Mailbox looks very solid, although I don't have long-term experience: https://mailbox.org
It provides email, online storage, video conferencing, calendar etc., all of it privacy-preserving by default. You explicitly don't have to provide any personal details.
You are looking at it from a product perspective.
From a scientific perspective, it just means the respective benchmark is meaningless, so we don't know how well such a model generalizes.
Not so! From a scientific perspective the result you can achieve matters, no one is a blank slate.
For humans this is true as well. The way you teach matters. Look at how the bell curve got absolutely demolished for example when math was taught this way:
Another way to look at this is: The first assembly language compiler was handcoded in binary to begin with, and then that compiler's machine code was translated to the more expressive language (assembly). Similar for Fortran/C/etc. from assembly code. Progressively, more expressive languages have been bootstrapped from prior lower-level languages. In a similar way, perhaps a more concise LLM can be built by utilizing a less efficient one?
I am not a biologist, but the first sentence seems bold to me:
> Millions of years of evolution have led mammalian brains to develop the crucial ability to store large amounts of world knowledge and continuously integrate new experiences without losing previous ones.
My impression has always been that humans have been good at selective forgetting, hence keeping relevant memories and dropping others.
Edit: it looks like none of the authors has a biological background either. How serious do they mean the "neurobiologically inspired" claim?
It's weird that's the first sentence of the abstract is different:
> In order to thrive in hostile and ever-changing natural environments, mammalian brains evolved to store large amounts of knowledge about the world and continually integrate new information while avoiding catastrophic forgetting.
I think that's more aligned with selective forgetting
Experts in the field might know more specialized tools, or how to train an actually better Transkribus model without deep technical knowledge required.
I wanted to say this, but could not express it as well.
I think what your points also reveal is the biggest success factor of ChatGPT: it can do many things that specialised tools have been doing (better), but many ChatGPT users had not known about those tools.
I do understand that a mere user of e.g. OCR tooling does not perform a systematic evaluation with the available tools, although it would be the scientific way to decide for one.
For a researcher, however, the lack of knowledge about the tooling ecosystem seems concerning.
> When a user sends an attachment (e.g., an image) on Signal, it is uploaded to cdn2.signal.org.
Why is that even the case? I had understood that (binary) attachments are embedded into the encrypted message and hence transferred directly from sender to receiver.
Obviously, retrieving media from an external location saves bandwidth at multiple positions. I am not a security expert, but it seems almost trivial to see how storing message data on an external server conceptually facilitates attacks like this one.
Isn't that the same reason a link preview is generated at the sender first and then embedded into the message as an image?
I did not read till the end yet, but "woke" is also a very successfully weaponised word for anyone to help push their ideology to further extremes, both left, right, not center. Woke is also a very good detractor from rich and poor discussions.
