Lies, lies, lies

It looks like it is time to revisit the topic of critical thinking.

Background story

A friend sent me a Twitter/X video where a young lady starts off with “…we are actively gaslight by AI on a daily basis…”

She then goes on to talk about a recent Stanford research on AI hiring tools that a job seeker encounters.

She teased: the Stanford study ‘…is uncovering a massive web…(of deceit???) that’s answering the question that all of us are asking, “Why does my resume keep getting rejected for jobs I am completely qualified for?”‘

She described how the process works: “…you apply to a job and then the company sends your resume to a third-party AI tool and that AI platform gives you a score and then it sends that score back to the employer.”

And then the scary part: “But the score doesn’t disappear or get regenerated with every application you send. Sometimes it holds on to that score for 330 days so when you apply to another company that uses the same platform, you are not getting a fresh start…you’re just getting that same score.”

Her conclusion? “…the rejection rate was a little too high to be coincidental, so they are calling it systemic rejection.

My response

My initial response was what a terrible thing happening and it is all AI false! Well, no but a lot of people are against AI.

I did think the story was terrible and wanted to learn more about what was going on. I always want more details and learn how prevalent the news is in the environment. Are the reporters reporting on this? What kind of reporters? What do they all converge on? What are the details popping up here and there but not reported on by everyone?

So I did a browser search on “stanford study on ai hiring tools”.

The study’s conclusion of “racial” bias jumped out at me, a very different take on the conclusions. The young lady said nothing about racial bias.

Next, when digging into a few articles and into CoPilot’s summary, I learned resumes played no role. Instead, Pymetrics tested the candidates through games.

Finally, none of the articles mentioned anything about scores being held for 330 days.

Are the articles talking about the same Stanford study as the young lady? They did include some numerical metrics the lady used: 4 million applicants, 156 employers.

So, I went to Google Notebook and asked it to find me sources on “Stanford study on AI hiring tools using resumes”. It brought up 10 sources which on first blush looked like articles talking about racism.

I asked Notebook if any of the sources discussed Stanford study, AI hiring platform and resumes in one article? And it said no.

But it was nice enough to point out the original study paper.

Original research paper

The paper took about a couple of hours to read and here’s what I gathered:

• 4 million applicants and 156 employers were part of the study.
• The AI hiring tool studied was Pymetric but the researchers emphasized there was a more popular AI hiring tool (HireVue) used by more companies.
• No resumes were sent to Pymetric. Instead, the applicants played Pymetrics’ 12 to 16 games.
• Each game received a score.
• The combination of scores added up to a “recommend”/”do not recommend” binary response.
• Pymetrics held the scores for 330 days.
• Researchers found bias against Blacks and Asians – “systematically disadvantaged Blacks and Asians”.

I felt like the paper’s conclusion was materially different than the Twitter video.

What do we do?

I don’t really have an answer for this. Obviously, you can’t do this kind of extensive research to dig at what is the “truth” out there with everything you read or hear. Reading that study was not a 5-minute endeavor; it was a couple of hours.

I only did that extensive research because I was interested enough to dig deeper. I wanted to have indepth details before I start spreading such findings.

I don’t know the young lady’s role in this. She might have heard from someone else about the study and was just reporting on what she heard. Or she might have intentionally omitted salient information or twisted some facts.

If something catches your eye and it bothers you enough, maybe do a little research first before you disperse what you learn.

It’s a rough world out there.