Artificial intelligence is far from infallible, but you'd think it would at least be smart enough to correctly match a politician's headshot with their name.
As BC NDP candidate Honieh Barzegari recently discovered, however, that is not always the case.
With just days to go before the provincial election, the West Vancouver-Capilano hopeful discovered that ChatGPT was confusing her for other women — not once, but multiple times.
In an interview with Sitka Media, Barzegari explained that after returning from a long day of door-knocking on Sunday, a friend and fellow canvasser informed her that people he'd spoken with had found her via ChatGPT.
"They took a screenshot and sent it to him, and he forwarded a screenshot to me because the woman in the picture is absolutely not me," she said. "Then I started to ask ChatGPT myself, and I asked lots of my other friends, not only in BC, but across the globe, to ask the same question: who are the West Vancouver Capilano candidates in the BC provincial election?"
Barzegari, a physician with 20 years of experience in healthcare, business, and public policy, said ChatGPT provided numerous pictures, none of which were her, and some of which may have been AI-generated.
While the AI platform managed to correctly provide an image of BC Conservative candidate Caroline Elliott, OneBC candidate Naomi Chocyk was also replaced, though at least her headshot was of party leader Dallas Brodie.
Barzegari, who ran for Vancouver city council in the 2022 municipal election, made it clear that the error could not have been made due to a lack of available images.
"If people google only my first name, it will show my pictures and my real identity," Barzegari explained. "At the same time, I am a candidate with BC NDP, and my picture is on the BC NDP's website — and lots of other resources are available about me online with my real picture."
She said she tried to tell ChatGPT to retrieve her image from the BC NDP website, but her efforts came up fruitless.
"It felt so bad," Barzegari said of the experience. "I felt a very bad feeling in my gut, started to have palpitations, and I started to sweat because the first thing which came to me is I have been misrepresented."
She added that knowing there was misinformation about her online made her feel "vulnerable."
On Tuesday, Barzegari posted a video to Instagram detailing the situation. It has since received thousands of views and been shared countless times.
"Before sharing, believing, or making decisions based on AI-generated content, take a moment to verify the facts using reliable, independent sources," she warned. "AI is a powerful tool, I use it myself however, critical thinking is still our responsibility."
As of Thursday evening, ChatGPT is still displaying the incorrect image of Barzegari. When the same question was posed to Grok, Claude, and Gemini, however, the responses were accurate.
