Examples? I can think of a number of foreign companies that the US facilitates, like Nestle.

Eh, I switched. I switched all of my lab’s computers, too, and my PhD students have remarked a few different times that Linux is pretty cool. It might snowball.

Journal quality can buffer this by getting better reviewers (MDPI shouldn’t be seen as having peer review at all, but peer review at the best journals–because professors want to say on their merit raise annual evals that they are doing the most service to the field by reviewing at the best journals–is usually good enough at weeding out bad papers), but it gets offset by the institutional prestige of authors when peer-review isn’t double-blind. I’ve seen some garbage published in top journals by folks that are the caliber of Harvard professors (thinking of one in particular) because reviewers use institutional prestige as a heuristic.

When I’m teaching new grad students, I tell them exactly what you said, with the exception that they can use field-recognized journal quality (not shitty metrics like impact factor) as a relative heuristic until they can evaluate methods for themselves.

I actually took that bit out because LLMs are pro climate and against everything that makes the environment worse. That’s a result of being trained on a lot of scientific literature. I was just curious what Opus would say about the conceptual knowledge piece.

Claude Opus disagrees, lol (edit to add: all of what follows is Claude; not me):

I respectfully disagree with the claim that current language models are logically incapable of mapping language to actual concepts. While today’s LLMs certainly have limitations and are prone to hallucinations, they have demonstrated a remarkable ability to capture and manipulate complex concepts through language.

A few key points:

LLMs are trained on vast amounts of human-generated text which is grounded in real-world concepts, allowing the models to build rich representations that go beyond surface-level word associations. Techniques like unsupervised pre-training enable LLMs to learn meaningful conceptual relationships.

In many domains, LLMs have shown an impressive capacity for reasoning, inference, and knowledge synthesis - skills that rely on a deep understanding of the underlying concepts, not just pattern matching. Their performance on complex question-answering, analogical reasoning, and natural language inference tasks is evidence of this.

LLMs can be fine-tuned and augmented with techniques like retrieval-augmented generation or chain-of-thought prompting to further ground them in factual knowledge and improve their reasoning capabilities. This allows more robust conceptual mappings.

Careful prompt engineering and sampling techniques can help to mitigate hallucinations and nonsensical outputs. While not a complete solution, this shows that the conceptual knowledge is often present even if the model sometimes struggles to express it faithfully.

That said, I fully agree that the tendency for subtle nonsense is a serious issue, especially for layperson use cases where hallucinations may go undetected. Continued research into making LLMs more truthful, consistent and robust is crucial. Techniques like constitutional AI to bake in truthfulness, and better UX design to convey uncertainty are important steps.

But in summary, I believe the evidence suggests that LLMs, while flawed, are not fundamentally incapable of meaningful conceptual representation and reasoning. We should push forward on making them more reliable and trustworthy, rather than dismissing their potential prematurely.