291–300 of 551 posts
Does this include any code base you are running Claude Code with (where parts of code are sent as part of the context)? I'm not hugely clear on how my private codebase is exposed to Claude in the first place when using Claude Code.
Earlier quoted context omitted.
Personally, I'm able to write code I wasn't able to before, like functions heavy with math. For game development, this has been super helpful, when I know basically what inputs I have, and what output I need, but I'm not able to figure out how the actual function implementation should be. Add a bunch of unit tests, let the LLM figure out the math, and I can move on to more important features. For me this been a prett…
Sure, I’m more productive with it in certain aspects of my work as well. Does that make it a net positive for humanity? From the energy consumption impact on climate change alone I would say the answer is clearly no. And that’s before we even talk about the impact on the next generation’s job opportunities. And tons of other issues like how Big Tech is behaving.
Everyone is so cynical these days.
Earlier quoted context omitted.
Under the FAQ: > Previous chats with no additional activity will not be used for model training.
that "with no additional activity" seems backdoor-ish though. If I said some things in a chat expecting privacy per the agreement, then they change the agreement, does that mean they can collect my data from that chat going forward or does it mean they can collect it retroactively?
Claude assists me in my math research. The scenario that concerns me is that Claude learns unpublished research ideas from me as we chat and code. Claude then suggests these same ideas to someone else, who legitimately believes this is now their work. Clearly commercial accounts use AI to assist in developing intellectual product, and privacy is mandatory. The same can apply to individuals.
> Claude assists me in my math research. > Claude then suggests these same ideas to someone else, who legitimately believes this is now their work. Won't this mean that claude assisted you with someone else work? Sure it's not from a "chat" but claude doesn't really know anything other than it's training data
I've seen cases where Claude demonstrates novel behaviors or combines existing concepts in new ways based on my input. I don't think it's as simple as memorization anymore.
Earlier quoted context omitted.
From Anthropic communication: > If you’re an existing user, you have until September 28, 2025 to accept the updated Consumer Terms and make your decision. If you choose to accept the new policies now, they will go into effect immediately. These updates will apply only to new or resumed chats and coding sessions. After September 28, you’ll need to make your selection on the model training setting in order to continue…
Under the FAQ: > Previous chats with no additional activity will not be used for model training.
In my opinion, training models on user data without their real consent (real consent = e.g. the user must sign a contract or so, so he's definitely aware), should be considered a serious criminal offense.
And when talking specifically about AI, one could argue that learning from interactions is a common aspect of intelligence, so a casual user who do not understand details about LLMs would expect so anyways. Also, the fact that LLMs (and other neural networks) have distinct training and inference phases seems more like an implementation detail.
Earlier quoted context omitted.
I am pretty sure they try to do it all the time between themselves. Most of the real sauce in AI coding comes from reinforcement learning, usually done by armies of third world outsourced developers tediously doing all kinds of tasks with instructions to detail their reasoning behind each chance. Things like: "to run this python test in a docker container with the python image we need to install the python package xy…
> those developers are not well paid (usually around 30/40 USD hour, no benefits) $40/hour for a full time would put you just over the median household income for the US. I suspect this provides quite a good living for their family and the devs doing the work feel like they’re well-paid.
For comparison, I live in a place that is typically considered as tier 3 or 4 out of 4 in the US by employers (4 being the cheapest). Costs of living are honestly more like tier 2 cities, but it’s a small city in a poor state. 7 years ago, the going rate for an unlicensed handyman was $32/hour, often paid under the table in cash (I don’t have more recent numbers because I find DIY better and easier than hiring someone reliable).