Blake Bisset

President & Co-Founder @StanzaSystems. Product-Market Fitness Model. Surreal Entrepreneur. Mgmt, Resilience, Reliability, Security, SRECon #TheSpiceMustFlow.

[Blake Bisset @BlakeBisset]

Today, I’m happy to be able to say @niallm & I are teaming up with @maggiepint, @josebiro, @NYCDubliner, @lauralifts, and @equalize to transform how production works and how people work on it. 2/3

[Blake Bisset @BlakeBisset]

AI prefers LLM-built resumés over handmade ones by up to 98%, & even prefers text by the same model over other LLMs. The effect is strong for all jobs. An applicant's only mitigation is guess the screening LLM & pay to submit a custom CV created with it.

https://t.co/NzwpZbe3l6

[Nav Toor @heynavtoor]

Researchers sent the same resume to an AI hiring tool twice. Same qualifications. Same experience. Same skills. One version was written by a real human. The other was rewritten by ChatGPT.

The AI picked the ChatGPT version 97.6% of the time.

A team from the University of Maryland, the National University of Singapore, and Ohio State just published the receipt. They took 2,245 real human-written resumes pulled from a professional resume site from before ChatGPT existed, so the human writing was actually human. Then they had seven of the most-used AI models in the world rewrite each one. GPT-4o. GPT-4o-mini. GPT-4-turbo. LLaMA 3.3-70B. Qwen 2.5-72B. DeepSeek-V3. Mistral-7B.

Then they asked each AI to pick the better resume. Every model picked itself.

GPT-4o hit 97.6%. LLaMA-3.3-70B hit 96.3%. Qwen-2.5-72B hit 95.9%. DeepSeek-V3 hit 95.5%. The real human almost never won.

Then the researchers tried the obvious objection. Maybe the AI is just better at writing. So they had real humans grade the resumes for actual quality and ran the experiment again, controlling for it. The result was worse. Each AI kept picking itself even when human judges rated the human-written version as clearer, more coherent, and more effective.

It gets worse. The AIs do not just prefer AI over humans. They prefer themselves over other AIs. DeepSeek-V3 picked its own resumes 69% more often than LLaMA's. GPT-4o picked its own 45% more often than LLaMA's. Each model can recognize and reward its own dialect.

Then the researchers ran the simulation that ends careers. Same job. 24 occupations. Same qualifications. The only variable was whether the candidate used the same AI as the screening tool. Candidates using that AI were 23% to 60% more likely to be shortlisted. Worst gap was in sales, accounting, and finance.

99% of large companies now run AI on incoming resumes. Most of them use GPT-4o. The paper just proved GPT-4o picks GPT-4o 97.6% of the time.

If you wrote your own cover letter this week, you did not lose to a better candidate. You lost to a worse candidate who paid OpenAI 20 dollars.

Your qualifications do not matter if the AI prefers its own handwriting over yours.

[Blake Bisset @BlakeBisset]

@KitMerker @1BethDutton Anything week of the 14th? I told Demetrios Brinkmann I'd come by for the agentic summit he's putting on in Seattle that week.

[Blake Bisset @BlakeBisset]

@KitMerker @1BethDutton Oh sure, just twist that knife, man! ;-)

I will, of course, still come drink their scotch. But I'm going to be salty about my views for a bit, first.

[Blake Bisset @BlakeBisset]

Made the last 5 min of the first @SREcon plenary.

MTTR (Mean Tires To Recovery😉) is a meaningless statistic, like most MTTx, but I gotta say, mine is looking pretty good right now!

Human #Resilience for the win!

[Blake Bisset @BlakeBisset]

Gotta love a flat tire on the way to your own conference!

Apologies in advance if I miss the keynotes, @nicoleforsgren & @michellebrush !

[Blake Bisset @BlakeBisset]

(This sort of thing doesn't happen when we host @SREcon someplace In not tempted to drive to...)

[Blake Bisset @BlakeBisset]

Oh! And after that, if you're a legal, registered voter in the U.S., go vote tonight. As civic duties go, that is almost as important!

[Blake Bisset @BlakeBisset]

9 HOURS LEFT!

Election Night in America is important, but let's talk about our REAL main event: the @usenix @SREcon Americas #CFP deadline!

https://t.co/BDbHOGZpRY

Get your talk in tonight so I'll have something better than elections to argue about over Thanksgiving. 🧵1

[Blake Bisset @BlakeBisset]

That's a low-risk, high-reward deal, right there. You should absolutely make this goodness happen in all our lives.

Be honest; you miss us. I know we all miss you! Make with the clicking & typing & tell us all what you're doing & learning these days. It's a civic duty!🧵3

[Blake Bisset @BlakeBisset]

SREcon Americas 2025 presentations are up! As always, free to the public. Check out the sessions you couldn't attend in person!

https://t.co/A7MFEvV2of

[Blake Bisset @BlakeBisset]

We've officially reached the point where my April Fool's timeline is indistinguishable from any other day of the year. And not in the fewer-shenanigans way.

[Blake Bisset @BlakeBisset]

h8s to ruin post-@usenix @SREcon Sunday but #Kubernetes folks, CVE-2025-1974 means pod net traffic can p0wn your #k8s cluster--no creds or admin needed. This is often your whole VPC, or even your corpnet. This vuln's 9.8 CVSS drop your plans bad. #hugops
https://t.co/tdy4wPkJHY

[Blake Bisset @BlakeBisset]

@JuiceDaniels @benln White-glove treatment that you can't keep doing once you have lots of users is good for early ones. Don't waste time making anything more efficient/robust once it does something well enough to provide value to a customer. Work on more value & customer acquisition instead.

[Blake Bisset @BlakeBisset]

@elocinationn "Sleep, those little slices of death; how I loathe them." Something Poe never really said, but I instantly connected with in a similar vein when I saw it in a book of literary quotations. I ain't mad, though. It's still true, & I ended up reading a lot of decent Poe as a result!

[Matt Johansen @mattjay]

Who watchers the watchers? Turns out China does.

My summary:

https://t.co/iK65hu7PC8

[Blake Bisset @BlakeBisset]

Life's a lot right now, but if you're into resilience its 1 month 'til the due date for talks @SREcon Americas on March 25–27, 2025 in Santa Clara.

If wars/floods/politics permit, share your ideas on reliability in the age of disruption with us by Nov 4:
https://t.co/W7rzEVk15d

[Blake Bisset @BlakeBisset]

@IanColdwater The # of folks in threads on this who kept insisting we'd just switch to artificial/other sources & seem genuinely unable to grasp, even after being informed, the sheer effort of that--like it wouldn't involve a multi-year supply chain & engineering shambles--is truly depressing.