@arstechnica it doesn’t need work it needs a fundamental rethink of whether the technology makes sense outside of specific research or narrow use cases.
It should never have made it out of research labs or opt-in curiosities for technologists.
None of these details are interesting and almost aren’t even worth reporting on.
This is a stupid, stupid bubble and saying they need to work on parts of it is like saying we’re close but just need refinement which is concretely untrue.
i’m very excited about the interpretability work that #anthropic has been doing with #LLMs.
in this paper, they used classical machine learning algorithms to discover concepts. if a concept like “golden gate bridge” is present in the text, then they discover the associated pattern of neuron activations.
this means that you can monitor LLM responses for concepts and behaviors, like “illicit behavior” or “fart jokes”
this is great work. i’m excited to see where this goes next
i hope #anthropic exposes this via their API. at this point in time, most of the promising interpretability work is only available on open source models that you can run yourself. it would be great to also have them available from #AI vendors
"AI has not created a situation where we need new principles. The established principles remain the crucial ones. But we do need to think through the changing situations anew, in order to figure out how to respond well to a rapidly changing landscape." Says Hallvard Fossheim from @UiB
At PyCon Italia 2024 Ines Montani is presenting her talk "The AI Revolution Will Not Be Monopolized: How open-source beats economies of scale, even for LLMs" 🐍
Although i am very enthusiastic about the functional aspects with regard to Microsoft Recall, it's a potential security nightmare. We should not be fooled with promises about local encrypted data, that can be compromised, also through federated learning, data could still reach microsoft. Also probably part of the compute will be in the cloud. #Microsoft#Recall#cybersecurity#privacy#AI