Claude le arruinó el negocio a los detectores de IA

Value of the Information & Strength of the Argument

The video provides valuable insights into the implications of AI watermarking and the broader context of AI in education and society. The host argues that watermarking is not inherently bad and that the real issue is the misuse of AI by both students and educators. He emphasizes the importance of learning and practical experience over mere certification. The argumentation is persuasive but relies heavily on personal opinions and anecdotes rather than empirical evidence. The host’s critique of the education system and the certification culture is thought-provoking, but it lacks rigorous data to support the claims.

Scientific Rigor, Source Quality, Title Accuracy

The video does not cite specific scientific sources or provide references for the claims about Anthropic’s watermarking. The host mentions that the information is based on public announcements and speculation. The title accurately reflects the content, focusing on the impact of Claude’s watermarking on AI detectors. The video includes promotional segments for EDteam courses, which are clearly marked. The overall scientific rigor is low, as the content is primarily opinion and commentary rather than evidence-based analysis.

The title is catchy and relevant, as it focuses on the impact of Claude's watermarking on AI detectors, which is the main topic of the video.

6/10

The video is an expert opinion and commentary rather than a rigorous scientific analysis. It discusses AI watermarking and its implications, but relies on anecdotal evidence and personal views. The information about Anthropic's watermarking is based on public announcements and speculation, with no primary sources cited. The discussion on education and certification systems is subjective and not backed by data.

• 00:00Intro
• 08:36Promo: Nuevo curso de Playwright Testing End to End con IA
• 09:03Promo: Nuevo curso de agentes IA con Eve y Vercel con Beto Quiroga
• 09:32Promo: EDcamp 11 en Bogotá, 16 y 17 de octubre
• 11:41Anthropic va a marcar con IA todo el texto que genere Claude
• 15:04La marca de agua va en los tokens y ningún skill puede quitarla
• 17:13¿Qué le recomiendas a alguien en séptimo semestre de ingeniería?
• 23:36¿Por qué la marca de agua solo debería asustar a quienes no aprenden?
• 29:25La policía del Perú quería comprar certificados sin aprender nada
• 33:42La mayoría paga para que le hagan la tesis y a nadie le importa
• 38:53Los papers académicos también están bajo ataque de la IA, y es normal
• 40:36El MINSA puede digitalizar todo, el estado no lo deja
• 56:41Promo: Nuevo curso de Google AI Studio en EDteam
• 57:05Los agentes propios de empresa son la siguiente evolución de la IA
• 1:03:44El caso Indecopi explica qué son los logs mejor que cualquier manual
• 1:07:02¿Puede la IA crear y desplegar un Airbnb completo sin errores?
• 1:10:52Los juniors siguen teniendo trabajo si aprenden y no solo pasan la IA
• 1:34:02Promo: Becas del 50% y cursos de IA con Álvaro Felipe en EDteam
• 1:35:43La IA genera slop y la UE obliga a marcarlo a partir de 2026
• 1:44:06Anthropic no anunció las marcas de agua: alguien encontró el cambio en su documentación
• 1:49:41Así funciona la marca de agua de Claude a nivel de tokens
• 1:57:19Los detectores de IA como Turnitin son un negocio que no funciona
• 2:09:40El camino correcto no es prohibir la IA sino enseñar a usarla bien

• Adopción de IA en las empresas (Gobernanza) — Mentioned as a course related to AI governance.
• Creacion de agentes IA con EVE y Vercel — Promoted as a new course on AI agents.
• Playwright: Testing end-to-end con IA — Promoted as a new course on testing with AI.
• Apps empresariales sin código con Power Apps — Promoted as a course on no-code business apps.

• Anthropic's announcement on watermarking — Official announcement of Claude's watermarking feature.

• Criticism of AI detectors — Research showing AI detectors are unreliable and biased.