David Asboth

Intro time! I'm David. I call myself a "data generalist" because I haven't found a label that fits. I've been a software dev, data scientist, and my real passion is education so now I do Python and data training. I'm currently most interested in skills that data people *actually* need to know.

I think you can have one change that worked if the issue is: I made a tweak to my cinnamon bun recipe that stopped them tasting like ass. I don't think you can if it's like I was lost in grief & then I took up Nordic walking

I've been away a couple of weeks and I come back to Firefox replacing every search bar and UI element with rounded versions. Stop it, I don't like change anymore.

Claude just said something is load-bearing, despite my custom instructions being very clear about that phrase but at least it had the sense to add: "sorry — but literally, here, it is: the whole plan rests on it". 2026, man, what a time to be alive.

The Guardian has printed articles week after week foregrounding the narrative that 'councils are at breaking point' because of the 'burden' of SEN kids having the gall to ask for support that allows them to access the same education as their peers. The focus always on cost, not human dignity.

If you are convinced that AI is, despite its flaws & ethical failings, a reliable research tool, ask it a specific question in a domain you know very well. Most people - understandably - ask LLMs questions about areas they're not experts in. They can't judge if the information is accurate.

Half Stack Data Science by @davidasboth.com and @shaunmcgirr.bsky.social might be my new favourite podcast.

New episode! How will we learn things in an age of AI? Recorded it a while ago, but luckily the topics haven't been made obsolete by recent developments in AI (always a risk). open.spotify.com/episode/3Ch2... #databs #datasky

I think as a society we went wrong when we started using the word "curious" to mean "weird" rather than "inquisitive".

BRB, grieving something that I didn't know didn't exist.

Excellent.

What I am suggesting is that, in the same way that people turning to AI for therapy suggests a failing of both public mental health institutions & our own capacity as a society to provide individual & community support, that so many would turn to AI suggests a systemic failure to value the arts.

Book: Test-Driven Data Analysis (by @njr0.bsky.social) Chapter 1 Data analysis as if the answers actually matter ht @cghlewis.bsky.social cc @rmcelreath.bsky.social book.tdda.info/book/chapter...

i think some pro-AI people are also taken aback by what feels like the sudden implication that they’re not cool when they’ve been told the last couple of years how they’re on the cutting edge and anyone who didn’t get on board would end up looking like a loser when maybe it’s the other way around

My book on TDDA is out. TDDA is like test-driven development (TDD) for data science with reproducible research. The tdda library has cross-language command-line tools and Python APIs. 20% off at www.routledge.com/Test-Driven-... with code 26SMA1. book.tdda.info #data #testing #reproducilibility

terrible. 20 years ago, the internet was full of hobbyist blogs and forums. over time, these have been replaced by paywalled substacks and private discord channels, each a walled garden. google's decision to prioritize AI just makes it less likely people will create free, public info

R, Python, Stata, SPSS, SAS, Matlab

What happens when everyone can build software by chatting to an LLM? How does that change the dynamic of internal tools? Who is the Magical Glue Person and why might they emerge victorious? These and other questions answered in our latest episode: halfstackdatascience.com/s5e2-the-mag... #databs

Film you've watched more than six times with gif hard mode: no star wars, star trek, lotr, dc, marvel, disney, pixar, dreamworks, spielberg, cameron, wachowskis, or ghibli

Now that LLMs can spit out practice questions and become tireless tutors, where does that leave humans (e.g. me) teaching things like Intro to Python? My hope is that as educators, our focus can move further towards helping people *apply* their knowledge to real project work. Thoughts? #databs

I need to get back to this too actually. What's more human than making your ridiculous ideas, the ones that make you *you*, a reality?

I know AI has people divided and there are a lot of nuanced discussions to be had, but can we at least all rally around the idea that "AI dungeon master for D&D" is an absolutely awful, soul-destroying idea that can get straight in the bin? No, I will not be linking to it but it exists 😬 #databs

What's your favourite origin/root of a word? (in any language) I love etymology so much. It's like history & conspiracy theories combined. Our language has secret messages & lizard DNA baked into it. Like how 'gargoyle' comes from gargle / gurgle, because of the water spewing out their mouths

If you could experience any video game again for the first time, what would it be? I can't name just one but whatever, it's my feed, so I'll mention a few: - Grim Fandango - Fallout 4 - the Mass Effect trilogy

Milla Jovovich (yes, *that* Milla Jovovich) releasing an AI memory product that beats benchmarks is further proof we're in the weirdest timeline. #databs #datasky github.com/milla-jovovi...

If you use GitHub (especially if you pay for it!!) consider doing this *immediately* Settings -> Privacy -> Disallow GitHub to train their models on your code. GitHub opted *everyone* into training. No matter if you pay for the service (like I do). WTH github.com/settings/cop...

I love the framing of AI users as nannies of the LLMs. "Misbehavior is blamed on the nanny, or the parents, never the little boy." It's so true that especially in AI hype circles every mistake the thing makes is a "skill issue". #databs #datasky

I keep a Trello board of random project ideas I want to do at some point. About two months ago I created one titled: Raspberry Pi "idea generator" box Please help - what on earth did I mean by that? 🤣

Saw a data analyst ad that had "Familiarity with ChatGPT or other AI tools for data analysis" as a nice to have 🤪 I guess we're at the point where employers think ChatGPT is a tool for data analysis. Cool cool cool cool cool. #databs #datasky

Here’s the thing about folks speedrunning the creative process with AI: I could go to the beach every morning and put a grain of sand in a cup. They’d take their cup to the beach and fill it one day; but the point isn’t to have a cup of sand. It’s to fill my life with a thousand trips to the beach.

👋This is also how I feel! Hello friends, please enjoy my dumb little thoughts ❤️

Trying to break my addiction of checking football news multiple times a day even though I don't really watch football anymore. But I will briefly break my attempted football hiatus to say: it would be SO funny if Arsenal bottled it and didn't win the league. That is all.

One of my biggest fears as a data educator is running my "Intro to APIs" material in preparation for a class only to find that the API has been deprecated since the last time I used it. Didn't happen today but it's happened enough times! #databs #datasky

One dark UX pattern that really irritates me is when the logout/sign out button is hidden deep in Settings. It's not a setting, it's a primary action! Insert "old man yells at cloud" meme.

Researchers face this dilemma all the time. You can talk to the media with care & nuance, but you will always be outflanked by anyone willing to simplify, exaggerate, speak with undue certainty, answer questions outside their expertise & extrapolate in ways unsupported by the data.

Are you using open models on a regular basis? If yes, I’d love to hear about it! - what’s your use-case? - local or through an inference provider? - how has the experience evolved over time? - anything you miss from the closed models, if you’ve used them? Would appreciate reposts for visibility 🙏

Louder for those at the back: once you start to think about LLM outputs as optimising for *plausibility*, it's easier to put into context what LLMs can/can't do and how to use them properly. #databs #datasky

A succinct and useful list of rules for teaching data science, including some of my favourites: "teach with real data" and "have students do projects". #databs #datasky

BAHAHAHAHAHAHAHAHAHAHAHAJAHAHAHAHAHAHAHAHAHAHAHAHAHAHA *breathes* AHAHAHAHAHAHAHAHAHAHAHAHAHAHAAHAHAHA

We recorded an end of year review episode. We talked about what happened on the podcast, the world of AI, and what great sci-fi we read in 2025. Listen wherever you listen! Looking forward to recording more in 2026! #databs #datasky