203 private links
UX and AI, but no single speaker addressed the training data sources, the energy requirements,
But never once did the question arise of whether it’s ethical to even use these tools.
One topic was expressed: the AI slop
There’s a quote by Finnish architect Eliel Saarinen that UX designers like repeating:
Always design a thing by considering it in its next larger context. A chair in a room, a room in a house, a house in an environment, an environment in a city plan.
As Molly White states:
There are no ethical uses of current large language models.
Similar to a gardener.
Play around with ideas, follow intersting threads and see where it goes.
When something start to make sense, a TODO list follows.
As for staying motivated during the build process, while I don't have anything prescriptive I did write a post on what works for me.
Les coûts directs:
- légaux ou juridiques: amendes ou sanctions.
- les corrections
- les surcoûts techniques lorsque l'infrastructure n'est pas optimisée.
Les coûts indirects:
- la communication sur les incidents, SAV, etc...
- les pertes commerciales (dédommagement)
- la perte de productivité ou l'absentéisme
- la formation
engendrent du manque à gagner
- la visibilité (publicité supplémentaire nécessaire)
- la non fidélisation (churn) des clients existants
- l'image
Il y a aussi des coûts de surqualité !
Le sous-ensemble latin étendu a 395 caractères et pèse 395ko.
Pour le chinois ou le japonais, il peut y avoir de 3 000 à 80 000 caractères.
L'optimisation habituelle consiste à créer un sous-ensemble, mais il faut néanmoins au moins plus de 5 000 caractères pour certaines langues.
L'idée est donc de créer une police qui peut être segmentée, "contenant uniquement le sous-ensemble de ce qui est critique, puis de lui adjoindre des additions qui complètent la police en cours de route."
Un streaming de caractères de police en somme. Cette idée est actuellement un draft au W3C: Incremental Font Transfer
Optimization is not always a progress in every field
🏳️⚧️ Meuf trans 💻 Développeuse web 📚 Passe son temps libre à lire des romans. 👂💍 Peut parler des heures de piercings et tatouages 💪 Active dans le milieu queer
Les quatres manières de lire:
- indexation
- skimming
- analyse
- sémiologique
Concernant le code:
- Ne pas attribuer à la malveillance ce que la bêtise peut expliquer ;
- Ne pas attribuer à la stupidité ce qui est expliqué par le manque d'attention ;
- Ne pas attribuer au manque d'attention ce qui peut être expliqué par un contexte non dit ;
- Ne pas attribuer à un contexte non dit ce que le système autour de la personne peut expliquer ;
major.minor.patch works well but not for applications.
- Simplicity is essential
- Solve problems instead of creating them
- We are not smarter than others, others are usually not smarter
- Do everything yourself
- Strive for robustness
- Do not think you can make computing "secure"
- Use input devices when they make the most sense
- Avoid all ornaments
- Tools are just tools
- Be humble
- Don't work for free if you do not enjoy it
- Do not listen to others
About high and low-programming languages
I go often for the state based version, based on component states.
Nearly all the state is handled in the view component.
Are we web yet? Well yes but with extra work. A lot of work.
The author provides its experience.
All the things I love about Django, could we have those for a Rust web framework so that we can reap the benefits of Rust without having to go needlessly slowly?
An attempt: https://git.sr.ht/~ntietz/newt
Never mind that you can also harvest code from any of your shelved projects. I mean why rewrite the Fisher–Yates shuffle algorithm if you have it already in a shelved game? Code for switching the monitor depth (on those early Macs) I moved from game to game… Polygon-point collision code, a sine lookup-table for quick trig functions, a dot-product routine, cross-product routines…
was the whole exercise of my guerilla programming technique a wash? Maybe. But it always served me well
There were times too when a coworker might have said, “You should have used a Bloom Filter” and I was able to come back with, “Yeah, already tried that but the typical data we are seeing is so small that the performance gains were negligible and added unnecessary complexity to the code base so I tossed it.” Boom!
A.B.I Always Be Iterating.