5 ms·

If bard will provide at least somewhat useful suggestions to quirks of particular language (I'm looking at you, c++) and/or there will be whitelisted or mirrore

by galkk 3y ago

If bard will provide at least somewhat useful suggestions to quirks of particular language (I'm looking at you, c++) and/or there will be whitelisted or mirrored sites like cppreference/godbolt, then I suspect I won't need actual big internet for work at all. Most of the reference documentation is actually internal project/other google sources and internal docs/guides/design docs.

weinzierl 3y ago
I find that idea fascinating but I doubt it will fly. Todays LLMs are no suitable replacement for documentation, in my experience, because their knowledge is so sparse. You will not notice it immediately because they fill the gaps with plausible nonsense. Also training a model with domain specific knowledge is not a realistic option for most of us as of today. For having all the reference documentation locally (possibly indexed in a vector database and accessible to the LLM) I'm doubtful as well, since it is so hard to determine scope beforehand. A couple of years ago I tried to program off-grid and prepared a MacBook with Dash (an OSX offline doc reader) and all the reference docs I thought I need. It was a nightmare, and that is from a dude who learned programming before the Internet, based on offline docs solely.
- vbezhenar 3y ago
  I worked in offline environment for a few years. It was not a nightmare. It's less productive, that's for sure, but not by far margin. And it compensates by lack of online distractions.
- wakeupcall 3y ago
  Working offline is perfectly doable. I've grown not to rely on connectivity when I was living in a rural area with spotty/slow 3g connectivity at best. Not relying on network connectivity in my entire pipeline is still today a big discriminator in the stack I choose. The biggest downside was, and still is, documentation. I had a script that scavenged the installed packages and downloaded all relevant documentation packages automatically, along with a local indexing service (I'm still using and recommend "recoll"). There is a humongous gap between libc/posix (anything instantly accessible with "man" with stunning quality) and pretty much everything else. Even python, which has a decent "online" (built-in) reference, is not as convenient and easily searchable when looking at the html manual itself. A local search index doesn't have the inference power of something like google to point you to the right example you're looking for something in cppreference. And once you pass the top 10 best documented packages you have on system, you realize how downright bad everything else is documented and how much you rely on the search engine to fill the gaps from unofficial sources. For me at least, documentation was always THE problem.
LoveMortuus 3y ago
I work in customer support and while I haven't yet used Bard (because at the time it wasn't available to me), I have used ChatGPT to tell me what the customer is actually asking of me, when I didn't understand their emails. And thus far, it's been quite helpful. I even asked if once to rewrite the greeting part of the email so that it included now empathy (I'm kind of emotionally inhibited, as my doctor put it) and judging based on customer response, they were quite pleased.
- Philip-J-Fry 3y ago
  How are people getting things like this past management? I work in a big company and it is stressed to us that the last thing we ever want to do is paste proprietary code, customer info, internal documents, etc. into a cloud LLM. Especially since there's every chance that the prompts could become part of the future training set.
  - withinboredom 3y ago
    They probably don't even know, and why would they care? OP never said they were pasting personal information or proprietary code.
    - arghnoname 3y ago
      At my job, any information we get from customers is treated as private/personal. It would be a fireable offense to load it up into some third party database or service. I have not, will not, and am not allowed to copy and paste anything into chatGPT from my company that I wouldn't be comfortable putting on twitter. I agree with this policy.
    - Philip-J-Fry 3y ago
      A support ticket or live chat with a customer is customer data. With things like GDPR you need to know exactly where customer data is stored and for how long. Once you're pasting this in cloud LLMs that data is out of your control and you're breaching GDPR. The last thing someone expects is that when they're replying to a support ticket is Google or OpenAI being fed that info and it's being put into a training dataset. We all value our privacy but it seems accepted to breach other people's privacy for the sake of using an LLM to make your job easier.
mike_hock 3y ago
cppreference can be downloaded and used locally. godbolt would be a weird choice to whitelist if you go through the trouble of airgapping in the first place. All it takes is one accidental copy&paste of a sensitive code snippet and it's there for the world to see with no undo.

by galkk 3y ago

weinzierl 3y ago

I find that idea fascinating but I doubt it will fly. Todays LLMs are no suitable replacement for documentation, in my experience, because their knowledge is so sparse. You will not notice it immediately because they fill the gaps with plausible nonsense. Also training a model with domain specific knowledge is not a realistic option for most of us as of today. For having all the reference documentation locally (possibly indexed in a vector database and accessible to the LLM) I'm doubtful as well, since it is so hard to determine scope beforehand. A couple of years ago I tried to program off-grid and prepared a MacBook with Dash (an OSX offline doc reader) and all the reference docs I thought I need. It was a nightmare, and that is from a dude who learned programming before the Internet, based on offline docs solely.

vbezhenar 3y ago
I worked in offline environment for a few years. It was not a nightmare. It's less productive, that's for sure, but not by far margin. And it compensates by lack of online distractions.
wakeupcall 3y ago
Working offline is perfectly doable. I've grown not to rely on connectivity when I was living in a rural area with spotty/slow 3g connectivity at best. Not relying on network connectivity in my entire pipeline is still today a big discriminator in the stack I choose. The biggest downside was, and still is, documentation. I had a script that scavenged the installed packages and downloaded all relevant documentation packages automatically, along with a local indexing service (I'm still using and recommend "recoll"). There is a humongous gap between libc/posix (anything instantly accessible with "man" with stunning quality) and pretty much everything else. Even python, which has a decent "online" (built-in) reference, is not as convenient and easily searchable when looking at the html manual itself. A local search index doesn't have the inference power of something like google to point you to the right example you're looking for something in cppreference. And once you pass the top 10 best documented packages you have on system, you realize how downright bad everything else is documented and how much you rely on the search engine to fill the gaps from unofficial sources. For me at least, documentation was always THE problem.

LoveMortuus 3y ago

I work in customer support and while I haven't yet used Bard (because at the time it wasn't available to me), I have used ChatGPT to tell me what the customer is actually asking of me, when I didn't understand their emails. And thus far, it's been quite helpful. I even asked if once to rewrite the greeting part of the email so that it included now empathy (I'm kind of emotionally inhibited, as my doctor put it) and judging based on customer response, they were quite pleased.

Philip-J-Fry 3y ago
How are people getting things like this past management? I work in a big company and it is stressed to us that the last thing we ever want to do is paste proprietary code, customer info, internal documents, etc. into a cloud LLM. Especially since there's every chance that the prompts could become part of the future training set.
- withinboredom 3y ago
  They probably don't even know, and why would they care? OP never said they were pasting personal information or proprietary code.
  - arghnoname 3y ago
    At my job, any information we get from customers is treated as private/personal. It would be a fireable offense to load it up into some third party database or service. I have not, will not, and am not allowed to copy and paste anything into chatGPT from my company that I wouldn't be comfortable putting on twitter. I agree with this policy.
  - Philip-J-Fry 3y ago
    A support ticket or live chat with a customer is customer data. With things like GDPR you need to know exactly where customer data is stored and for how long. Once you're pasting this in cloud LLMs that data is out of your control and you're breaching GDPR. The last thing someone expects is that when they're replying to a support ticket is Google or OpenAI being fed that info and it's being put into a training dataset. We all value our privacy but it seems accepted to breach other people's privacy for the sake of using an LLM to make your job easier.

mike_hock 3y ago

cppreference can be downloaded and used locally. godbolt would be a weird choice to whitelist if you go through the trouble of airgapping in the first place. All it takes is one accidental copy&paste of a sensitive code snippet and it's there for the world to see with no undo.