nullbio

13 ms·

361.
by nullbio 3mo ago
Agreed. There needs to be full transparency on the capabilities of the models. Being given access is one thing, but you can't have labs building powerful models that could be manipulating the entire planet without the public knowing. T
362.
by nullbio 3mo ago
This is bad. Where is the transparency? So a small group of technocrats get together behind closed doors and secretly share their AI breakthroughs, and determine whether it's too powerful or not for the plebs in the public. Who is watc
363.
by nullbio 3mo ago
This will accomplish nothing.
364.
by nullbio 3mo ago
It's wild that people will continue to pay Anthropic when there's a better, faster model available for a tenth of the price.
365.
by nullbio 3mo ago
Not FUD, just fact.
366.
by nullbio 3mo ago
Just to be clear, this is not speculation, it is fact.
367.
by nullbio 3mo ago
Playing with fire.
368.
by nullbio 3mo ago
You sound like a cigarette company rep. "I've seen an uptick in vague ominous attacks on Marlboro saying that our products kill people. This must be a coordinated PR effort from <insert boogeyman>." Or perhaps, Anthropi
369.
by nullbio 3mo ago
Yep. Can confirm I do not work for OpenAI. Anthropic does all of this terrible shit and the first instinct of this guy is to assume anyone who calls it out works for OpenAI. Hilarious. Let me uno reverse it, he must work for Anthropic hence
370.
by nullbio 3mo ago
Exactly. They can also feed you a stupider model to goad you into handing over more of this training guidance as well. The incentives are aligned for evil behavior. Open source really needs to win this race, or we need much tighter scrutiny
371.
by nullbio 3mo ago
"be that through web scraping or by purchasing books directly" - or by creating synthetic iterations of the inputs your LLM API service receives.
372.
by nullbio 3mo ago
Anthropic is doing the same thing but at an even larger scale than OpenAI, ironically. OpenAI was just unlucky enough to get caught.
373.
by nullbio 3mo ago
It's exactly what you've said. I'm speaking about Anthropic though, I have no idea whether OpenAI does it. > Anthropic agrees that Customer (a) retains all rights to its Inputs, and (b) owns its Outputs. Anthropic disclaim
374.
by nullbio 3mo ago
No one is going to admit to having an inside source on HackerNews. Read between the lines.
375.
by nullbio 3mo ago
That's not at all how they do it. They wash the data. The end result is they can steal your IP insights without it being explicitly tied to your business. All of the decisions you made to build your product? Those become the standard s
376.
by nullbio 3mo ago
The universe is a giant transformer.
377.
by nullbio 3mo ago
Next up: Lizard nails.
378.
by nullbio 3mo ago
Reddit and X are flooded with them.
379.
by nullbio 3mo ago
> do you think they are just lying Yes. They're constantly lying, and constantly getting caught for it. They have a reputation for it. Why do you think this would be any different? Their standard opt-out agreement frames it as if th
380.
by nullbio 3mo ago
They do all of this, plus far worse. Including but not limited to flooding the internet with guerrilla marketing and sentiment-shifting bots across all social media platforms. This is the sort of company they are, and it's just the tip
381.
by nullbio 3mo ago
This is the key to training better and better models. You collect all of the direction prompts and anti-direction prompts, and you finetune the model to behave that way for the particular domain, under those particular conditions, as a defa
382.
by nullbio 3mo ago
This is a drop in the ocean compared to what Anthropic does behind closed doors.
383.
by nullbio 3mo ago
Obviously they are looking at your IP and code. Anthropic trains on your data regardless of you opting out, I know that one for certain. There's no coincidence they "keep your data temporarily despite opting out" - because th
384.
by nullbio 3mo ago
Data is the goal!
385.
by nullbio 3mo ago
I'm purely talking about the technology - not their business strategy. I actually think their business strategy is blatantly obvious and atrocious.
386.
by nullbio 3mo ago
There's algorithmic ways to accomplish the same thing without requiring monitoring of the brain, if there's enough data to pull from. Semantic analysis + engagement monitoring is more than enough. Peoples emotions expressed via co
387.
by nullbio 3mo ago
I want my model to help me build up its own infrastructure that instills it with the sort of constraints I want for my project, rather than have it behave generically and automatically for everything. It should follow instructions incredibl
388.
by nullbio 3mo ago
Agreed that they shouldn't be writing vulnerable code to begin with. You'd think the models would be trained to know when they are working on something that has security implications, and to validate the security of what they&#x27
389.
by nullbio 3mo ago
This is already happening at scale by the social media feed algorithms. We don't need generated content to accomplish this. In a sea of user created content, plenty of it is already at peak activation.
390.
by nullbio 3mo ago
Someone has to know. Would be nice if an insider would drop some hints so that the open-source space could make some good progress.

More ›