I was interviewed by an AI bot for a job

If you talk to 5 different people during one interview cycle you'll have 5 bias sources

If you talk to 100 instances of chatgpt during 100 separate interviews you'll have 1 single bias source

So you might think, but no. The LLM contains a large number of biases, coming from different training texts. Depending on how you structure the question, you can get biased statements.

For instance, if I discuss audio electronics with Google Gemini, depending on what kinds of questions I ask, I can get audiophile crackpot quackery out of it, or I can get solid electronic engineering statements.

The training data contains a vast number of narratives that are filled with different points of view. Generally speaking, you get the ones that resonate with your own narrative threaded through your prompts.

One way is if you ask loaded questions: questions which assume that some statements hold true, and are seeking clarification within that context. If the AI hasn't been system-prompted or fine tuned to push back on that topic, it may just take those assumptions at face value, and then produce token predictions out of narratives which express similar assumptions.

That's why a screening is needed. If people lie, it won't make me lower my standards.

I'm dealing with this all the time in recruitment. It can be done. People lie all the time or don't read the requirements. You need a way for the ones who really do know how to do the thing you need to demonstrate it to you.

Have you ever been in a situation where you had to hire someone to help you with something? Would you really follow your own advice? Your advice does not make sense for carpenters, cooks, or drivers. Why should it make sense for programmers?

I sent out screens to ~15-25% of resumes (a higher rate for new grads, lower for seniors, not wanting to unnecessarily rule them out just because they didn't have positive evidence of potential success and didn't know how to write a resume, only ruling them out of there was positive evidence they'd be unsuccessful). That amounted to ~100 per position filled. Around half of those completed the take-home. Some of the rest should have self-selected out and didn't, which is something I'd like to improve if I run a take-home again.

You're right the time commitment wasn't equal. Early on I spent much more time than the candidates designing and analyzing the test. Afterward, their 20 minutes would usually take me I did read every submission though. It wasn't wasted time for candidates.

Called legal aid. The bot that answered the phone submitted a complaint to the court and the management company which cited the correct historic documents and demanded compliance with them.

The management company bot responded to the court declaring that they're doing all they're required to do to correct the noise, and concluded with "the issue is not ripe for adjudication" -- whatever that means.

The court's bot agreed and binned the complaint "with prejudice" -- again, whatever that means, and sent me a fine for wasting their time.

Every day, the noise still happens.

And every day, the man from the management company still shows up to silence the noise.

I've come to know him fairly well.

It turns out that his name is William, although everyone calls him Bill. Bill is a nice guy who once studied computer programming, but the best-paying job he ever managed to get was slinging packages for Amazon back when that was still a thing that people did.

Most Thursday nights, if we don't have anything else going on, Bill and I go bowling at the AMF that's not too far down the road. It was his idea. We've been doing this about every week for long enough that I've learned to become a pretty proficient bowler. And while I still enjoy that part, we spend most of our time having a few beers and solving the world's problems.

A few months ago, we started talking about pinsetters and Bill mentioned that he read once that this was once a job that people did manually -- that rather than having a machine at the end of the alley, there were people behind the wall who would collect the scattered pins and put them back onto the painted dots on the floor. That sounded pretty archaic compared to the machines that I've seen doing this work for my entire life, but it seemed likely enough.

I started thinking about some other things about bowling: These days, we just walk in and our shoes are ready for us by the time we make it up to the front. We pick our own lane and just start bowling. After that, the machine sets the pins, keeps the score, and returns the ball. Pretty normal stuff.

And then, Bill pointed out the other people: There were a couple of small groups of people who were bowling, and one grizzled old fellah nursing what looked like a White Russian at the bar, but that was it. Nobody else was present; nobody actually worked there at all.

How long had it been since I asked for a pair of size 11 shoes, I wondered? When was the last time I talked to a bartender to order another beer? I hadn't paid for a thing using a card, or even carried anything like that with me for what seemed like eons. The self-cleaning bathrooms were certainly a welcome change, but how long ago were those put in and what happened to the person who used to clean them?

Neither of us could pick an exact timeframe for when these things changed. We both agreed that it wasn't important at the time, and that it seemed like a natural-enough progression.

Anyway, it was getting late again. After we put our shoes onto the mat for the sanitizer bot to deal with and started to walk out, the screens by the door told us what our tabs were, debited our accounts, and told us that it would see us next week.

I'm sure that Bill will stop by tomorrow afternoon to push the button and silence the noise from the electrical panel for another 24 hours, just like he always has.