Trung Phan

Este señor publica una newsletter semanal al más puro estilo @jefedelforo.
La temática no es puramente inversora pero si es verdad que suele hablar sobre muchas de las empresas más populares por aquí (mayormente tecnológicas). Este sábado me ha parecido muy interesante el resumen que hace del devenir de Alphabet con la Inteligencia Artificial.

Google’s AI journey has been…interesting. It was on the forefront by acquiring DeepMind in 2014 and three years later, Google researchers published the paper on transformers (“Attention is All You Need”), which became the basis for the current generative AI wave. However, in Fall 2022, OpenAI released ChatGPT and took all of Google’s AI mindshare. Google’s response has been slow — because it doesn’t want to disrupt its extremely lucrative 10 Blue Links search business (the best business model in the history of capitalism) — and filled with snafus, including a much-panned fake demo video of Gemini (the company’s ChatGPT competitor).

The entire situation reached a new level of farce last week when Google rolled out Gemini and users discovered that the model basically refused to create images of white people for prompts like “18th century British Royalty”. Or it generated “diverse” — but historically questionable photos — like black Vikings, brown German Nazi girls. It even turned the Google co-founders into two Asian dudes (spoiler alert: they aren’t Asian).

Remember, Google’s mission is to “organize the world’s information and make it universally accessible and useful” not “hey, let’s make sure the DEI department is happy”. Even worse is that many Googlers internally knew this could be bad PR but there’s no incentive to speak out because of Google’s bureaucratic left-leaning corporate culture.

Google said it “missed the mark” and paused the service after apologizing (good summary of the situation from Pirate Wires ).

Here’s the wild thing: despite all this bad PR, Google is in such a strong position to win with a powerful AI model (Gemini has a massive 1,000,000 token context window which means it can ingest and query more information than any other model out there) and unparalleled resources (its own TensorFlow AI chips, cloud computing, Android and the firehose of data from YouTube, Gmail and Search).

Anyways, the memes were out of control

He vivido lo suficiente para ver a una de las empresas más innovadoras del planeta sufrir en carne propia el “dilema del innovador”.

Que rápido va el mundo

Po zí.

La llegada a la luna se produjo sólo 66 años después del primer vuelo de los hermanos Wright (1903-1969). En el transcurso de una vida, los humanos pasaron de tener una tecnología de vuelo limitada a viajar ~239.000 millas desde la Tierra.


Today, we are bringing back one of my favourite internet topics: El Risitas (aka “Spanish Laughing Guy” aka the greatest meme template ever).

Also this week:

  • Buffett’s Annual Letter
  • Wendy’s Surge Pricing Fiasco
  • …and them fire X posts (including Dune 2 )


Lo de este mundo es incomprensible a todas luces, fuera de toda lógica y de toda excel y de todo índice


Leer sobre el Risitas en inglés me ha explotado la cabeza.


Trung Phan es muy top

Lo sigo desde hace años

1 Like

1 Like