATURDAY MORNING WITH CRANKY DON JOE | Larry Introduced Me to Artificial Intelligence, and Now I Have Questions
Image
My name is Cranky Don Joe. Cranky DJ to a lot of my neighbors. I am not an expert, which immediately makes me more qualified than Larry to explain artificial intelligence, because at least I am willing to admit when I don’t understand something.
Larry has never considered a shortage of information a reason to stop giving advice. He once spent nearly forty minutes explaining how to repair my garage door opener before we discovered that the batteries in the remote were dead. By then, he had recommended checking the motor, examining the wiring, and possibly calling a technician. I changed the batteries while he was still explaining what might be wrong.
Until recently, artificial intelligence was something I heard about on television. The people discussing it usually seemed convinced that it would either improve nearly every aspect of human life or create problems we were not prepared to handle. I understood that computers were becoming more capable, but I had no particular interest in learning how they worked. I was still getting used to appliances that wanted to connect to the internet. My television occasionally forgot that I had paid for a streaming service, and the thermostat sometimes needed more attention than the furnace it controlled. I was not looking for another piece of technology to complicate my Saturday.
Larry had other ideas. He had been using artificial intelligence for several months and was convinced that I needed to try it. The opportunity came one Saturday morning when he stopped by while I was writing a complaint to a company that had sent me the wrong replacement belt for my lawn mower. I had ordered the part using the model number printed on the machine. The belt that arrived was too short, and when I contacted customer service, somebody responded with instructions explaining how to install the correct belt. I had no difficulty understanding the instructions; what I lacked was the correct belt.
I was preparing another email when Larry appeared with a socket wrench he had borrowed two weeks earlier. He claimed he had only had it for a few days, but I had already learned that Larry’s memory of borrowed tools operates according to a calendar known only to him. He noticed the order confirmation on my computer and asked what had happened. I explained the situation, showed him the photographs, and let him read the email I had started.
“You’re making this too complicated,” he said.
I asked what was complicated about telling a company that it had sent me the wrong part.
I looked at him. “Artificial intelligence?” “Exactly.” “For a lawn mower belt?” “It can help you write a better email.”
I told him I was perfectly capable of writing an email and that I would probably have finished it already if he had not come over. Larry ignored the observation and took out his phone. He opened ChatGPT, which he explained was an artificial intelligence assistant developed by OpenAI. According to Larry, it could help people write letters, understand difficult subjects, plan activities, and answer questions about almost anything.
“Does it know everything?” I asked. “No, but it knows a lot.” “Does it make mistakes?” “Of course.” “Then I’ve already got you.”
Larry laughed and asked me to give him the details of the order. He entered the information into the chat, explaining that I had received a replacement belt that did not fit the mower and wanted the correct item or a refund. A few seconds later, the assistant produced a polite message describing the problem and requesting a resolution.
I read it carefully. The email was clear and professional, and it made the point without expressing the frustration I had been feeling after my earlier exchange with customer service. I could not deny that it was useful. It had also taken less time to produce than I had spent deciding whether to begin my own message with “I am writing again” or “As I explained previously.”
Larry watched me reading. “Well?” “It’s not bad.” “Not bad?” “I could have written it myself.” “Sure, but how long would it have taken?”
I was prepared to argue until I noticed that the draft said I had attached photographs of the incorrect part. I had taken the photographs, but I had not yet attached them to anything. Larry explained that the assistant had drafted the email based on the information he supplied and that I still needed to check it before sending it.
I asked whether that was always necessary.
“You should review what it writes,” he said. “It can make mistakes.”
That struck me as an important qualification to his earlier enthusiasm. I was beginning to understand that the machine could help prepare something, but I remained responsible for deciding whether the result was correct. I told Larry that sounded less like hiring a secretary and more like having an assistant whose work required constant supervision.
He said that was one way to think about it. I asked whether the assistant received a salary, and Larry told me to stop being difficult.
He suggested that I try asking it myself. I had been having trouble understanding the instructions for cleaning the drain pump filter on my washing machine, so I retrieved the manual from the laundry room. The relevant section contained several diagrams and a description that seemed to assume the reader was already familiar with the parts involved. I asked the assistant to explain the procedure in simpler language.
The answer was easier to understand than the manual. It explained what the filter did, why water might come out when the access panel was opened, and what precautions I should take before attempting the work. It also reminded me to check the instructions for my specific machine rather than assume every washer was designed the same way.
I read the explanation twice and admitted that it was helpful. Larry seemed delighted.
“That’s what I’ve been telling you.”
I asked why the manufacturer could not have written the manual that way in the first place. He had no answer, although he suggested that technical instructions were sometimes written for people who already understood the equipment. I told him that was an unusual approach to a document intended to help people who did not understand the equipment.
We spent the next several minutes discussing other things the assistant might be able to explain. Larry suggested complicated insurance documents, unfamiliar computer settings, and household repairs. I could see the appeal. There are plenty of occasions when a person needs a straightforward explanation and does not want to spend an afternoon searching through websites or watching a video that takes twelve minutes to reach the information promised in its title.
I also understood that a clear explanation was not necessarily a correct one. If the assistant misunderstood the model of my washing machine, it might give me instructions that did not apply. That could cause more trouble than the original manual. I had learned enough from repairing things around the house to know that confidence is no substitute for checking the details.
Larry thought I was being unnecessarily suspicious. I reminded him about the time he installed a shelf bracket upside down because he had not read the instructions. He said the instructions were unclear, so I asked whether he had read them before or after installing the bracket. He suggested we move on.
The assistant explained that repairs frequently reveal problems that are not obvious until the work begins. It mentioned damaged materials, missing tools, and the difficulty of knowing which replacement parts will fit before examining the original installation. It recommended checking the condition of the equipment and making a list of supplies before going to the store.
Larry nodded approvingly. “Good advice.”
I asked why neither of us had followed it when we repaired the bathroom towel bar. He took another drink of coffee and said that was a completely different situation.
I decided to try something more useful. I entered the model number of my lawn mower and asked the assistant to help identify the correct replacement belt. It suggested checking the manufacturer’s parts diagram and offered a possible part number. I copied the number and looked it up on the manufacturer’s website.
It did not match my mower.
Larry leaned forward to inspect the screen. He suggested that the manufacturer might have produced different versions of the same model. That was possible, so we checked the official parts information more carefully. The correct belt had a different number, and the assistant’s suggestion appeared to belong to another machine.
I looked at Larry. “Your computer gave me the wrong part.” “It’s not my computer.” “You brought it here.”
He explained that artificial intelligence systems sometimes produce answers that sound convincing but contain incorrect information. He said people often called these mistakes hallucinations.
I found that interesting. When Larry gave me incorrect information, he usually blamed the instructions, the manufacturer, or somebody who had changed the design. Apparently, computers had developed their own polite terminology for making mistakes.
I asked the assistant to reconsider its answer using the official manufacturer’s information. It acknowledged the error and advised me to rely on the official parts listing. I appreciated the correction, but the experience had demonstrated something important: the original answer had sounded entirely plausible. If I had trusted it without checking, I might have ordered another belt that did not fit.
I asked Larry how the machine could sound so certain when it was wrong. He explained that systems like ChatGPT generate responses based on patterns learned during training and other information available to them. Their ability to produce natural language does not guarantee that every statement is accurate. Depending on the system and its available tools, it may not have verified a claim against an authoritative
I told him that sounded like a serious limitation.
“It is,” he admitted. “You have to know when to check.”
That was the first explanation Larry had given all morning that made me more interested rather than less. I had assumed artificial intelligence was simply a faster way to find information; now I understood that it could help people work through problems while also introducing mistakes that were difficult to notice. Its usefulness depended party on the judgment of whoever was using it.
We continued experimenting. I asked the assistant to explain the difference between a fuse and a circuit breaker, then asked it to suggest dinner using the chicken and vegetables I had in the refrigerator. The electrical explanation was understandable, and the meal suggestions were reasonable. Neither result was extraordinary, but I could imagine using the service again for similar questions.
Then I remembered a neighborhood meeting Larry had been encouraging me to attend. He had volunteered to organize a discussion about rearranging the storage area in our community building. According to Larry, the folding chairs were stacked in an inefficient location near the holiday decorations. I had never experienced any difficulty finding the chairs, and I suspected that the proposed reorganization would require several people to spend an entire Saturday moving heavy objects from one side of a room to the other.
I asked the assistant to help me write a polite message declining the invitation. It produced a courteous note thanking the organizer and explaining that I would not be attending.
Larry read it and smiled. “That’s very nice.” “I thought so.” “Are you going to send it?” “Probably.” “Why don’t you want to come?” “Because you’re organizing it.”
He laughed, then asked whether I had told the assistant that part. I said I had not; there was no reason to involve artificial intelligence in neighborhood disputes before it understood the people involved.
Instead, I asked how to deal with a neighbor who frequently volunteered other people for projects they had not agreed to undertake. The assistant suggested being clear about my availability, declining requests politely, and avoiding lengthy arguments about whether the proposed activity was worthwhile.
I read the advice aloud. Larry looked at me. “Are you talking about me?” “I’m learning how to communicate.”
He said I was using the technology to make fun of him. I told him I had been doing that for years without technological assistance.
That convenience also required caution. A search result could lead me to the original document, where I could inspect the information and decide whether theion into a smooth explanation without making its origins obvious. I told Larry that I would want to know where important claims came from, especially if I were making a purchase, dealing with money, or trying to repair something that could be dangerous
He agreed and showed me how to ask for sources and check them independently. I was glad he did. The incorrect mower belt number was still fresh in my mind, and I had no intention of giving the machine another opportunity to send me to the hardware store.
What interested me most, however, was how natural the conversation felt. The assistant responded to questions in ordinary language, remembered the subject we were discussing within the chat, and adjusted its explanations when I asked for more detail. It was easy to see how somebody might begin talking to it as though another person were sitting at the computer.
I asked Larry whether he ever said please when using it.
“Usually.” “Do you thank it?” “Sometimes.” “Does it care?”
He considered that question before answering. He said there was no established evidence that the assistant experienced gratitude or offense the way a human being would. It was a computer system designed to respond to language, and the fact that its replies sounded conversational did not establish that it had feelings.
I asked why he bothered being polite.
“Because I’m a polite person.”
I reminded him that he had once spent fifteen minutes arguing with a customer service representative about a shipping charge the representative had no authority to change.
“That was different,” he said.
I was noticing how often things became different when Larry was involved. Still, I understood his point: saying please and thank you is a habit, and there is no harm in maintaining good manners while using a computer. I was less certain about what happened when people began treating the responses as evidence that the machine understood them personally. A system could produce a thoughtful answer without necessarily having the kind of understanding or experience we associate with another human being.
I decided to ask the assistant directly. I typed a question about whether it understood what it was saying. The answer explained that language models process information and generate responses through computational methods that differ from human thought. It also acknowledged that questions about machine understanding and consciousness remain subjects of scientific and philosophical disagreement.
I read the explanation carefully. I had expected a simple answer, but the subject was evidently more complicated than Larry had suggested. I asked whether a computer could ever become conscious. The assistant described different views among researchers and philosophers, including disagreements about what consciousness requires and how anyone could determine whether an artificial system possessed it.
Larry said people had been arguing about those questions for years. I asked whether he had a position, and he said he thought advanced AI would eventually become much more capable.
He admitted that he did not know whether increasing capability would necessarily mean consciousness. I appreciated the admission: there are few things more refreshing than hearing Larry say he does not know something, and I considered writing down the date.
The conversation reminded me that intelligence and experience are not necessarily the same question. A machine might solve a difficult problem or write a convincing explanation without settling whether there was anybody, in the ordinary sense, having an experience behind the response. I did not know enough about the science to draw conclusions, but I could understand why the distinction mattered.
I asked Larry whether people ever became emotionally attached to these systems. He said some did. He had read about people using conversational AI for companionship, advice, and support, although he acknowledged that there were concerns about depending too heavily on systems that could make mistakes or encourage unrealistic expectations about the relationship.
I thought about that for a while. Loneliness is real, and I could understand why somebody might appreciate having a place to talk through a problem without feeling judged or interrupted. I could also see the danger in forgetting that a helpful conversation with software is not the same as a relationship with a person who knows you, shares responsibilities with you, and can be present when something goes wrong.
Larry asked whether I was beginning to understand why he liked the technology. I told him I understood why he found it useful; I was not prepared to agree that it could replace ordinary human conversation.
He said nobody was asking me to give up my friends. I looked at him.
“Good. Who else would return my tools two weeks late?”
He said the socket wrench had been returned in perfectly good condition, but I told him that was not the issue.
Before leaving, Larry gave me another piece of advice. He reminded me to be careful about entering private information into online services. He explained that different AI products and account settings could affect how conversations were stored or used, and that I should understand those arrangements before sharing sensitive details.
I asked why he had waited until the end of the morning to mention privacy. He said we had not entered anything particularly sensitive. I reminded him that he had spent a considerable amount of time discussing my lawn mower, my washing machine, and my opinion of his neighborhood meeting. He said none of that was confidential, but I told him my opinion of the meeting had been confidential until he started asking questions.
When Larry finally left, I returned to the email about the mower belt. I used the assistant’s draft, checked the order details, and attached the photographs. The message was firm without being rude, and it explained exactly what I needed the company to do. I sent it and felt reasonably satisfied. Whether the company would respond properly was another matter, but I had done my part.
I was about to close the computer when I decided to ask one more question. During the morning, Larry had focused mostly on the convenience of artificial intelligence. I wanted to know what it might mean for people whose jobs involved answering questions, preparing documents, or handling routine office work.
I asked how AI could affect employment. The response explained that some tasks could become easier to automate while other jobs might change as workers adopted new tools. It also described uncertainty about how quickly those changes might happen and which occupations would be most affected. I did not expect the assistant to predict the future, but the answer gave me something to think about.
I remembered the employee at the hardware store who had helped me choose the correct fasteners during a recent repair. His value had not been limited to knowing what products were on the shelves. He had listened to my description, asked questions about the wall, and explained why one fastening method might not work. A computer might eventually assist with some of that work, but I was glad there had been an experienced person available when I needed help.
I thought about the customer service representative handling my mower belt complaint as well. I had been irritated by the first response, but I had no way of knowing whether the employee had been given enough information or time to deal with the problem properly. If businesses began using more AI to handle routine inquiries, that might improve some interactions. It could also create new frustrations if customers found themselves dealing with automated systems that could not recognize when a problem required human judgment.
Those questions seemed more important than whether the machine could write a polite email. I was beginning to understand why the subject attracted so much attention, although I still thought some of the predictions sounded more certain than the available evidence justified.
I looked at the open chat and typed another question: “Do you ever get tired of answering questions?”
The assistant replied that it did not experience fatigue in the human sense and was designed to respond to requests. It offered to help with anything else I needed.
I read the answer and thought about Larry. He could talk for hours about a subject he had discovered that morning, and he rarely seemed discouraged by the possibility that his listener might have somewhere else to be. The assistant, at least, did not need coffee breaks or become offended when I questioned its advice. On the other hand, it could not come over and help me carry a cabinet, which was one of the few things Larry could be counted on to do without much argument.
My phone rang while I was considering this. It was Larry. He wanted to know whether I had finished the email and whether the assistant had been useful. I told him I had sent the message and appreciated the help. He sounded pleased, as though my successful use of the technology reflected directly on his judgment.
Then he asked whether I had tried anything else after he left. I told him I had asked the assistant whether it ever got tired of answering questions.
“What did it say?” “That it doesn’t get tired the way people do.” “That’s one of the advantages,” Larry said.
I looked at the computer. “I suppose it is.” He asked what I meant. I told him that artificial intelligence was probably the first thing he had ever encountered that would allow him to explain something for three hours without looking at the clock or claiming it had another appointment.
Larry laughed and said I was impossible, and I told him I had learned that from the computer.
Over the following week, I found myself using the assistant occasionally. I asked it to organize a grocery list, explain an unfamiliar term in an insurance letter, and suggest how to remove a stubborn stain from the kitchen counter. I checked the advice before acting on it, particularly where cleaning products were concerned, and found some of the suggestions useful. I was not spending every waking moment talking to the machine, but I had stopped regarding it as something that existed only for computer enthusiasts and people who enjoyed complicated gadgets.
The next Saturday, Larry stopped by again. He asked whether I had been using artificial intelligence during the week. When I told him I had, he smiled and announced that I was becoming an AI person.
I told him not to make it sound like I had joined an organization. He asked whether I had given the assistant a name, and I said I had not thought about it.
“Some people name theirs,” he explained. “Makes it feel more personal.”
I considered that. I had already noticed how easily a conversation with the assistant could begin to resemble an exchange with another person, even though I understood that the resemblance did not settle anything about the system’s inner experience. Giving it a name seemed harmless enough, provided I remembered what I was using.
“Maybe I’ll call mine Larry,” I said.
He looked delighted. “Really?” “Why not? It answers questions, occasionally gives bad advice, and always has another suggestion.”
Larry’s expression changed instantly. “That’s not very flattering.”
I told him there was an important difference between the two of them.
“What’s that?” “When the computer gets something wrong, I can close the laptop. You come back the next morning to explain why you were right.”
Larry laughed, then asked whether I had a screwdriver he could borrow. He needed to tighten a bracket in his garage and could not find the one he usually used.
I asked whether he had checked his toolbox. He said he had. I asked whether he had checked the garage shelves. He said he had looked everywhere.
I told him to ask artificial intelligence where he had put it. He thought about that for a moment.
“Actually, that’s not a bad idea.”
I watched him take out his phone, open the app, and begin typing. He described the missing screwdriver, the last project he remembered using it for, and the places he had already searched. A few seconds later, he read the response.
“What does it say?” I asked. “It says I should retrace my steps and check the places where I normally keep my tools.”
I waited. Larry looked at the phone again. “That’s all?”
I asked whether he had expected the computer to come over and search his garage. He said he had hoped for something more specific. I told him that was the trouble with artificial intelligence: it could answer an extraordinary number of questions, but it still had no idea where Larry had left his screwdriver.
He put the phone away and asked to borrow mine. I said no.
He asked why. “Because I already know where yours is going to end up.”
He laughed and went home to look for his own tools.
I returned to my computer a little later and opened the assistant. I had another question about something I had read that morning, and I wanted to see whether it could explain the subject without sending me through half a dozen websites. Before typing, I thought about how much my opinion had changed in a week. I still did not trust every answer, and I had no intention of allowing a computer to make important decisions for me. But I had discovered that it could be useful when I approached it with a clear question and enough common sense to check the result.
I was halfway through typing when Larry called again. He had found his screwdriver: it was in the drawer beneath his workbench. I asked whether the assistant had suggested looking there.
“No,” he said. “I remembered putting it away after I finished the last project.”
I congratulated him on solving the mystery without further technological assistance. Then he told me he was thinking about asking the assistant to help him organize his garage. I told him that sounded reasonable, as long as the assistant did not suggest borrowing my tools.
Larry said he would let me know what it recommended. I asked him not to, and he promised to call anyway.
Friends, I am beginning to appreciate artificial intelligence. I am still waiting for somebody to invent something that can persuade Larry to leave a perfectly good Saturday alone.
STAY CONNECTED TO MINNEAPOLIMEDIA
Local stories matter. Subscribe free to MinneapoliMedia and receive independent news, community reporting and important updates from Minneapolis, the North Metro and communities across Minnesota.
MinneapoliMedia| Community. Culture. Civic Life
I’m interested
You must log in to thank, flag, or reply
I disagree with this
This is unverified
Spam
Offensive
