1. La distinción crítica entre Inteligencia Artificial y LLM.
Existe una creciente demanda por diferenciar los modelos de lenguaje de la verdadera inteligencia artificial, criticando el uso del término como una estrategia de marketing.
- Reminder - the AI companies of today have no actual intelligence, and no cognitive reasoning abilities. These models do some interesting tricks using a lot of fancy math that can give an illusion of thinking, but there is no actual thought involved. Its LLM, not AI.
- Desperately hoping people start saying algorithm or script instead of AI for the sake of clarity. AI has been so fully co-opted at this point I avoid it if there's even a chance they meant an LLM.
- Technically LLMs are also AI but they are a very specific form of it that should use its own name, especially since most people hate that shit and the intelligence it emulates is a CS college dropout.
2. Capacidades operativas y el problema de las alucinaciones.
Aunque se reconoce su valor en tareas específicas como la programación, persiste la desconfianza debido a la falta de precisión y la necesidad constante de supervisión experta.
- LLMs are really genuinely decent at doing a lot of the tedious grunt work of coding and making it take way less time. Its bad at almost everything else, but that specific thing, it is a miracle machine.
- AI can be a useful tool for people who are already experts but AI researchers are right that LLMs arent getting any better and hallucinate alot still and will always need an expert to check their output.
- Using AI is not hard. Discerning if what it spits out is valuable and accurate is the difficult part. LLMs have huge limitations.
3. Impacto ambiental y la degradación del pensamiento crítico.
Los usuarios expresan una profunda preocupación por el consumo de recursos naturales y el riesgo de que la dependencia de estas herramientas atrofie las habilidades cognitivas humanas.
- LLMs are still computationally intensive, requiring tons of power-hungry hardware - AI datacenters are still destroying lives - hallucinations - other stuff once AI becomes sustainable, ill accept it.
- A primary risk of LLMs being used everywhere is that people will forget how to learn. We are so focused on output that were losing the joy of learning and discovery.
- The majority of AI usage is just using plagiarism bots to atrophy human brains.
4. El futuro incierto de la industria y la burbuja tecnológica.
Se debate intensamente si la tecnología ha alcanzado un techo de desarrollo y si el modelo de negocio actual es una burbuja financiera insostenible.
- The AI industry is collapsing in on itself. LLMs are very dumb, and the AI industry is collapsing in on itself.
- LLMs are a dead end technology that our economy now relies on. Even OpenAI has admitted that AI will always just make shit up, no matter how much data, training, or processing power it has because thats just how LLMs work.
- I really hope this AI bubble bursts but there is no way well be able to step back from the benefits of using LLMs in knowledge work whether wed like to or not.