Get the picture, using AI?

I am always suspicious of the promises of emerging technology. Existing tech allows me to do so much more than I could five years ago. Why bother? My writing, conversation, photography, research, and most recently sports entertainment all benefit from hi-tech resources and solutions. Social media algorithms however suggest I need to go further in my hardware and software exploration. They bombard me with the next generation of tech innovations promising so much more convenience and efficiency including artificial intelligence (AI); It’s calling me by name always and everywhere.

Seated at a lunch table recently a friend showed me a picture of a family gathering. I can’t share it with you here for privacy reasons. I can however share with you what AI thinks is going on in the picture. Here is an AI description of one particular family photo:

A large group of people, approximately 75 individuals, are gathered together outdoors for what appears to be a celebratory or reunion event. The group includes a wide range of ages, from young children (around 5-7 years old) to elderly adults, possibly in their 80s or 90s. There are both men and women present, and the group is diverse in terms of age and appearance, but most appear to be of similar ethnicity (predominantly white).

The photo is taken outside on a well-kept green lawn. The background is lush with greenery, including dense bushes, leafy trees, and a wooden fence running along the back. The area feels private and enclosed, likely a large backyard or garden. The weather appears mild and pleasant, as everyone is dressed comfortably for an outdoor gathering. The lighting is natural, suggesting the photo was taken during the day, possibly late morning, or afternoon.

The group is arranged in several rows, with taller people and adults standing at the back and shorter individuals, including children, at the front. Some people are kneeling or sitting in the front row. There is a mix of formal and casual attire: some men are in jackets, shirts, and ties, while others wear sweaters, polo shirts, or shorts. Women are dressed in blouses, cardigans, dresses, and pants, with a few in bright, patterned clothing. Children are dressed casually, with some in hoodies, t-shirts, or sportswear.

Many participants are wearing name tags, indicating this is an organized event where not everyone may know each other—possibly a family reunion, alumni gathering, or community event. The mood is warm, friendly, and inclusive. Most people are smiling, and some are standing close together with arms around each other or holding hands, suggesting familiarity and affection among the group. There are a few notable details:

– An elderly man at the front left is holding a cane, and another is holding a glass of wine.
– Two young women or teenagers are kneeling at the front right, smiling and holding hands.
– Several children are grouped together at the front right, some with their arms around each other.
– The group is diverse in terms of age but appears to be a close-knit community or extended family.

Overall, the image captures a joyful, multi-generational gathering in a beautiful garden setting, with everyone coming together for a special occasion. 

So there it is, according to AI. Do you “get the picture?” Seeing the picture myself, I am impressed. The only thing missing is an assessment of body type and identification of political persuasion. If a picture is worth a 1,000 words, well, we got 392. Not bad for a first outing.

Still thinking about my own AI use especially in my photography – you can now remove, adapt, or recreate anything with just a little practice. Given the manipulative potential of visual AI I now realize that I can never completely trust any image (mine or that of others) as truth-telling.

As I mulled over the family photo analysis above I stumbled across a piece in the Washington Post: AI bots took our tough reading test. One was smartest — and it wasn’t ChatGPT. Geoffrey A. Fowler describes a project that focuses on AI textual analysis and composition:

All of the most popular artificial intelligence chatbots have the ability to upload and summarize documents, from legal contracts to an entire book. The tech promises to give you a kind of speed-reading superpower. But do any of the bots really understand what they’re reading?

To figure out which AI tools you can trust as a reading assistant, I held a competition. I challenged five bots to read four very different types of writing and then tested their comprehension. The reading spanned the liberal arts, including a novel, medical research, legal agreements, and speeches by President Donald Trump.

All told, I asked 115 questions about the assigned reading to ChatGPT, Claude, Copilot, Meta AI, and Gemini. Some of the AI responses were astoundingly good. Others were so clueless they sounded like “Seinfeld’s” George Costanza.

I encourage readers to scan the original post as the detailed scoring and analysis are relevant and interesting. A few highlights are worth noting here:

Literature was the worst subject overall for the bots. Only Claude got all the facts right about Chris Bohjalian’s 2025 Civil War love story, “The Jackal’s Mistress.”

Gemini, which wrote very short responses to our questions, was most often guilty of . . .  inaccurate, misleading, and sloppy reading.

On average, all of the AI tools scored [best] at analyzing scientific research. In our test of two papers co-written by judge Eric Topol, less than two points separated the best and worst performances.

[Concerning Trump’s speeches] Trump’s speeches can be so meandering, they’ve garnered their own stylistic nickname: “the weave.” Cat Zakrzewski, a Washington Post White House reporter, judged whether AI could make out what he was actually asserting and analyze what it meant. [It seems that along with most humans, AI was and remains baffled by Trump, the man, and the speech.]

So is there an overall winner? Claude edged out ChatGPT and left the others in the dust. Overall winner Claude was also the only model that never hallucinated.

One flabbergasted evaluator exclaimed “Okay, I’m done. Whole human race is. Stick a fork in us,” he noted. This same commentator continued: “AI summaries frequently left out important information and overemphasized the positive (while ignoring the negative). Too often, you could “really see the robot hiding behind the human mask” pretending to be an expert in something it didn’t actually understand.”

If similar reports urge caution around accuracy, there are other concerns with AI itself. The demands for power and water are massive with AI. When I ask Adobe Lightroom to remove an object, a mainframe somewhere in the continental US responds to my request. AI as currently deployed is unsustainable. That said, this technology of convenience is here to stay. Along with future generations, we must earn to live with it.

Get the picture?

---

Visit the takenote.ca HOME page for a colourful display of hundreds of other blogs which may interest or inspire you.