It's exhausting when colleagues (who claim to know about design research) send you articles about using AI in research and it's all academic meta-study theme based tasks.
You're meant to know enough to know that's not what we do. You're not meant to be adding to my workload by refuting the obvious
I really didn't find The Drama to be a good movie, which seems to contrast with everyone online.
I'm resolving this dissonance by assuming everyone positive about it online is just a bot run by a marketing agency
It took me longer than I'd like to admit to realise which prime minister it was referring to
Fun workshop warm up thought exercise at work:
Would it be 1 Machamp for one courageous person, so all 4 fists? 1 Machamp and 4 people? Maybe 2 people per Machamp?
I’m sure they could do something thematically with machamp
It isn’t Pokémon role play?
An ad for bbwjobs.com but the letter spacing is extremely tight, so the second b look like an l and o instead.
keming. not even once.
That makes so much sense as a flow! And that’s such a great use case! A reference / template that can just be used as a go to! You made me realise that I’ve been doing roughly the same thing as my breakfast and lunch don’t tend to vary!
What was the UI jank? MF has some odd UX choices
A version of the Jennifer Coolidge 4th of July meme where she states that the sunny weather makes her want to automate her workflows with AI real bad instead
I gift you all this
Put my vote into the compartmentalisation box. I'd go far as to say national politics tends to be ignored if it doesn't align with the current business strategy
You know what, this sunny weather really makes me want to pivot to AI and automate my workflow processes
Gah! Probably on LinkedIn. I’m starting the rumour that Apple’s autocorrect was vibe coded with Copilot
Code is law or something like that. I’m sure someone has come up an expression, proton linkedin
Second cycle before work! Go me!
That’s an awesome chart!
Am I allowed to say that I want to know their secret so I can become one?
mind was inside on a Peloton. I'm not allowed outside (and who is insane enough to actually wake up early enough to cycle and then get to work before 9?)
Just one? That's not going to see you through to lunch
UXRs and Design Researchers, what tool would you pick between UserTesting, LookBack, and BallPark HQ?
Already did a quick cycle before work and tracked my calories!
I've really got to stop passively checking LinkedIn. There's some good content but we're talking about a 1:100 ratio with the amount of slop and just generally bad PR-style posts
2 weeks of absolutely no motivation to eat well and do any form of workout. This is me stating publicly that I'm back on it from today
Does anyone subscribe to Lenny's substack?
I'm curious what is after the subscription prompt on this post but his content isn't often relevant enough to me to actually pay
www.lennysnewsletter.com/p/how-to-do-...
Screenshot of an article from The Making of Making Sense by Dorian Taylor: First we should reiterate why LLMs work for writing software at all: Code can be verified cheaply, and at several levels, by other code (linter, compiler, unit tests). Mathematically proving it is another matter, but formal proofs aren’t actually necessary to be fit for purpose. There is an enormous body of source code to train on, with attendant documentation, going back decades. Most software just repeats similar patterns—they only vary slightly in the details—a phenomenon well-represented in the training data. Code does not have to reference empirical reality, it only has to have the right “shape”.
Oooo! bullet points reiterating the above!
Screenshot from a blog on The Sense of Sense Making by Dorian Taylor: A question worth asking is why do large language models perform relatively well with code, anyway? I agree with other commentators that it’s because there’s a right answer that is verifiable—quickly and deterministically—by external means. I’ll also add that code only has to be internally consistent: it (mostly) doesn’t have to reference facts about the world.
High-level summary of why LLMs work for code. It'll be useful to contrast with AI Generated insights that aren't quick to verify. If we already know enough to know an LLM is talking bollocks, then why would we be doing the research in the first place?
Source: buttondown.com/dorian/archi...
What did they do? I open blue sky, see this, and I assume it’s a trans thing?
I’m curious why the saucier ones are down the bottom, at infant height level, and the more bland ones are at adult level
Screenshot of text from a New York Times article by Ezra Klein published in March 2026: Researchers have drawn a distinction between “cognitive offloading” and “cognitive surrender.” Cognitive offloading comes when you shift a discrete task over to a tool like a calculator; cognitive surrender comes when, as Steven Shaw and Gideon Nave of the University of Pennsylvania put it, “the user relinquishes cognitive control and adopts the A.I.’s judgment as their own.” In practice, I wonder whether this distinction is so clean: My use of calculators has surely atrophied my math skills, as my use of mapping services has allowed my (already poor) sense of direction to diminish further.
I need to remember the term "Cognitive Surrender". I'm been struggling to express the difference between using Excel to crunch some numbers and just asking AI to crunch everything and produce a final report with a top line, and why that difference is important
Source: www.nytimes.com/2026/03/29/o...
That's a good way of remembering it!