611–620 of 740 posts
This isn't the reason I'm considering leaving Anthropic. I don't think I can tolerate its writing style anymore. Reading Claude output is starting to cause actual psychological harm. I have tried many ways to get it to stop writing in its stupid punchy linked-in marketing-team voice, and I can't. Is there a model out there that sounds sound this awful? It's like rubbing sand into the folds of my brain.
This isn't the reason I'm considering leaving Anthropic. I don't think I can tolerate its writing style anymore. Reading Claude output is starting to cause actual psychological harm. I have tried many ways to get it to stop writing in its stupid punchy linked-in marketing-team voice, and I can't. Is there a model out there that sounds sound this awful? It's like rubbing sand into the folds of my brain.
Earlier quoted context omitted.
Yep translation is going to be a free feature on all devices using on-device LLMs. I see no revenue flowing to AI companies.
Commercial grade translation between dozens of languages would take enormous amounts of disk space. But we'll see. Text editing tools are included on all digital devices, yet companies pay for commercial solutions like MS Office. Cameras and basic image editing tools are included on all smart phones, yet there are millions of advertising agencies around the world. And so on.
eg. processors are needed for everything on the planet. No Intel isn't going to generate annual revenue in the trillions because of that. In the case of frontier companies the moat is even smaller with dozens of players competing. LLMs will become a low-cost commodity. You meanwhile can make billions building applications using them though.
Non-LLM user here. Why? Apart from the ecological issues, I'm very uncomfortable giving my organisation's crown jewels to {random_internet__corp}. Look at the lengths they go to for training data - 10M for Spirit's call logs? Destroying millions of obscure books to scan them? They make meth-heads look scrupulous. Your prompts, especially if they contain your entire codebase, are _way_ more data-rich than an airline p…
Curious what you think your organization's crown jewels actually are? Because, almost certainly, it's not your source code.
Markets been waiting for something to sell off or correct in the short term over. Looks like it may have found its justification.
At this point, Fable is really just an experimental model (as it should be). It can do very useful things, but is it a broadly good general purpose model? Definitely no. Most users don't have problems hard enough for Fable outside of coding very large and complicated projects. Most users don't have 45 minutes to accomplish a task Sonnet can do well enough in five. There's not a PowerPoint in the world where Fable is the right tool to build it.
The secret sauce is going to be in letting Sonnet decide to delegate to Opus and Fable when they're the right tools for the job. But you can't train a model to do that until the bigger/better models exist and you can observe how your users actually take advantage of them. I'd bet money that's exactly what the rlhf going on at Anthropic looks like right now.
Economically, it makes sense. Sure, on paper you want users burning as many tokens as you can. But pushing users into burning tokens and taking a long time and getting a meh result is far worse then giving them the "fast and good enough" solution that occasionally burns more tokens automatically when the problem requires it, and getting a higher quality result out when you do. From an infrastructure capacity perspective, this is the dream: you stop measuring cost [for Anthropic] per token and measure cost per outcome, allowing you to use less hardware to accomplish the same abstract units of work.
Earlier quoted context omitted.
OP made the "electricity" analogy, and that's really all people want. I want to plug something into the wall and have it work. I don't want to have to worry that my electric company is going to rug-pull me because I plugged the wrong appliance in, or I didn't agree to some TOS, or I used the electricity to run grow lights for my pot farm, or this or that or the other.
The analogy is apt because I feel this is exactly what keeps Anthropic and OpenAI's owners up at night -- becoming the utility company the People want them to be. Ironically their behavior may accelerate their fears. And yes, the so-called safety features are ridiculously invasive and the worst is agreeing to have surveillance cameras installed in every room that occasionally detect any attempt to grow plants with LE…
Earlier quoted context omitted.
How many big tech companies let you talk to a human to get support. Automation is wonderful to cut cost for them but for the users being unable to get support is a horrible experience. But you cannot go elsewhere because they are the only player in town. How can small companies with 1000x less money able to provide live support, but if you pay 20, 100, 200 dollars for a subscription you dont have a phone number to ca…
>Automation is wonderful to cut cost for them but for the users being unable to get support is a horrible experience. But you cannot go elsewhere because they are the only player in town. I can't tell you how many times I've experienced this with comcast. The last time I had to deal with it, was when I bought a new cable modem. I call in to provision it, the automated system assumes I have one of their modems and fai…