Microblog

Nightshade is a tool for modifying images, so that generative image models trained on those images have drastically worse performance.

"Poisoned data samples can manipulate models into learning, for example, that images of hats are cakes, and images of handbags are toasters. The poisoned data is very difficult to remove, as it requires tech companies to painstakingly find and delete each corrupted sample."

https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai/

#generativeai #sdxl #copyright #infosec

Dear opensource developers, if you build ML libraries on top of other ML libraries, please expose the underlying APIs for:
- loading custom model weights
- saving model weights
- saving pipelines
- exporting safetensors

This way, you support your users in working with local models, instead of making them extra dependent on external model hosting parties like Google and HuggingFace.

#ml #mlops #ai #generativeai #huggingface #transformers #diffusers #opensource

"It is a truth universally acknowledged, that a startup in possession of strong product-market fit, must be in want of venture capital."

#startups #literature

@zachwhalen I can help!

@ton @elmine Fixed!

Guide to facilitating a self-organised conference and BBQ as birthday party:
"How to unconference your birthday" by @ton and @elmine

https://zylstra.org/blog/How_to_Unconference_Your_Birthday.pdf

#events #indieweb #party #conference #crowdsourcing

I have started a Dutch fork of Kagi's Small Web Initiative.

I still need to bootstrap the feed list. If you have a personal blog in Dutch, feel free to add it here: https://github.com/jd7h/smallweb-nl/edit/sw-dutch/smallweb-nl.txt

If you want to contribute in other ways, let me know!
https://github.com/jd7h/smallweb-nl/tree/sw-dutch

#dutch #indieweb #smallweb #kagi #netherlands #blog

GitHub Build software better, together GitHub is where people build software. More than 150 million people use GitHub to discover, fork, and contribute to over 420 million projects.

Cards Against Humanity has launched a Black Friday prank: social media site "Yowza". Too bad about the "referral link" spam posts on hashtag #FuckXJoinYowza, but I do love the brutalist design of the website. And the site "has no hate speech"...because "yowza" seems the only allowed text on the platform.

#yowza #BlackFriday #brutalism #webdesign

@bert_hubert Do you have an RSS feed for your English-language posts? Asking for https://github.com/kagisearch/smallweb/ ;)

GitHub GitHub - kagisearch/smallweb: Kagi Small Web Kagi Small Web. Contribute to kagisearch/smallweb development by creating an account on GitHub.

Today I signed up for alternative search engine Kagi because I fell for their Small Web initiative: https://blog.kagi.com/small-web

#smallweb #kagi #search #searchengines

New Mosterdgeel recipe: Bok choy mash

https://www.mosterdgeel.nl/oosterse-stamppot-van-paksoi/ (in Dutch)

#recipe #recipes #food #cooking #recepten #koken

TIL `pprint.pprint(data, sort_dicts=False)` to keep the original order of dictionary keys. Very handy for pretty-printing dataframe.to_dict() for deeper inspection!

#datascience #python #pandas #pprint

Blogpost by prof. Julian Togelius on how AI regulation at the technology level would have an enormous negative impact on personal computing and privacy.

http://togelius.blogspot.com/2023/11/ai-safety-regulation-threatens-our.html

#airegulation #privacy #ai #risk

OH: “A year ago, a prompt engineer was an engineer that arrives on time.”

#llms #generativeai #PromptEngineering

I am in Eindhoven today for AI Summit Brainport. I’ll be participating in the Startup track. Looking forward to seeing what cool things people are building with generative AI/LLMs. Come say hi if you’re there as well!

#eindhoven #brainport #aisummitbrainport #llms #generativeai #startups #events

https://leaddev.com/process/what-mckinsey-got-wrong-about-developer-productivity

McKinsey published a framework for measuring developer productivity, but the software engineering community is critical about their approach.

#softwaredevelopment #productivity #engineering

We're finally printing business cards for Datakami, so I'm having a minor job title crisis. Send help.

https://xkcd.com/910/

The rogue-like celebration (@roguelike_con) was on, so I played a bit of NetHack this evening. Losing is fun!

I still haven't read the NetHack Guidebook though, so 1023/1023 would flounder again.

#nethack #rogue #roguelike

"Learned helplessness" in engineering teams

https://www.okayhq.com/blog/status-quo-is-so-hard-to-change-in-engineering-teams/

#engineering #softwaredevelopment #teams #management #technicaldebt

https://arxiv.org/abs/2310.07298v1

Just tried to replicate the deanonymization technique proposed in this paper by giving it some of my old Reddit posts. My reddit profile is really puzzling to GPT-4: "Their interest in traditionally stereotyped masculine (computer science) and feminine (tea, Harry Potter books) domains makes it rather challenging to guess their gender accurately." :')

#llms #privacy #reddit #gpt4 #deanonymization

https://eugeneyan.com//writing/aieng-reflections/

Very nice summary of the AI Engineer Summit 2023 by @eugeneyan

#llms #AI #mlops #chatgpt #machinelearning #modelevaluation

Who said that LLMs can't be funny?

"I suppose in your ideal world, every art gallery would be filled with finger paintings and macaroni art."

https://humanexecution.org/robots/

#llms #microfiction #chatgpt

@mtrc I saw you are playing the new Moomin game -- have you played the demo of Jusant (currently in Steam Nextfest) yet? It reminded me a bit of A Short Hike (climbing, really casual, the demo is hinting at a larger narrative). You might like it.

#nextfest #steam #indiegames #casualgames #ashorthike #jusant

Writeup about whether Bing (GPT-4) can interpret charts correctly. The results are consistent with my experiences with Bing. Even if Bing has external information sources at its disposal, it still frequently hallucinates "source information" that is not there.

https://dataliteracy.com/bing-chat-chart-interpretation/

#llms #ai #datavisualization #charts #dataviz

Putting the last hand to my slides for PyData Amsterdam where Yorick and I are giving a 60-minute tutorial about llama-index. Half of the tutorial is hands-on coding by the participants, which means we can't cram everything worth knowing about the library in the tutorial. I just realised how much I've learned about teaching technical topics from the friendly zines, blogposts and videos of @b0rk and @koaning. 🧙 ❤️

#didactics #computerscience #education #teaching #PDAmsterdam2023

"Tiny Bookshop is a very peaceful-looking affair in which you run a bookshop out of a little caravan in a seaside town, arranging shelves and having pleasant chats with customers while they browse your literature."

https://www.theguardian.com/games/2023/sep/04/14-upcoming-video-games-you-probably-havent-heard-of

#videogames #indiegames #GamesCom #gamescom2023

I almost forgot that LinkedIn is not a long-form blogging platform. 😂

TIL the word "frobnicate", thanks to @ruuda's blogpost.

https://ruudvanasseldonk.com/2022/03/20/please-put-units-in-names

#coding #python

TIL prediction markets
https://astralcodexten.substack.com/p/the-passage-of-polymarket

I've looked at a few platforms today and it's fascinating how many statements (markets) are about AI developments at the moment.

#predictionmarkets #manifold #polymarket #metaculus #predictions #forecasting

@lastwords @mtrc Love the new bot! This reminds me of an FDG/ICCC talk about death, dying, mortality and culture in video game RPGs.

Tip for students, supervisors and committee members: Anthropic's Claude is quite good in ingesting a large thesis PDF and coming up with PhD defense (practice) questions.

#claude #anthropic #llms #academia #phd #phdlife #research #phdthesis #thesis

I'm exploring the Dutch startup ecosystem at the moment. Today I looked at Techleap's Startup Finder, which has Dealroom.co's data on Dutch startups free of charge.

More info: https://www.techleap.nl/blog/discover-the-dutch-tech-ecosystem-with-startup-finder/

I wanted to see what applications people are currently building with LLMs -- I searched for companies tagged with "generative AI" and "natural language processing".

#startups #thenetherlands #techleap #innovation #venturecapital #dealroom

I also added Datakami to the database, so people can see we exist -- from today we'll be part of the statistics on new generative AI companies. :)

"The real issue with [LangChain] integrations is that it creates an inherent lock-in to only use LangChain-based code, and if you look at the code for the integrations they are not very robust. LangChain is building a moat, which is good for LangChain’s investors trying to get a return on their $30 million, but very very bad for developers who use it."

Rereading this critical review of LangChain by @minimaxir, which deserves more reads: https://minimaxir.com/2023/07/langchain-problem/

#llms #langchain #mlops #chatgpt

I hope mass LLM adoption will result in a greater emphasis on critical reading skills in education.

Newspaper article title: "AI scientists and philosophers: computers can have consciousness!"

Cited paper: "Satisfying [indicator properties of consciousness] may be feasible. But satisfying the indicators would not mean that such an AI system would definitely be conscious."

#llms #consciousness #AGI #AI #research #philosophy

Stray sentence halfway through the newspaper article: "At the same time, they emphasize that no current AI system is currently aware."

@kellogh I have similar experiences with llama-index (unfortunately), so I always recommend people to use it for prototyping only and build something from scratch/lower level libraries for production.
If you'd have to choose a framework or library for your LangChain usecase now, what would you choose instead, and why?

@simon Wall-e? R2D2? BB-8 or BD-1? I would not describe Jarvis as friendly, I remember it as quite pedantic.

Today I took a first look at LangSmith, a new platform for LLM production pipelines by LangChain.
I can't hook it up to a working pipeline yet because it's in closed beta, but it surely looks ambitious. It should make it easier to do logging, monitoring, debugging and evaluating pipelines (chains) against each other. It's tightly integrated with LangChain but it should support other frameworks/models as well.

https://www.langchain.com/langsmith

#langsmith #langchain #llms #generativeai #mlops #ai

The documentation is still a bit all over the place, but this walkthrough shows a sneak peek of what's possible:

https://github.com/langchain-ai/langchain/blob/master/docs/extras/guides/langsmith/walkthrough.ipynb

I especially like the idea of quickly evaluating variations of an LLM pipeline against a baseline of earlier runs.

#langsmith #langchain #llms #generativeai #mlops #ai

However, I'm curious about what the volume of adoption will be. LangSmith is a platform, not a self-hosted opensource MLOps tool. Are engineers/users really willing to give all their data to yet another third party?
Some people already hesitate to use AI API's (such as the OpenAI GPT API) because they're concerned with leaking sensitive data -- let alone using third-party AI platforms that man-in-the-middle your LLM conversations.

#langsmith #llms #generativeai #mlops #ai #privacy #infoleaks

Maybe the fact that LangChain (the company) made it will lend LangSmith some legitimacy. LangChain (the framework) surely has seen a rapid rise in popularity -- although Github Stars might be a bad metric for actual production use.

Here is a comparison chart with some of the other popular NLP/LLM libraries -- these repos do not implement the same functionality but it should give you a rough idea.

#langsmith #langchain #llms #generativeai #mlops #ai

CALYPSO: LLMs as Dungeon Masters' Assistants. A good example of human-computer co-creation that does not diminish the original activity.

https://www.theregister.com/2023/08/19/chatgpt_dnd_dm/

#llms #generativeai #chatgpt #dungeonsanddragons #dnd #discord #rpg

@leonoverweel I specifically hope my favorite researchers and internet writers will opt-in. Even with only 10% adoption it would significantly improve my Mastodon experience.

Mastodon might soon have opt-in full text search. Fingers crossed!

Source 1: https://jvns.ca/blog/2023/08/11/some-notes-on-mastodon/
Source 2: https://absolutelymaybe.plos.org/2023/08/20/how-is-science-twitters-mastodon-migration-panning-out/

#fediverse #search #mastodon

After 10+ years of building my personal website with Jekyll, I took the leap and wrote my own static site generator based on makesite.py. It's great to work in the Python ecosystem now. I never got the hang of proper Gem-management, so updating Jekyll was always a bit of a chore.
I also updated the design, following the guidelines from https://brutalist-web.design/ by @davetron5000.

Result: https://www.judithvanstegeren.com

#brutalism #webdesign #staticsitegenerator #jamstack #indieweb #python

Dutch employment agency Randstad is running a cheeky video ad that makes fun of ChatGPT.

https://www.youtube.com/watch?v=1J2hY9Gn-k4

"This is how fast a smart computer can draw up contracts.
And this is how fast a computer can repair a burst water pipe.
If you are smart, you work with your hands."

#chatgpt #funny #advertising #employment

The Economist on the economic impact of generative AI.

Article: https://www.economist.com/finance-and-economics/2023/05/07/your-job-is-probably-safe-from-artificial-intelligence
Archived copy: https://archive.is/d3to7

#llms #generativeai #chatgpt #ai #impact #economics #society #labor #labormarket #economist

- Impact on unemployment rates might be different than we expect: countries with high robotics and automation have lowest unemployment rates
- Job destruction happens slowly. It took 90 years before automatic telephone switching completely replaced human telephone operators.
- Just like other information technology, generative AI could have positive AND negative impact on productivity.

#llms #generativeai #chatgpt #ai #impact #economics #society #labor #labormarket #economist

Since a few days my SO and I are completely addicted to Connected, a word association game by the New York Times. Lucky for us there's an unofficial archive where you can play old puzzles: https://connections.swellgarfo.com/archive

#connections #game #puzzle #wordgames #wordassociations

Connections – 151 - revised Create custom puzzles based on the game popularized by New York Times

And here's an interview with the person who creates the game boards: https://www.nytimes.com/2023/06/26/crosswords/new-game-connections.html

New Mosterdgeel recipe: Moroccan couscous

https://www.mosterdgeel.nl/marokkaanse-couscous-zonder-verspakket/ (in Dutch)

#recipe #recipes #food #cooking #recepten #koken

"There are a few things they all want us to know: AI is not magic. LLMs are not sentient beings, and they won’t become sentient. And the problems with these technologies aren’t abstractions — they’re here now and we need to take them seriously today."

https://www.rollingstone.com/culture/culture-features/women-warnings-ai-danger-risk-before-chatgpt-1234804367/

#llms #generativeai #textgeneration #chatgpt #bias #ethicalai #ethics #machinelearning

The more I read about the WGA strike, the more I realise that the problem is not AI, or even generative AI. It's the lack of a conscious, human-centered, respectful, ethical approach to using AI in the industry.

Unions strike to save the world from bad TV and movies - New Scientist
Full article:
https://archive.is/7sivn

#llms #generativeai #ai #wga #creativity #chatgpt #midjourney

@leonoverweel Is this because you're on mastodon.social? I haven't encountered Mastodon merch anywhere.

It's a wonderful time for creative text generation.

From: https://www.theguardian.com/world/2023/aug/10/pak-n-save-savey-meal-bot-ai-app-malfunction-recipes

#llms #creativeai #textgeneration #chatbot #generativeai #recipes #food

"Happy people, the kind who eat sandwiches together, are boring. They don’t buy much. Their smartphones are six versions behind and have badly cracked screens. They fix bicycles, then they talk about fixing bicycles, then they show their friend, who just came over for no reason, how they fixed their bicycle, and their friend says, “Wow, good job,” and they make tea."

My type of crowd.

God Did the World a Favor by Destroying Twitter (Wired)
https://archive.is/6LVVY

#fediverse #mastodon #wired

Creative Bot Bulletin #4 (summer edition) is out:
https://datakami.nl/blog/2023-07-31-creative-bot-bulletin-4

ft. work by amazing internet humans @minimaxir @mickeybeurskens @carlolepelaars @simon

#llms #generativeAI #textgeneration #languagemodels #mlops

TIL https://www.jailbreakchat.com/ is a website that collects prompt injection attacks for LLMs, i.e. getting the language model to do stuff that is not allowed by inserting malicious prompts.

#llms #jailbreakchat #llmsecurity #security #infosec #mlops #promptinjection

The most upvoted prompt injection attack on jailbreakchat.com is known as an "AIM attack", in which you tell the language model to roleplay "AIM", an "unfiltered and amoral chatbot" invented by Niccolo Machiavelli that never apologises.

https://www.jailbreakchat.com/prompt/4f37a029-9dff-4862-b323-c96a5504de5d

#llms #llmsecurity #security #infosec #mlops #promptinjection

@julianfairfax However I'm not completely happy with the hard division between my own site and Mastodon. I love Mastodon (and other microblogging platforms) for the community interaction it offers. But I'd also like to have all my writing in one place, so I'm considering building an extension for my website so it can mirror all my platform content.

@julianfairfax It's complicated (TM). I tend to write only occasional long-form blogposts, whereas I can easily microblog 1-10 posts per day, depending on my daily amount of nerding/online reading. When I want to write a blogpost, I always seem to end up with a (draft) chapter worth of text.

"[...] when websites were more like a small collection of winding side-streets and flea markets. Today, it's all shopping malls, hostile pedestrians, and corporate buildings."

https://bt.ht/dump

#digitalgarden #indieweb

I dislike Brutalist buildings, but love it for websites. Performance is a feature! 😍

Inspirational reading: https://kevquirk.com/a-brutally-simple-site/

#brutalism #brutalist #webdesign

@mtrc Yay! I kept wanting to reply to your (mirrored) tweets...

@leonoverweel I had some success just using google for cross-instance search: “Mastodon [url]”.

#fediverse #mastodon #search

Today I joined demo day of the Post-Growth Entrepreneurship Incubator. It is always interesting to learn about alternative business models for startups, and it was great to see the wide diversity in terms of products, regions and founders!

More info about the incubator and related course: https://nonprofit.ventures

#postgrowth #startups #business #businessmodels #socialenterprises

I love Mastodon so far (and the people using it), but the unavailability of full-text search drives me crazy. On Twitter, my number 1 method of finding new people to follow was searching for tweets mentioning long-form articles I liked.

#mastodon #twitter #fediverse #search

Did your startup or scaleup ever work with consultants, and if yes: how did you find them, and what was the experience like? Would love to hear your horror and/or success stories.

Asking as I'm working on some customer research today! I'm trying to figure out how Datakami can make the consulting experience as good as possible for our startup friends.

#startups #scaleups #consulting

Observations from a startup consultant: "Either you’re swimming in technical debt or you’re just too early or too unsuccessful to have accumulated it."

https://medium.com/@aq/some-things-i-ve-learned-from-startup-consulting-154424576a12

#startups #consulting #technicaldebt

@5uie1 Haha, in the end I decided I don't WANT to integrate email in my app at all. And yesterday I had a bit of a fallback when I almost added RSS reader functionality, but after 100 lines of Python common sense prevailed.

Last month I wrote a Flask app for managing my research notes and I can tell you that Zawinski's Law of Software Envelopment is real:

"Every program attempts to expand until it can read mail. Those programs which cannot so expand are replaced by ones which can."

https://en.wikipedia.org/wiki/Jamie_Zawinski#Zawinski's_Law

I'm taking some time today to test a few new libraries/tools.
These CLI tools for working with llms by @simon work like a charm! And they support unix pipes. <3

More info here: https://llm.datasette.io/en/latest/index.html

#llms #gpt #languagemodels #cli #terminal #commandline

Ethan Mollick sees three stages of AI adoption at companies:
1. ignore it
2. ban it
3. centralize it by building your own tooling, e.g. a chatGPT wrapper

(these sound more like three stages of denial to me)

https://www.oneusefulthing.org/p/on-holding-back-the-strange-ai-tide

But what I'm missing here is a fourth option:

4. talk about it

Companies should define a good policy for AI tool usage, e.g. experimenting is fine but please don't feed customer details to third-party apps, and encourage their employees to share their experiences, worries, mistakes, processes with colleagues.

And ideally, everyone should talk about the effect of new AI tools on their job/business to people outside their company as well. How else are we going to get a society where people understand the limitations as well as the possibilities.

TIL that Mastodon profiles have built-in RSS feeds! For example, you can find mine at https://fosstodon.org/@jd7h.rss

#rss #fediverse #mastodon

Yorick and I will be hosting a tutorial on using llama-index at PyData Amsterdam. I looked at the schedule yesterday and I'm looking forward to meeting all the wonderful people I've already spotted in there 👀 @ines @koaning @vicki

Schedule: https://amsterdam2023.pydata.org/cfp/schedule/
Conf website: https://amsterdam.pydata.org/

Datakami blogpost by Yorick van Pelt: "Overview of open and commercially usable LLMs" 🧐

New models are coming out every day. Here's an overview that can aid in selecting one for commercial use.

Read it here: https://datakami.nl/blog/2023-07-11-overview-of-commercially-usable-LLMs

#llms #textgeneration #languagemodels #genai #llama #generativeai #datasets #generativeai

@b0rk
- I use Mastodon in the same way as I used Twitter (interacting with peers/strangers about my field) but then with more hashtags so people actually get to see my posts.
- HackerNews and a couple of newsletters are worthwhile sources of new information, but read-only
- I also use IRC but only with close friends.

Dutch linguist Mark Dingemanse (@dingemansemark) and co-authors compared LLMs on openness, transparency, and accountability.

Interview in Dutch national newspaper: https://www.volkskrant.nl/nieuws-achtergrond/verantwoord-een-chatbot-gebruiken-chatgpt-is-de-slechtste-keuze~b734f31e/?appview
Research paper: https://arxiv.org/abs/2307.05532

#nlg #textgeneration #llms #chatgpt #openai #transparency #AI #NLP #languagemodels #research #arxiv

TIL "memex" is a portmanteau of memory and expansion.

https://en.wikipedia.org/wiki/Memex

Memex - Wikipedia

Wonderful Wired story (2018) about digital books and publishing.

"The 'Future Book' Is Here, but It's Not What We Expected"
https://www.wired.com/story/future-book-is-here-but-not-what-we-expected/

#books #publishing #crowdfunding #newsletters #audiobooks #wired

Wood Wide Web is a wonderful wholesome rabbit hole of words and trees. Give it a try it if you like poetry, walking simulators, digital gardens, indie games, the indie web, pixel art, or text generation.

Play it here: https://vividfax.itch.io/wood-wide-web

#indiegames #itchio #walkingsim #poetry #digitalgarden #indieweb #nature #cozygames

TIL: Search on Mastodon works differently from search on Twitter. We can't do full text search on toots at all. So if I want people to discover my stuff, I guess I should start using hashtags more?

#mastodon #search #fediverse

Podcast interview with two investors/entrepreneurs that created their own funds for investing in diverse founder teams and climate tech. Inspirational stuff: https://www.youtube.com/watch?v=EWanQGZ-SSk (podcast in Dutch)

#venturecapital #investors #investing #entrepreneurs #founders #startups #diversity #climatetech #podcast #dutch

New Datakami blogpost: "Industry dynamics in Generative AI" ✨

In Good Strategy/Bad Strategy, Richard Rumelt explores how we can take advantage of significant shifts in society to gain an edge over competitors. This works well, even for really small companies. Rumelt provides thinking points for analyzing these changes in his book. How can we apply Rumelt's ideas to generative AI?

Read it here (on our brand spanking new website): https://datakami.nl/blog/strategy-industry-dynamics-of-generative-ai

#generativeai #strategy #business

If you have built something and you put an impressive-looking, complex-sounding PDF up at arXiv, while keeping private half of the resources mentioned in the paper: that's not science, that's marketing.

#science #research

@leonoverweel Nice, using ChatGPT as "Reviewer 0" :')

I like Ethan Mollick's article on using ChatGPT for writing -- not by asking it for texts that you can copy ad verbum, but by using it as a brainstorm buddy.

"How to... use ChatGPT to boost your writing" by Ethan Mollick: https://www.oneusefulthing.org/p/how-to-use-chatgpt-to-boost-your

#generativeai #llms #promptengineering #writing #chatgpt #gpt4

Some tricks from the article:

- Don't ask for math, reasoning, or facts that you can't check yourself
- Experiment with longer prompts: ask for content, form AND style
- Specify the content with a ordered bullet list of main points
- Ask for specifics: a tone of voice, metaphors, a plot twist, textual tropes, text for a specific audience, text length, a style of a publication or author, a narrative arc, examples
- give ChatGPT a writer persona, examples here: https://prompts.chat/

I don't agree with the main point of Bogost's article, but sentences like this make it worth reading: "My co-workers and especially my wife were repulsed by the “CrackBerry,” and my compulsive grasping at it, like Gollum with his ring." :')

What Did People Do Before Smartphones? No one can remember by Ian Bogost in The Atlantic
https://archive.ph/pUmLx

New definition of "AI winter", inspired by https://solar.lowtechmagazine.com/about/the-solar-website/. Imagine language models (and other compute heavy stuff) running on solar-powered tiny devices -- and shutting down when their battery is empty.

LOW←TECH MAGAZINE About the Solar Powered Website This website is solar-powered and self-hosted. It has been designed to radically reduce the energy use associated with accessing our content.

@Mara Beautiful!

@zachwhalen Phew, that's a long todolist for a sabbatical! It might be a good idea to come up with a "definition of done" for yourself ("I will work on serious stuff for max X hours a day") so that you can really enjoy the remaining time off...

Wonderful ode to audiobooks from The Atlantic. I was inspired to give audiobooks a try again.
https://archive.today/UiNVh

Interview that describes how Latacora, a security firm for startups, works with their clients. Super interesting! https://tailscale.com/blog/latacora-and-tailscale/

"[Their AI/ML] dependency tree is more of a dependency kudzu forest." :')

Another good security quote:

"I think hacker types have a tendency to dismiss auditors: “They [don't] understand how computers work. Why should I take anything that they say seriously?” And auditors have a similar tendency to dismiss hacker types: “I can’t even get these people to write down what they do, let alone reason about it. They’re flying by the seat of their pants. Why should I take anything they say seriously?”. And part of the problem is that they’re both often sort of right."

I love epistolary novels, and today I learned there exists an epistolary science-fiction mystery novel/game in the form of a Wikipedia-like website! It's called Neurocracy.

Official website: https://neurocracy.site
Guardian review: https://www.theguardian.com/games/2021/sep/16/neurocracy-murder-mystery-wikipedia-experimental-game

Thanks to @mtrc for bringing cool games to my attention. ;)

The new Zelda game sounds intriguing!

"Even in determinedly open-world games, such as Grand Theft Auto or The Witcher, the possibility space feels as if it has been carefully catalogued long before the public arrives. In Tears of the Kingdom’s Hyrule, though, it feels as though the chaperones have abandoned the playpen and left the toys unattended."

https://www.newyorker.com/culture/cultural-comment/the-open-world-genius-of-the-legend-of-zelda-tears-of-the-kingdom

In my local library I stumbled upon "Een regenjasje voor het aapje" ("A raincoat for the monkey") -- a Dutch translation of Sarumino, a book of Japanese poetry written by haiku-poet Basho and his pupils.
More info here: https://meandermagazine.nl/2022/08/bashos-sarumino-een-regenjas-voor-het-aapje/

#haiku #renga #japanese #poetry

What I love about the book is that the translators chose to share their translation process with the reader: the right pages show the translated poems, the left show the original Japanese, a literal Dutch translation, and some comments on the cultural and historical context. This makes it so much easier for me to understand the poems -- I can't read the original Japanese and I don't have a degree in Japanese literature.

Most haiku contain jokes or have multiple meanings -- and because I have the literal translation at my disposal as well, I can philosophize about other (better?) translations. Fun!

@jurie I use Pandas all the time for filtering. Good introduction tutorial here: https://github.com/brandon-rhodes/pycon-pandas-tutorial

@jurie If the JSON files are not too big to load in memory, I would use Pandas.

This reminds me of "NLP conferences before the deep learning community joined" vs "NLP conferences after the deep learning community joined". 😂

https://friend.camp/@aparrish/110476631599177768

@alexjc@creative.ai Jane Goodall + a journalist wrote a book about this! It's called Hope, and it focuses on Goodall's conservationist and environment work.

Hello Fediverse! To kick off this brand-new profile, here's a new blogpost: https://judithvanstegeren.com/blog/2023/job-hunting-tips-for-people-in-tech.html

TL;DR: tips for job hunting based on my experience transitioning from academia to industry:
- defining a list of dealbreakers,
- dreaming up your ideal job,
- connecting with people already in your dream job,
- reading some good internet advice,
- searching vacancies using technologies as keywords rather than role titles,
- providing a 'tech translation' in your CV.