presende

AI Engineering Manager @OutSystems - working on AI for low-code and DevEx

[presende @pt_resende]

@cruncher990 @MegaCrit Casual? I almost take offense to that… Otherwise I very much share your excitement!

[François Chollet @fchollet]

All the great breakthroughs in science are, at their core, compression. They take a complex mess of observations and say, "it's all just this simple rule".

Symbolic compression, specifically. Because the rule is always symbolic -- usually expressed as mathematical equations. If it isn't symbolic, you haven't really explained the thing. You can observe it but you can't understand it.

[presende @pt_resende]

@EmpyreanHuskarl @pmarca That’s not the logic here… The logic is that the market may have disconnected from job openings due to the promise of AI, not that AI is causing a reduction in job opens.

[presende @pt_resende]

@thomasbroadfoot @karpathy Sutton’s argument is that imitation by itself is not enough. There are periods in life when it plays a role, but when and how much is defined by the rewards that the overall generic learning system defines for that at that particular point in time.

[presende @pt_resende]

https://t.co/0o3i24iW5v is finally out!

[Jeremy Howard @jeremyphoward]

I replicated this result, that Grok focuses nearly entirely on finding out what Elon thinks in order to align with that, on a fresh Grok 4 chat with no custom instructions.
https://t.co/NgeMpGWBOB

[presende @pt_resende]

@__marlonw @burkov Why would one use perplexity instead of a regular llm? I feel like they (the best ones, at least, like Gemini) can achieve the same results and then do much more on top of it. Genuinely curious because I’ve tried perplexity several times now and it never stuck

[Ruben Hassid @rubenhassid]

BREAKING: Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all.

They just memorize patterns really well.

Here's what Apple discovered:

(hint: we're not as close to AGI as the hype suggests)

[dono da c🅾️🅾️perativa ™ @donocooperativa]

🚨Não é habitual partilhar temas pessoais por aqui. Mas há coisas que para mim não devem passar em claro. Encomenda no Continente online entregue no dia 19 de abril, véspera de domingo de Páscoa. 4 x 2 paletes de hambúrgueres entregues com a validade do dia seguinte, 20 de abril.

Que abuso, pensei. Vou reclamar, pedi reembolso ou a troca da mercadoria. Demoraram a responder. Tentaram ligar-me quase uma semana depois, não consegui atender, responderam por e-mail para me dizer que me tentaram ligar. Obrigadinho.

Pediram-me uma melhor hora para contacto, respondi por e-mail, a partir das 18h. Não lhes devia ser conveniente, a eles, nunca mais me ligaram de volta. Entretanto, domingo abri nova reclamação.

Reponderam hoje por e-mail. Destaco o seguinte trecho: "Relativamente à validade da carne, informamos que, sempre que possível, enviamos os artigos com maior validade. No entanto, e conforme o procedimento, a validade mínima de 2 dias, incluindo a data de entrega, foi cumprida."

E complementam "Para situações futuras, sugerimos que indique, nos comentários da encomenda, a validade mínima pretendida" . Ou seja, eu é que tenho a obrigação de dizer que não quero validades à queima para a carne que encomendei. Como se não fosse um comportamento normal num consumidor. A subversão completa da lógica!

Achei indecente. Sou cliente há vários anos, nunca reclamei de nada, nem de uma fruta um pouco mais passada do que deveria, ou de uns iogurtes com validade curta.

Neste caso, pelo abuso de confiança, mas também pela resposta, dou hoje por terminada a minha relação de décadas com o Continente.

Já pedi o encerramento da conta online, do cartão e também para apagarem todos os meus dados pessoais. Relações de má-fé, não estou interessado, obrigado.

Quando se preza um cliente, não se fazem destas coisas. Quando por lapso acontece, corrige-se com o menor incómodo possível para o cliente.

O Continente não fez nenhuma das duas. É porque não precisa de clientes. Não faz mal. Eu também não preciso do Continente.

Até nunca...

o dono da cooperativa

[Yuchen Jin @Yuchenj_UW]

I have some CS PhD friends who'd been single their entire lives but suddenly all got girlfriends. When I asked how, they said: "AI."

Yep, AI.

They used to spend thousands of dollars on Chinese dating courses. Now, they just paste WeChat messages into ChatGPT and copy AI's responses back.

Turns out AI flirts way better than they ever could. A good sign that the Turing Test has been officially passed.

It's time for a dating benchmark. AI might already beat most tech nerds.

Not sure if this counts as "cheating", but at least AI finally found a way to boost the human birth rate.

[presende @pt_resende]

@Mylovanov What? No, it’s not… It’s about the rare earth minerals and money, nothing else, from both sides. Same thing with Greenland. Ideology plays second fiddle to money and power, and is mostly used as the excuse.

[eu/acc @euacc]

Top LLM models sources of funding:
- OpenAI, private investment
- Claude, private investment
- DeepSeek, private investment
- Meta AI, private investment
- Gemini, private investment

Meanwhile, the 🇪🇺 EU:
- "Let's build an LLM with government funding"

Makes no sense

Instead, the EU should immediately repeal the EU Artificial Intelligence Act that literally bans most AI companies from operating within the EU

If Europe wants to be competitive in AI (or tech in general), the answer isn't spending more European taxpayer's money

The answer is removing the ridiculous regulations that prevent Europeans from building AI and tech companies right now

[François Chollet @fchollet]

Many Americans apparently believe they are engaged in some kind of civilizational battle against China. I really don't think that's the case. The global fracture line is autocracy vs democracy, and the US will find itself increasingly aligned with (and allied with) China & Russia in that regard

[presende @pt_resende]

@zharta Looking amazing! Congrats guys 🚀🩷

[Zharta Finance @Zharta]

A bold new era for Zharta and NFT Finance starts NOW.

Today, we unveil our new P2P NFT lending protocol and a complete rebrand!

Doubling down on NFTs.

Make your move 👇🧵

[presende @pt_resende]

@zharta Can’t wait to see what you’ve been cooking, I’m really curious!

[presende @pt_resende]

@zharta Hyyyyyype 🔥

[presende @pt_resende]

@MonkusAurelius @RaoulGMI @OpenAI People who talk about AGI like this either: have no idea what they’re talking about, are farming engagement or, most likely, both. No company that is developing AI expects AGI any time soon, or believes that current models are able to achieve it

[Zharta Finance @Zharta]

Just launched: Rent BAYCs, MAYCs, HV-MTLs & more to pump your Dookie Dash score!

Players get GOAT 🤝 Non-players EARN

@BoredApeYC x @farawaygg just dropped the game of the year, and we’re backing it

Not renting? Not playing.
@CryptoGarga, @GordonGoner , @apaley 👀

[presende @pt_resende]

@zharta @NFCsummit @BoredApeYC 💚😄