My theory:
if it ever happens that enterprises improve productivity using LLMs (which I doubt), chatbot vendors will just raise the subscription price to collect the suprlus, absorbing all the gains.
There is no way to win anything from this tech.
✅ Fact:
The #ClaudeCode logo represents the orifice that produced its own code
Donc, en étendant les chiffres de cet article sur 6 mois, ca leur a rapporté 2,2 millions et couté 2,7 milliards... Sans compter le cout environmental bien sur. Et les utilisateurs voulaient pas payer 1000x plus cher ?
Heureusement que y'avait des bon cotés, comme ....
Ah ben non.
https://www.lesnumeriques.com/intelligence-artificielle/openai-tue-sora-six-mois-de-vie-2-millions-de-dollars-de-revenus-et-un-partenariat-disney-parti-en-fumee-n253481.html
#sora #ia #ai
nuget indexing seems to be lagging...
Version 0.3.0 of Fastoch directly integrates HMR to avoid indirect dependency on React.
Fastoch is a virtual DOM for #fsharp Fable Elmish with minimal dependencies.
According to wikipedia, starlink has 3 millions customers in may 2024, so it is 1 satellite for 500 customers!
The environmental cost of this thing is abysmal
Et oui, ce sera en #fsharp, off course.
Demain je serai à #devoxxfr2023 pour présenter
Fun(ctional) Domain Driven Design avec du #ukulele dedans !
C'est à 16h45.
https://cfp.devoxx.fr/2023/talk/QAD-8702/Fun(ctional)_Domain_Driven_Design_(featuring_Ukulele)
My small contribution to #fsharp is now released in #fsharp 7.0:
The compiler now emit branchless code for 'compare x y'.
The return value is 0 if x=y, <0 is x<y and >0 if x>y.
It is tempting to implement it as y-x, but this doesn't always work because of overflow.
To fixe this it emits (cgt x y) - (cgt y x).
Branchless version is 3x faster on average. It is used heavily in Map and Set implementation.
Here it is:
