Replying to on tech.lgbt
@Lydie@tech.lgbt @jt_rebelo@ciberlandia.pt @RegGuy@mstdn.social Used to connect to BBSs at that speed 👍 "Good enough" but not really, it was painfully slow.
Remote
Ricardo Tavares
@t_var_s@phpc.social
mastodon 4.7.3Passionate for the promise of an open web.
I blog at https://viewfromtheweb.com
0 Followers
0 Following
50 Posts
Joined May 06, 2022
I love writing,
https://blog.viewfromtheweb.com/is-it-a-clanker-writing-this-blog-96f4a978/
but I also love just making things with computers.
#llm #writing #art #ai #blog #illustration
Replying to
@gafanhoto@masto.pt "No bairro do amor, há quem pergunte a sorrir: será que ainda cá estamos no fim do Verão?" 😭
"The rise of artificial intelligence is the next industrial revolution."
So... decades of growth powered by innovation, widespread suffering and environmental damage, followed by worker organization and political reform that eventually secure a broader share of that wealth across society?
That is, if we get there. That industrial revolution?
No wonder you get booed for saying that.
Replying to
@agreeable_landfall@mastodon.social @zzt@mas.to @joXn@wandering.shop @benfry@information.garden @cstross@wandering.shop @mjtsai@mastodon.social @mwl@io.mwl.io This is when the big ecosystem drops an OS/store update and you have to keep shipping at least every year just to keep your software afloat.
FBI: "HACKERS CAUSE CHAOS"
Hackers: "hello sir yes you have a security bug here ok I'll wait"
the road to hell is paved with good employees
Welcome to the Zero Pushback Podcast! Where we talk to so-called experts for five hours to mine all the clickbait and ragebait out of their grifts and crazy takes. Brought to you by whatever health supplement we're currently making money from.
How much ambiguity can be useful to your projects and where do you place it?
https://blog.viewfromtheweb.com/the-ambiguous-intelligence-of-coding-agents-9a2d96f5/ is the latest post on the blog
Turns out you can solve #math through a judicious approach of just trying lots of stuff: https://openai.com/index/model-disproves-discrete-geometry-conjecture/
Seems to line up with the #mythos #llm hype where it still matters how you deploy this new artillery of resources, but it's possible to automate getting some results.
But how much of this is just money being poured into domains that usually have very little investment? #ai is far from being the perfect excuse to start doing the right things, but we might as well take it.
I keep saying that any serious plan to introduce #ai in schools (whatever that means), has to include learning #philosophy as early as possible.
Put down the google/microsoft subscriptions. Ban smartphones in classrooms. How do we know anything is true? Can you think for yourself and ask good questions? Can you feel how your choice of words molds your rationale?
Replying to
@maxleibman@beige.party This is the part when Sam Altman gives you money in exchange for scanning your eye balls. But yes, nobody even asks the question to hear about this answer.
Replying to
@splendorr@mastodon.social @baldur@toot.cafe I recognise the elephant in the room, but I like to think about web components or smaller frameworks.
I prefer not having a build step, but there's a whole world of options that opens up once you do.
Replying to
Just build moar, bro: https://portugaldispatch.com/article-housing-foreign-capital
Add more caviar to the menu, that will surely help people starving.
#Portugal #Housing #Inequality
Practical harm is way more relevant right now than theoretical harm.
Given enough time, even if #ML researchers can't do nothing with it, we're saving lives by simply not having #AI accelerate the people-crushing machines currently at the center of our societies. https://toad.social/@KimPerales/117339458314218196
Replying to
@slightlyoff@toot.cafe @baldur@toot.cafe Still developers aren't completely wrong when thinking of components as bits of HTML + CSS + JS
"Before we go folding our clothes in a stack,
let's see what the robots think.
Before we go patting ourselves on the back,
let's see what the robots think.
Before we start working or dancing again, before we start writing love letters to send, before we start fighting or finding a friend, let's see what the robots think."
-- Bo Burnham (Egghead, 2013)
Social media is just some websites on the internet that people keep going to.
And governments like to make lists of websites they don't want people to go to.
But they let you go to the website if it tells you to vote for them and that works.
Replying to
@jcoglan@mastodon.social There is prompt caching, although I'm not entirely sure how it works.
Basically everyone wants to solve that problem, but it's still easier to manage state by just shoving the whole conversation again down that pipe.
Replying to
@carnage4life@mas.to 3 out of 10 people is still a lot. I heard 19% from a different poll.
Developers everywhere are using LLMs to get unstuck or try something new. You can get the flamethrower and ban all the Claude commits but that's only glancing the surface. People don't need to ship LLM code to do security audits, pinpoint bugs and bounce ideas, specially if they're working alone. We can all be benefiting from LLMs and still criticise abusive data centers and the many AI grifts. It's ethical consumption under capitalism all over again.
I wonder if the people comparing the latest #ai wave to the birth of the internet were there when the BBSs opened up to a whole wide web of new connections. I wish it was the same, but all the arrows are pointing in the opposite direction: less jobs, more centralisation.
Not sure if agentic coders are able to grok this: https://www.selc.org/news/trump-administration-attempts-massive-power-grab-in-defense-of-xai/
"hey, it's a cheaper 5.5"
Replying to
@LilaHexe@fedi.lilahexe.top @algernon@come-from.mad-scientist.club Right now I'm using B2 with rclone, since I was already using Backblaze
Pelo Represas, batam em latas. Rompam aos saltos e aos pinotes. Façam estalar no ar chicotes. Chamem palhaços e acrobatas.
Mas não é o fim.
#Portugal
Since the first time we zapped a rock, you could always say that all of this amounts to nothing when we're just zapping rocks.
Replying to
But of course https://www.nytimes.com/2026/08/05/climate/data-centers-pollution-trump-ai-energy.html?unlocked_article_code=1.3VA.suy8.Oa2GGtzD2uG8 no chance can be wasted to make everything worse instead. Dropping average life expectancy for the sake of unequally distributed GDP is an horrific national agenda.
Replying to
A balanced stance considering that a lot of members who could vote have abstained or simply not cast a vote.
Also lets historic projects like linux, vim, rsync, etc. incorporate LLMs without confronting that issue head on.
Enforcement is difficult, so this decision mostly gives codeberg another knob to ease pressure on their infrastructure.
Not here to compete with GitHub, much like Mastodon isn't here to compete with Bluesky.
Replying to
Inflation is also rising artificially in #Portugal with a scheme that has already failed in Scotland, Malta, the Netherlands or Ireland. As always, "Volta" starts by charging more for plastic packaging and then eventually you can redeem it. But there's no incentive to make redemption actually work, so you just get charged more. Everybody profits except you. Let the boycotts begin.
Replying to
@inthehands@hachyderm.io @baldur@toot.cafe As an abstract concept, I guess LLMs as an environmental disaster are somewhere between video streaming and cruise ships.
Replying to
@eduzsh@mastodon.social Projects like vim or rsync eventually have an increasing percentage of code touched by LLMs.
https://blog.viewfromtheweb.com/just-using-a-chat-app-for-my-to-dos-0cb188be/
Let's talk about something else on the blog 📝
People keep saying that frameworks like #tailwind or even #stylex are good for agents, but are they working in our current reality where even frontier LLMs still suck at frontend and #CSS?
I'm not even saying that you should look at the code, but often it helps a lot if you're able to inspect element into something that's wrong and be able to tell an agent what to do with actionable words. Like, this class, label or id, lose the padding, don't use absolute units, etc.