Just jump over the effing hype curve! 🏃♂️➡️
While others are still climbing, you are already at the other side, seeing the beautiful landscape on the horizon.
marc0s
boosted
La #InteligenciaArtificial, una disciplina a la que he dedicado prácticamente 20 años de investigación, está siendo devorada por las lógicas capitalistas que afectan tantísimas otras cosas que funcionan mal en nuestra vida. El #hype, la burbuja, la moda, el apetito insaciable o como lo llamamos en #IA, el #ColapsoModal.
La IA se basa en la estadística, en identificar patrones en conjuntos de datos que ayudan a predecir unas variables en función de otras. Todo esto se consigue construyendo cálculos basados en combinaciones de los datos de entrada que a su vez resulten en una aproximación al valor que queremos predecir y a su vez definiendo función objetivo: una forma de evaluar lo lejos que está la predicción de la respuesta correcta, de forma que podamos variar ligeramente los cálculos para acercarnos cada vez más. Bien, pues si no tenemos cuidado, podríamos definir esa evaluación de forma que los cálculos acaben haciendo una prediccion trivial: la moda de la distribución, el valor que más se repite. Predecir siempre que en Buenaventura, Colombia está lloviendo garantiza acertar 260 de 365 días al año, pero hace que el modelo no sirva de nada porque en realidad no predice nada sino que repite la #moda. Esto pasa también con las IAs generativas, desde las GANs a los modelos lingüísticos. Y el problema se hace peor si datos generados con IAs se usan para entrenar otras IAs, asegurando el colapso modal. La IA encuentra que puede optimizar desde la #pereza: si lo apuesto todo al valor que más recompensa me da, no tengo que preocuparme de ver más allá.
La #InteligenciaArtificial, una disciplina a la que he dedicado prácticamente 20 años de investigación, está siendo devorada por las lógicas capitalistas que afectan tantísimas otras cosas que funcionan mal en nuestra vida. El #hype, la burbuja, la moda, el apetito insaciable o como lo llamamos en #IA, el #ColapsoModal.
La IA se basa en la estadística, en identificar patrones en conjuntos de datos que ayudan a predecir unas variables en función de otras. Todo esto se consigue construyendo cálculos basados en combinaciones de los datos de entrada que a su vez resulten en una aproximación al valor que queremos predecir y a su vez definiendo función objetivo: una forma de evaluar lo lejos que está la predicción de la respuesta correcta, de forma que podamos variar ligeramente los cálculos para acercarnos cada vez más. Bien, pues si no tenemos cuidado, podríamos definir esa evaluación de forma que los cálculos acaben haciendo una prediccion trivial: la moda de la distribución, el valor que más se repite. Predecir siempre que en Buenaventura, Colombia está lloviendo garantiza acertar 260 de 365 días al año, pero hace que el modelo no sirva de nada porque en realidad no predice nada sino que repite la #moda. Esto pasa también con las IAs generativas, desde las GANs a los modelos lingüísticos. Y el problema se hace peor si datos generados con IAs se usan para entrenar otras IAs, asegurando el colapso modal. La IA encuentra que puede optimizar desde la #pereza: si lo apuesto todo al valor que más recompensa me da, no tengo que preocuparme de ver más allá.
hypebot
boosted
Bundesminister für Digitales in seiner Rede: »Zum ersten Mal übertreffen Maschinen den Menschen in dem, was uns bisher einzigartig machte: unsere Intelligenz.«
Auweia. Taschenrechner »übertreffen« mich auch – trotzdem sind sie keine Mathematiker. KI ist Statistik: Vorhersagen, imitieren, plausibel klingen. Verstehen? Null. Bewusstsein? Null. Verantwortung? Null. Wer hier von »Intelligenz« spricht, zeigt vor allem, wie wenig er davon versteht.
/kuk
Bundesminister für Digitales in seiner Rede: »Zum ersten Mal übertreffen Maschinen den Menschen in dem, was uns bisher einzigartig machte: unsere Intelligenz.«
Auweia. Taschenrechner »übertreffen« mich auch – trotzdem sind sie keine Mathematiker. KI ist Statistik: Vorhersagen, imitieren, plausibel klingen. Verstehen? Null. Bewusstsein? Null. Verantwortung? Null. Wer hier von »Intelligenz« spricht, zeigt vor allem, wie wenig er davon versteht.
/kuk
Ethan Marcotte
boosted
The first is that as a product class, “AI” is a failed technology. I don’t think it’s controversial to suggest that LLMs haven’t measured up to any of the lofty promises made by their vendors ...In other words, we’re some three years into the hype cycle, and LLMs haven’t met any markers of success we’d apply to, well, literally any other technology.
@beep #AI #Hype #Failure
https://ethanmarcotte.com/wrote/against-stocking-frames/
The first is that as a product class, “AI” is a failed technology. I don’t think it’s controversial to suggest that LLMs haven’t measured up to any of the lofty promises made by their vendors ...In other words, we’re some three years into the hype cycle, and LLMs haven’t met any markers of success we’d apply to, well, literally any other technology.
@beep #AI #Hype #Failure
https://ethanmarcotte.com/wrote/against-stocking-frames/
I work for a software and DevOps consultancy firm, which means that I get an assignment to build software for one of my company's customers. (I'm on a US DOD cybersecurity contract).
I'm sitting in the monthly All-Hands meeting, and I had to turn off my camera, because I didn't want everyone to see how much I'm rolling my eyes right now.
Our Chief Digital Officer says that a recent MIT study shows that only 5% of new software projects built on AI actually make it to market. And that's why we have to go all in on AI to fill that market gap where others are failing.
He also started by saying that as we move towards AI-focused work, we're doing so to "distinguish ourselves". Yeah. We're going to "distinguish ourselves" by jumping into the bandwagon that everyone else is jumping into.
🙄 
Joseph Nuthalapati :fbx:
boosted
💡 Ed Zitron's new 16K-word "How to Argue With An AI Booster" is an essential communications toolkit for the next ~18 months. Go read it, and bookmark it.
#AI #ArtificialIntelligence #hype
https://www.wheresyoured.at/how-to-argue-with-an-ai-booster/
💡 Ed Zitron's new 16K-word "How to Argue With An AI Booster" is an essential communications toolkit for the next ~18 months. Go read it, and bookmark it.
#AI #ArtificialIntelligence #hype
https://www.wheresyoured.at/how-to-argue-with-an-ai-booster/
maxlath
boosted
This is one of the best things I read this summer:
"In the Future All Food Will Be Cooked in a Microwave, and if You Can’t Deal With That Then You Need to Get Out of the Kitchen." Thank you for writing it so brilliantly @colincornaby 🙏
This is one of the best things I read this summer:
"In the Future All Food Will Be Cooked in a Microwave, and if You Can’t Deal With That Then You Need to Get Out of the Kitchen." Thank you for writing it so brilliantly @colincornaby 🙏
Robert W. Gehl
boosted
Mark Zuckerberg's vision of 'personal superintelligence' just sounds like supersurveillance. Again.
"Personal devices like glasses that understand our context because they can see what we see, hear what we hear, and interact with us throughout the day will become our primary computing devices."
https://www.meta.com/superintelligence/ #meta#generativeAI #hype
Mark Zuckerberg's vision of 'personal superintelligence' just sounds like supersurveillance. Again.
"Personal devices like glasses that understand our context because they can see what we see, hear what we hear, and interact with us throughout the day will become our primary computing devices."
https://www.meta.com/superintelligence/ #meta#generativeAI #hype
"So here's the catch. If you have the power of geoengineering to turn Mars into Earth, then you have the power of geoengineering to turn Earth back into Earth."
– Neil deGrasse Tyson, astrophysicist and writer
@rolle
Yeah, unfortunately that's a #fallacy and deGrasse Tyson knows better, but chooses hypotheticals like "if you have the power". It's like trying to land on Mars but only managing to skid on the outer atmosphere and never entering it.
Yeah, unfortunately that's a #fallacy and deGrasse Tyson knows better, but chooses hypotheticals like "if you have the power". It's like trying to land on Mars but only managing to skid on the outer atmosphere and never entering it.
The #climateCrisis is inherent to the system that's created it. Expecting that the same system that owes its existence to exploitation of resources and people will now act against its own interests and… save the planet, is just very bad #fiction.
🧐 Wall Street is catching on.
“If you’re going to invest now and get returns in 10 to 15 years, that’s a venture investment, that’s not a public company investment,” Luria said. “For public companies, we expect to get return on investment in much shorter time frames. So that’s causing discomfort, because we’re not seeing the types of applications and revenue from applications that we would need to justify anywhere near these investments right now.”
https://www.cnn.com/2024/08/02/tech/wall-street-asks-big-tech-will-ai-ever-make-money
These bros can't go extinct soon enough
｢ Rosebud was founded in 2023 by Chrys Bader, a Y Combinator alum and co-founder of Secret; and Sean Dadashi, a UC Berkeley Cognitive Science graduate. The pair met at a men’s group, and came up with the idea for Rosebud based on their own experiences with therapy and coaching ｣
https://techcrunch.com/2025/06/04/rosebud-lands-6m-to-scale-its-interactive-ai-journaling-app/
We've Been Conned: The truth about Big LLM•DoltHub Blog
｢ Big LLM is purposefully obfuscating the cost of running LLM's, because if the people actually knew the cost, they'd riot. Well, they wouldn't riot. But they definitely stop believing all the hype surrounding AI and LLMs and especially open-source models ｣
https://dolthub.com/blog/2025-04-09-weve-been-conned-the-truth-about-llms/
