Ahmed Hasan

The extent to which AI FOMO is warping people's decisions is rather stunning, actually. I can see why the Anthropics and Googles would want this, but the other tech co managers going along with it ...

Experimental evidence of the effects of large language models versus web search on depth of learning Abstract. The effects of using large language models (LLMs) versus traditional web search on depth of learning are explored. A theory is proposed that when

When people learn with ChatGPT instead of following their own searches, they end up knowing less, caring less, and producing worse advice, even when the facts are the same.

Friction is an essential ingredient for learning! Convenience makes us shallow.

academic.oup.com/pnasnexus/ar...

If your random seed is 42 I will come to your office and set your computer on fire🔥 Figuratively. More likely you'll get a stern talking to.

We need to have a conversation about random seeds. Don't use 42.
blog.genesmindsmachines.com/p/if-your-ra...

Screenshot of text reading: Significance Large Language Models (LLMs) are used in evaluative tasks across domains. Yet, what appears as alignment with human or expert judgments may conceal a deeper shift in how “judgment” itself is operationalized. Using news outlets as a controlled benchmark, we compare six LLMs to expert ratings and human evaluations under an identical, structured framework. While models often match expert outputs, our results suggest that they may rely on lexical associations and statistical priors rather than contextual reasoning or normative criteria. We term this divergence epistemia: the illusion of knowledge emerging when surface plausibility replaces verification. Our findings suggest not only performance asymmetries but also a shift in the heuristics underlying evaluative processes, raising fundamental questions about delegating judgment to LLMs.

We. Told. You. So. But also, that "May Rely" in the significance section ( 👇🏾 ) is *way* too… let's go with "generous;" lexical associations and statistical priors are what these systems are literally made of, and all that they do. Literally. doi.org/10.1073/pnas...

I am convinced that Microsoft is run by people who do not interact with digital devices. Somebody holds up a computer and uses it for Satya Nadella, he just sits and scratches his chin and says “make it do more”

tweet that reads: Lady behind me at The Thing screening last night to her partner "why are they shooting at the dog?" and her partner says very firmly "have you never seen a film before? You watch it and information is revealed"

i think about this post every time i watch a movie

CHOTINER: This game you like to play, what's it called?

ME: Slay the Spire.

CHOTINER: And you play it on the computer?

ME: Yeah. Well, I also have it on my phone. I played another 300 hours.

CHOTINER: And these other games in your Steam library. Have you finished them?

ME: Now hold on a second,

El volquete del metal #71: Deafheaven tomaron una mesa Los discos de metal que tienes que escuchar

Ahmed just stumbled on a Substack article written in spanish with a positive review of Nighttime Heists/Daytime Getaways written by a total stranger this is such an incredible feeling :' ) I love making music

www.hipersonica.com/p/el-volquet...

Balancing off-target and on-target considerations for optimized Cas9 CRISPR knockout library design The continued development of high-dimensional CRISPR screen readouts, such as single-cell RNA sequencing and high-content imaging, necessitates compact libraries to enable functional interrogation at ...

Doomscrolling pause to chat CRISPR libraries! Preprint describes our new, data-driven approach to combine on-target and off-target predictions much more intelligently for *selecting* guides, which we use to develop our newest Cas9 knockout library, Jacquere. Thread: www.biorxiv.org/content/10.1...

One of the ways that LLM-authored code improves productivity is by merely SAYING it does things. It's way faster than the whole time-consuming process of actually doing things. This is real code someone sent to me for review. A screenshot of a code editor. The line numbers are in the lefthand column, running from 121 to 125. There is a comment that says "Verify packages were created and check sizes" Then line 123 says echo Verifying Lambda packages were created... and Line 124 says echo checkmark emoji colon All Lambda packages verified successfully. There is no line in here that actually did anything.

UNLIMITED VELOCITY

Academic authors, here's a peek into the black box of journal publishing from an journal editor if you can bear it:

Excellent. "If you’re pushing slop or eating it, you wouldn’t read it anyway. You’d ask a bot for a summary and forget what it told you, then proceed with your day, unchanged by words you did not read and ideas you did not consider."

Everyone else, please read.

🫡 aye aye - hopefully will help with post-lunch debugging woes

Ahh I remember giving it a listen when it first came out, but it didn’t stick with me at the time. I’m due for a revisit though! Greg was on guitar for that too iirc which is neat

lol how did I forget Black Bubblegum. One of the catchiest DEP numbers ever

Yeah Greg-era TDEP didn’t explore that side of their sound too much overall unfortunately. They were real good at it too! There are some similar tunes like Unretrofied and Dead as History but most of the discog is abrasive mathcore

I only discovered Elliott a few years back and I’m grateful I did!

I have some memory of you posting about Elliott Smith some time back, and so will forgive these unkind words about an album I love 😭

Hey now, the ending of Hero of the Soviet Union is a great description of certain unscrupulous Silicon Valley characters

Big tech’s selective disclosure masks AI’s real climate impact Google claims to have disclosed new information proving its own efficiency. But it has hidden the bigger picture. Guess what: I’ve got the bigger picture for you right here in this big old po…

HEY friends: recently seen heaps of "disclosures" from AI companies of the energy cost per question you type into chatbots (most recently from Google)?

Presenting per-query 'efficiency' is part of big tech's efforts to hide its extremely real and happening-right-now climate impacts!

NEW POST -->>>

Scientists no Longer Find Twitter Professionally Useful, and have Switched to Bluesky Synopsis. Social media has become widely used by the scientific community for a variety of professional uses, including networking and public outreach. For

Hey everybody! @drjuliawester.bsky.social and I have a new paper!

We surveyed over 800 scientists, science communicators, and science educators who use social media.

Conclusion: Scientists no longer find Twitter useful or pleasant, and many have switched to Bluesky! 🧪🌎🦑

doi.org/10.1093/icb/...

Air Canada consoles stranded passengers with pictures of their executive bonuses OTTAWA - With thousands of passengers still facing travel delays as flight attendants strike for fair working conditions, Air Canada executives have decided to comfort customers by sending photos of t...

Air Canada consoles stranded passengers with pictures of their executive bonuses

The LLMentalist Effect: how chat-based Large Language Models rep… The new era of tech seems to be built on superstitious behaviour

I will never pass up a chance to re-share this piece. Best way I ever found to talk about AI to anyone: softwarecrisis.dev/letters/llme...

PhD-level intelligence or the graduate student from hell AI companies love to tout that their models are approaching—or have reached—PhD-level intelligence. This is blatant nonsensical marketing geared towards an audience that deeply misunderstands what a P...

clauswilke.substack.com/p/phd-level-...

Excited to share my newest quarto revealjs plugin: imagemover

Easily reposition and resize images directly in your quarto revealjs slides for a much smoother slidecrafting experience

github.com/EmilHvitfeld...
#quarto

An abbreviation (ABB) in a journal article (JA) or Grant Application (GA) is rarely worth the words it saves. Every ABB requires cognitive resources (CR) and at my age by the time I'm halfway through a JA or GA I no longer have the CR to remember what your ABB stood for.

AI2027 is a kind of intelligence test in that if somebody (like Peggy Noonan did in the WSJ today!) cites it as "proof" of anything you can immediately dismiss them as a rube. It's totally nonsensical.
houseofmirrors.substack.com/p/the-spectr...

The Prodigal Techbro Prodigal tech bro stories skip straight from the past, when they were part of something that—surprise!—turned out to be bad, to the present, where they are now a moral authority on how to do good…

remember there’s always a lucrative position for the seeming critic of the tech industry who doesn’t actually challenge its power and foundational ideas.

the prodigal tech bro is one example. there’s a whole industry of academics and journalists who serve in this role too.

THAT WAS ONE OF THE BEST GOD DAMN SHOWS I'VE EVER SEEN

There is some justice in the world. The new Thank You Scientist vocalist kicks so much fucking ass!!!!

Excited to finally have the pre-print up for this work on evolutionary rescue from my PhD at UBC!

Previous theoretical work on the effects of negative density-dependent growth on evolutionary rescue has uncovered inconsistent results, leading to an incomplete understanding of its influence.