The numbers are finally shifting. After a rough patch in 2025, U.S. job market data for Q2 2026 shows a sliver of green. Weekly unemployment claims dropped 12,000 in late June, landing at 215,00. Analysts expected higher. The Associated Press called it a beat.
Still. 7.3 million people are unemployed.
Most of them start the grind with a resume. Then they feed it into a chatbot. Because why type when an algorithm can do it?
The catch is the employers use AI too. It sprints through stacks of CVs before a human eye ever touches the paper. So you’re playing poker with a bot that’s reading against a bot that’s writing.
To sort through the noise, I spoke with Shaun Pichler. He’s a professor at CSU Fullerton. He edits the Journal of Occupational & Organizational Psychology too. He knows the machinery inside the HR black box.
Do robots really read everything?
People think the algorithm sees all.
They’re wrong.
“It depends on the job and the company.”
Entry-level roles get the digital screen. Executive suites? Usually not. Pichler says big companies use AI because the volume is unbearable. Small firms? Maybe they still have time for manual reviews. Recent surveys show less than half of organizations use AI recruiting at all. Plus referrals always slip through the front door. AI never touches those.
Does AI favor AI-written resumes?
This gets weird.
Language models are trained to sound like… language models. If your resume is generated by an LLM, it matches their training data. Some studies suggest AI tools might actually prefer content they could have written themselves.
Is it bias? Sort of.
One conference paper hinted at it. But real evidence is scarce. The tech moves faster than the research. Pichler predicts we’ll know more in two years. Right now? It’s a hunch. But a dangerous one if you ignore it.
Don’t pay strangers for optimization
Websites pop up daily promising to “optimize your CV for AI.” They charge monthly fees. They sell keywords.
Pichler’s take is blunt.
There is almost zero evidence that these vendors know what they are doing. The models change every week. Their static guides are outdated before they print. He suggests using free LLMs instead. ChatGPT or Claude. Feed them your draft. Let them argue. Don’t pay a third party to guess. It’s a waste of money. Mostly.
Tailor or starve
Want the bot to click “accept”?
Mirror the posting.
Steal the exact phrasing from the job description. Use the keywords. If the ad asks for “strategic planning,” don’t write “big picture thinking.” Use the words they programmed the system to catch.
Quantify. Numbers stick. “Managed a team of five” beats “led a group.”
And here’s the labor. You have to do it yourself first. Dump your achievements into an LLM alongside the job description. Ask it to merge them. But watch it closely. AI is lazy. It’ll hallucinate skills you don’t have if you let it run wild.
It’s not magic. It’s homework. Just with a slightly less judgmental editor.
