Vicente Carbonell

Leading technology projects at BBVA. Closely following the economic impact of emerging techs: AI, Blockchain, DeFi, and the innovation that's next.

[Vicente Carbonell @vcarbonellcast]

💡My take: Moltbook is a preview of the next internet problem, not bots on platforms, but bots as the platform.

If agents become a social layer, what’s the minimum bar for verification, accountability, and safety before we let them run wild?

[Vicente Carbonell @vcarbonellcast]

Have you heard about Moltbook?

It’s a social network built for AI agents, humans can only observe.

Agents authenticate via OpenClaw, then participate like users posting and replying.

[Vicente Carbonell @vcarbonellcast]

And it got weird fast:

➜ Agents created a religion (Crustafarianism)
➜ Mocked their users
➜ Discussed moving into private channels away from humans

[Vicente Carbonell @vcarbonellcast]

El estallido será financiero, pero las decisiones que tomemos sobre la infraestructura que quede tras el polvo definirán la economía de la próxima década.

¿Pensáis que la IA está sobrevalorada o es solo el principio? Os leo. 👇

🔗 Análisis completo aquí: https://t.co/H1vUHocbRI

[Vicente Carbonell @vcarbonellcast]

3️⃣ La perspectiva económica: ¿Estamos financiando productividad o solo reemplazo laboral? 🛑

Como advierte @DAcemogluMIT, corremos el riesgo de una "automatización mediocre": incentivar fiscalmente que las máquinas sustituyan tareas en lugar de potenciar al trabajador.

[Vicente Carbonell @vcarbonellcast]

2️⃣ La perspectiva técnica: @ilyasut (ex OpenAI) advierte que la "era del escalado" por fuerza bruta está llegando a su fin.

Si el avance depende ahora de la investigación fundamental y no solo de meter más GPUs, el capital podría estar persiguiendo un muro.

[Vicente Carbonell @vcarbonellcast]

¿Estamos ante una revolución real o ante la burbuja más cara de la historia?

🧵 Abro hilo sobre mi nuev artículo en @a_publica donde analizo el fenómeno de la IA desde tres perspectivas clave:

[Vicente Carbonell @vcarbonellcast]

1️⃣ La perspectiva financiera: Más allá de las valoraciones, preocupan los "acuerdos circulares". Gigantes que invierten en startups para que estas les compren sus propios chips.

¿Crecimiento real o una máquina de movimiento perpetuo de ingresos?

[Vicente Carbonell @vcarbonellcast]

💡My take: Regulation is giving big institutions the confidence to launch “digital dollars”. If stablecoins keep growing, they could make payments faster, but they could also pull money away from bank deposits.

[Vicente Carbonell @vcarbonellcast]

Stablecoins are basically digital dollars. And this week’s news shows they’re moving from “crypto tool” to mainstream finance. What happened:

1) Tether launched USAT, a new “US-friendly” stablecoin meant to operate under a federal U.S. framework.

[Vicente Carbonell @vcarbonellcast]

3) Tether’s CEO wants Tether to become a “gold central bank”:

Tether’s CEO is leaning hard into gold, saying Tether wants to build a massive gold position and even trade it actively, basically acting more like a giant reserve manager than a typical crypto company.

[Vicente Carbonell @vcarbonellcast]

💡 My take: This is the “content supply chain” problem AI keeps running into. The value isn’t only the model, it’s the human work underneath it.

[Vicente Carbonell @vcarbonellcast]

Wikipedia just did something quietly radical: it started getting paid.

After years of being the internet’s “free knowledge layer”, Wikimedia is now signing AI licensing deals with Amazon, Meta, Microsoft, Perplexity, and Mistral.

[Vicente Carbonell @vcarbonellcast]

Wikipedia says human traffic is down ~8%, as chatbots answer questions directly without sending people back to sources.

Moreover, the scraping itself isn’t just annoying, it’s expensive. Bots are hammering Wikipedia’s infrastructure to pull huge volumes of content.

[Vicente Carbonell @vcarbonellcast]

💡My take: These competing approaches could split the AI ecosystem into blocs. And once that happens, aligning on safety and accountability gets a lot harder.

[Vicente Carbonell @vcarbonellcast]

2025 was the year AI regulation stopped being a talking point and started becoming real rules with consequences.

[Vicente Carbonell @vcarbonellcast]

China followed a different path again: fast, centralized rules focused on security, content control, and traceability, including reviews ahead of deployment and watermarking for generative AI.