It’s kind of disheartening to hear supposedly #AI researchers going “Yeah, I know #LLM‘s are trained to predict the next word, but I kinda think may be they are intelligent agents..” You kind of begin to understand why alchemy survived so damn long..Subbarao Kambhampati
A lot of AI folks are not only not impressed by Meta’s Galactica, but are saying it’s downright dangerous:
How MetaAI’s Galactica just jumped the shark by Gary Marcus.
Galactica is what they call a large language model (LLM)…
Fundamental point that ~all people who see LLMs as “AI” seem to be missing: The *only* knowledge that an LLM can truthfully be said to have is information about the distribution of word forms.Emily M. Bender
Why dangerous? Galactica generates text that’s grammatical and feels real. This text will slip into real scientific submissions. It will be realistic but wrong or biased. It will be hard to detect. It will influence how people think.Michael Black
The problem of #galactica is not statistical nonsense or limitations of #LLM. It is a fundamental wrong conception of scientific research. Is not about putting citations on a paper. Is reading, reflecting on what others have done. This is upside-down science. A farce @MetaVirginia Dignum
An EV is only a decarbonizer to the extent that it offsets both gas-powered driving and the emissions needed to make it, a process that leaves a far larger carbon footprint than that for a gas-powered car. The only way for the machine to cover its carbon, so to speak, is in miles.
Amazon CTO publishes the 1998 Distributed Computing Manifesto: “a canonical document from the early days of Amazon that transformed the architecture of Amazon’s ecommerce platform.”