The Talmud says: “The liar’s punishment is that even when he speaks the truth, no one believes him.”
When I asked my doctor if the test he was recommending would be covered by my health insurance, he suggested I check first. I anticipated being put on hold for an indeterminate length of time on the phone.
I took a short cut.
“United Health Care explicitly considers Garden reveal tests to be medically unnecessary,” I read aloud. I sighed. Failing to notice the autocorrect – I had been looking up Guardant reveal tests, not garden – I continued. “If you are referring to a gender reveal test in your garden, or a reveal to your friends of your newly redone landscaping… Wait, what?”
My doctor and I burst out laughing.
What I was actually looking up was a liquid biopsy that measures residual disease. It’s a test my doctor and I had previously discussed, some months back, but at the time, he said I would need tissue from my original biopsy. I had already called and written to the powers-that-be at the NHS multiple times to no avail. My hospital in Birmingham sent electronic test results to my California doctor, but no rice-sized piece of my breast tissue made it across the Atlantic Ocean.
I hadn’t been hugely concerned because my doctor warned me that had the test been returned positive, there would have been nothing to do. I would know cancer DNA was circulating in my body, but that’s about it. All we could do was play the waiting game. Would my breast cancer recurrence manifest in weeks? Months? A year? In my kidney? My spine?
Science can work magically fast; there’s still no cure (in other words, once the metastatic cat is out of the bag, it’s not going back in), but it turns out that now there is something they can do. A medication can target the cancer cells, and, ideally, hold them at bay – perhaps until the next scientific discovery.
Would my insurance pay for this $5,000 test? Here is why I was typing my question into Google, and here is why I was sighing. And then laughing. Yet again, AI had foiled me.
Everyone seems to trust AI. My doctor uses it but assures me his version is a medically specific one and “very good”. My students say they use it only for “polish” but I can see they’re outsourcing their writing – and, frankly, thinking. My colleagues cite its efficiency as they whip up syllabi and letters of recommendation. And hey, obviously, I am no different.
The problem is: it is so bad! This past weekend, I got into a massive fight with Gemini when I put a photo into Google lens and asked the artificial agent to identify the location. The photo was from 2006 and featured a young, pregnant me with my husband and toddler on a trip to southern California. There were only three possibilities as to where the photo had been taken.
Gemini: “This photo was taken at Pismo Pier.” That was not one of the three possibilities.
“I’ve never been there. Where else could it have been taken? Please pay close attention to the concrete barrier and single wooden beam railing.”
“You’re absolutely correct! Thank you for this suggestion. Now I can confirm that this photo was taken at Ventura Pier.”
Huh? I reframed my question. “Is it Santa Monica, Venice, or Santa Barbara?”
Gemini: “You can rule out Santa Monica, Venice, and Santa Barbara. This photo was taken at Pismo Pier.”
“I said I’ve never been to Pismo…”
“Ventura.”
“No!”
“Pismo.”
“I SAID NO!!!”
Luckily, my husband stopped me from driving to Venice Pier – the correct answer, which I finally figured out sans some stupid, insistent, argumentative, non-intelligent agent – and throwing my phone into the Pacific Ocean.
Instead, I used my phone to call Guardant. An AI agent picked up the phone and said I could ask it for anything I wanted.
“PERSON!” I yelled. “PERSON PERSON PERSON.”
Unfortunately, the person proved unhelpful, so I tried calling my insurance company. There, too, I encountered an AI. “PERSON!” I shouted in what was beginning to feel like my new refrain. “PERSON PERSON PERSON.”
Well, at long last, I spoke to a person. And guess what? The person told me that “United Health Care explicitly considers Guardant reveal tests to be medically unnecessary”.
So, what do you know? AI was right. Or else… the person was just reading the answer to the question that he, too, put to AI.
To get more from Life, click here to sign up for our free Life newsletter.
