I struggle with the concept of artificial intelligence — perhaps because in my 77 years I’ve had to cope with what intelligence I can perceive and distill from teachers, friends, art, science and the natural world.

I’ve had to consider and process these inputs, place them in context and sort out what’s true and worthwhile and what’s not.

Technology, like a manual tool, is supposed to make life easier. But if the energy required to learn and use the tool exceeds the energy it saves using it, it is not a useful tool, as I mentioned in an earlier column.

The explosion of AI software that tries to satisfy our natural human need for help and advice is said to save billions for industry but, so far, has done little for humans seeking answers.

Will AI make the world a better place or simply make business more profitable? In deploying AI, will businesses prioritize customer and human needs or company policy and profit?

The company I cofounded and ran for 25 years provided local call center service to major broadcasters and several publishers. At our peak, we employed 80 agents who were thoroughly trained on our clients’ products. Our goal was to answer all calls by the third ring and we largely succeeded. An agent might answer, “Thank you for calling The History Channel; how can I help you?” Agents knew intimately the client product lines and had screen backup for all product inquiries.

We sold the company in 2008 to a private equity firm that proceeded to wring out costs, transferred the call center to Southeast Asia, where agents were answering calls for dozens of other companies, and tried to implement primitive AI. The customer service deterioration was so marked that clients began leaving in droves and the company and its 175 jobs were gone within another few years.

Have you ever called the toll-free number helpline for QuickBooks, Adobe, your bank, the IRS, Medicare, or any of the other countless services on which we increasingly rely? Studies show that 15% of customers hang up after being on hold for 40 seconds. When a customer succeeds in complaining to a company about its product or service, does that information ever reach decision-makers?

Remember when you could call a travel agent and be booked from Burlington to Iceland and back and be sent the details of your trip. Now, you must work directly with the airline or a travel aggregator — good luck with that.

One persistent airline customer put a company's phone line to the test after being placed on hold for two hours. He decided to wait and see how long the business would keep him on the line. Some 15 hours later, the call was finally pushed through and he was notified that his original request (which put him on hold in the first place) had been denied due to an error.

My own record to date is waiting 47 minutes to get a human to answer the Quicken helpline in the Philippines. We then talked for 128 more minutes after I gave her online access to my computer, and she finally acknowledged that the system was having “issues” and that I should call back later.

As it turned out, I lost the ability to process 22 years’ worth of financial records. Luckily, I had backed up the data.

To apply to American colleges, our thoroughly literate, trilingual Serbian foreign exchange student, Mina, who has earned straight A’s at Champlain Valley Union High School as a senior, needed to take a standard test administered online to determine her English fluency.

The first effort failed because of a scheduling confusion regarding Euro-military time standard vs. American time.

The second effort, which she started a half-hour early, was on her home computer. It required extensive rewiring. She passed the technical test and then, minutes before starting, was instructed to download an extension. After the download completed, and as the test was starting, the extension created onscreen ambiguities. She was told that, if the extension caused a problem, she should click on the "help" icon. There was no help icon. We could not resolve the technical issue, so Mina was disconnected from the site.

We set up the third trial at CVU in its computer lab with a staff tech helping. Mina got there an hour early. They resolved the myriad technical issues, which included having to move to another site in the school, and the test began. She completed the first half and was told to take a 10-minute break. She did, returned, and began to start again, but the test site told her that she could not complete the test because she’d taken her cellphone to the bathroom (to text me that the test was working).

She was denied access for the second part and called me in tears. So after three trials and charges for each, she could not access the test required of all foreign exchange students wishing to apply to an American college — our country’s loss.

I recently went to a restaurant that had no menu. I had to get the Wi-Fi password from the waitress, enter it into my cellphone, scan a QR code, and read a linear menu that listed my lunch options and prices. I’ll never go back there. Maybe I’m just old.

Going to Canada to pick up my wife at the airport, I filled out the ArriveCan online entry form, a prerequisite for any traveler to Canada now. I filled it out with all my personal data and then it asked for a credit card. I was surprised at the charge, but heedlessly entered my Amex card. That failed. So, I entered a Visa card. I then got an email saying my payments had not worked and an agent would be in touch within 72 hours. Planning to be in Canada in 60 hours, I responded that I needed help before then.

The answer came from Russia and both cards had been ripped. I canceled both and went to Canada with no credit cards. The fake site perfectly mirrored the legitimate Canadian one, just adding the card capture.

I am an early adopter. I wrote my first novel on a portable computer the size of an overnight case 35 years ago. But I worry about our aging population, among whom I count myself. Many now lack access to the once-common human help (“navigators,” I think we called them). How does a person of Medicare or Social Security age navigate on their own the myriad complexities of using systems designed essentially to minimize personnel costs?

Thanks to our local telephone company, we have fiber-optic cable to our home. But we know many Vermonters who still have dial-up.

If health care continues its migration to telemedicine, helplines don’t get answered, and emergency room waits get longer, what happens to the many whose lifelong access to medicine has been through a local primary care doctor? How many people will simply give up? At a certain age, the energy and perhaps the capacity to relearn how to cope with all of life’s tasks wanes.

Artificial intelligence holds promise in many areas, but if it’s designed solely to wring human costs out of business enterprise and gather personal data to be remarketed at a profit, it will not serve humanity well.

AI must be designed with people (consumers) in mind and with the knowledge that information technology can be designed either to help or to exploit, depending on how it is deployed.