Ricardo Lopes
@ricardolopes.net
Software engineer focused on architecting and building secure and scalable services
8yo asked for a maths game, so I vibe coded one. In 1-shot (and 1 fix) we got a nice game that he loved.
Tried later on a different PC. Never worked, going in circles with fixes.
This perfectly illustrates LLMs: they either work like magic, or feel useless. No in-between
Tried later on a different PC. Never worked, going in circles with fixes.
This perfectly illustrates LLMs: they either work like magic, or feel useless. No in-between
October 16, 2025 at 4:44 AM
8yo asked for a maths game, so I vibe coded one. In 1-shot (and 1 fix) we got a nice game that he loved.
Tried later on a different PC. Never worked, going in circles with fixes.
This perfectly illustrates LLMs: they either work like magic, or feel useless. No in-between
Tried later on a different PC. Never worked, going in circles with fixes.
This perfectly illustrates LLMs: they either work like magic, or feel useless. No in-between
Reposted by Ricardo Lopes
Some of the anti-AI stuff feels a bit like when people would say "don't use Wikipedia as a source." It's just like anything else, a piece of information that you weigh against multiple sources and your own understanding of its likely failure modes
April 26, 2025 at 1:23 PM
Some of the anti-AI stuff feels a bit like when people would say "don't use Wikipedia as a source." It's just like anything else, a piece of information that you weigh against multiple sources and your own understanding of its likely failure modes
The deepseek app is topping charts, but it uses V3, not R1, which is the new breakthrough everyone's talking about.
Am I missing anything? V3 was released last month and doesn't use CoT, so it's closer to 4o than o1. Why are people using it now?
Am I missing anything? V3 was released last month and doesn't use CoT, so it's closer to 4o than o1. Why are people using it now?
January 27, 2025 at 8:28 PM
The deepseek app is topping charts, but it uses V3, not R1, which is the new breakthrough everyone's talking about.
Am I missing anything? V3 was released last month and doesn't use CoT, so it's closer to 4o than o1. Why are people using it now?
Am I missing anything? V3 was released last month and doesn't use CoT, so it's closer to 4o than o1. Why are people using it now?
Reposted by Ricardo Lopes
If AI development stopped this week we would have 5-10 years of absorbing the impact of current models on education, culture, healthcare, and business.
But this week has also suggested that development is not stopping.
But this week has also suggested that development is not stopping.
December 11, 2024 at 11:03 PM
If AI development stopped this week we would have 5-10 years of absorbing the impact of current models on education, culture, healthcare, and business.
But this week has also suggested that development is not stopping.
But this week has also suggested that development is not stopping.
Blue screens too 🙂
December 6, 2024 at 11:18 AM
Blue screens too 🙂
Pair programming is your chance to have others listen to your "uh this is weird" and "oh it's a different error now"
December 5, 2024 at 10:03 PM
Pair programming is your chance to have others listen to your "uh this is weird" and "oh it's a different error now"
Remote rubber ducking is wild. You write paragraphs of text in a slack DM, only for the writing chain of thought to deliver you the answer, so you delete everything, ship the fix, and the dev who saved your day won't even know about it!
December 5, 2024 at 10:00 PM
Remote rubber ducking is wild. You write paragraphs of text in a slack DM, only for the writing chain of thought to deliver you the answer, so you delete everything, ship the fix, and the dev who saved your day won't even know about it!
Whoa, OpenAI's new o1 model looks mindblowing!
Here is a fun o1 test. I gave it this XKCD comic & the prompt: "make this a reality. i need a gui and clear instructions since i can't code. that means you need to give me full working software"
It took less than 15 minutes and it didn't get caught in any of the usual LLM loops, just solved issues.
It took less than 15 minutes and it didn't get caught in any of the usual LLM loops, just solved issues.
December 5, 2024 at 9:24 PM
Whoa, OpenAI's new o1 model looks mindblowing!
Hello, world!
Trying out bluesky in case it takes off.
I'm c.im/@ricardolopes on Mastodon and x.com/ricardoplopes on ex-twitter
Trying out bluesky in case it takes off.
I'm c.im/@ricardolopes on Mastodon and x.com/ricardoplopes on ex-twitter
November 21, 2024 at 1:50 AM
Hello, world!
Trying out bluesky in case it takes off.
I'm c.im/@ricardolopes on Mastodon and x.com/ricardoplopes on ex-twitter
Trying out bluesky in case it takes off.
I'm c.im/@ricardolopes on Mastodon and x.com/ricardoplopes on ex-twitter