HN user
Not sure if it’s using the same FSD decision matrix but my model S chimed at me to drive into the intersection while sitting at a red light Last night with absolutely zero possibility it saw a green light anywhere in the intersection.
Perfectly isn’t a descriptor I would use. But this is just anecdotal.
Having handed over control of my vehicles to FSD many times, I’ve yet to come away from the experience feeling that my vehicle was operating in a safer regime for the general public than within my own control.
Something like 400 people just died because of a claim about nuclear weapons which is not backed up by evidence. Claims that have been echoed for decades…
Purportedly 400 people just died… Was it because a sovereign country wants to have Nuclear power? Maybe? Maybe not? Was it because Israel already has Nukes? Who knows… But it’s not a simple end of story situation unless lives have no value.
Model Control Protocol is not Model Context Protocol. We might be suffering from too many acronyms. (Just to be clearer, I think you’re referencing on the wrong MCP)
Where positive impact is purely defined as a reduction in investment on public initiatives and ignoring the much more complicated returns those investments may have produced both fiscally and socially.
Amazing project.
I’m sure the 30k is below cost assuming Tesla makes some revenue on the taxi service over the lifetime of the vehicle.
The vehicle decided to go into oncoming traffic to clear a light change?! What type of an excuse is that. The vehicle should not have been in an intersection it is unable to clear.
As much as the headline leads to satirization, I think this is a good thing. I’d much rather have someone who’s leading the forefront of a technical innovation on the board than not. At least they can participate and inform the board what is being done and what isn’t being done. What can be changed and what we don’t yet have the tools to do.
A board that has Sam Altman on it is going to be much better able to craft safety guidance for the use and misuse of AI than a board that doesn’t have access to such a domain expert.
Just, you know, don’t give him veto power.
There’s a sentiment which isn’t really being expressed here so far.
I don’t begrudge people who are less fortunate and need help.
I don’t enjoy my food any less because someone who is hungry can get free food from a food bank.
There are so many ways that the government has helped the more successful. I for one was a beneficiary of PPP loan forgiveness (not by my choice but an election of my managing partner).
This attempts to balance the scales of what type of support people get who are going through different situations and at different times with different needs isn’t particularly productive. It might feel good in some self centered bitter way, but it doesn’t speak to whether this is a policy measure that has a positive effect on many people and is a net benefit to the economy as a whole.
“OpenTofu Project was recently made aware of a letter by HashiCorp’s lawyers, alleging that OpenTofu was not respecting the terms of its BSL license governing its Terraform codebase.
OpenTofu vehemently disagrees with any suggestion that it misappropriated, mis-sourced, or otherwise misused HashiCorp’s BSL code.
Indeed, it seems that HashiCorp may be conflating code that it had previously been open-sourced under the MPL and more recently developed code it published under the BSL.
OpenTofu’s maintainers have investigated this matter, and intends to issue a written response providing a more detailed explanation of its position in the coming days.”
Just leaving this here without making simplistic claims of what constitutes a fact.
Having worked at Facebook, albeit a long time ago. It’s far more likely that a conflict of events caused a false positive of an Automated spam system than Facebook giving a crap about privacy campaigns that hurt revenue.
But alas, we humans love our conspiracy theories because they tell a more interesting story.
I’m running and finishing missions daily. There are bugs, yes, but it is imminently playable.
35mm film is equivalent to about 21MP.
I would think it has to do with the IP owners using the likeness of the actors to continue to market the show year after year.
All knowledge is lossy compression.
My thinking “the enemy gate is down” when considering the tokens “Ender’s Game” is my recalling a learned association of those tokens to the given token string.
My knowing that doesn’t strip the copyright. My telling someone the meaning and context of the phrase generally doesn’t strip the copyright away from Orson Scott Card. I’m not reproducing his work but my knowledge of it. And it’s dependent on what I do with that knowledge and how if I’ve violated his copyright.
We are prosecuting the LLMs for possessing fragments of knowledge. And we’re assuming that the recall of some of those fragments means a copy of that work is in fact contained within the weights.
I believe LLMs should be allowed to read/view/consume content and learn from it even if that content has a copyright.
We phrase it like somehow the material is being copied into the LLM, but that’s not what it’s doing. It’s building a neural graph from the experience of consuming that content.
What would the world be like if humans couldn’t learn, train the weights of the interconnects of their neural tissue, from any material with a copyright?
I’ve been trying to install git on the hypervisor of my smartOS install intermittently for a year. I run into system package conflicts or key signing errors and then I give up and switch to a zone that isn’t so borked with package issues.
Then I go lament in the corner that I really should be serious about migrating this to 200TB home lab to anything else and then realize that the 200TB is not something easily backed up in the case FreeBSD or TrueNAS doesn’t understand my Zpool versions and all of those zones I created will need to be rebuilt as jails…
And my lamentations turn to tears. All because I wanted to just checkout some random project.
Except that generative AI is actually generative. It isn’t recreating a work by shape and line and form. It is using the knowledge gained from studying many works to determine when given a random pixel, what color pixel should it and its neighbors transform into to best fit the requested parameters.
Stable Diffusion starts with random noise and attempts to denoise that random information into a coherent picture. That it has in the past learned that certain types of pictures should have say something resembling a watermark is just it being incredibly dumb but ultimately no different than it learning that horses should have something resembling manes or people’s hands having something resembling fingers.
And the other 30% is unknown so we can’t say they selectively comply. It’s quite possible the other 30% are invalid warrants, in which case there’s nothing to comply with.
I’m biased because I used to work for Facebook a long time ago. So take my words with a grain of salt.
But Facebook research really wants to build amazing tech to power new experiences. The company has attracted many smart and amazing developers who want to pioneer innovation and want to see their work thrive in the broader tech community.
I’m sure this will have the same issues as the other Large Language Models and there’s a lot of work to be done to figure out how to distill a reliably useful system from something trained on so much human language which is unreliable.
I’m heartened to see them talk about needing to research how to remove bias, toxicity, disinformation and hallucinations.
This is something we need to focus on for all LLMs so I’m happy to see more focus on that in the community.
I gave it a picture of the Mahi Mahi and jambalaya I was having in a restaurant. It showed me pictures of food. Fair enough. But it’s completely disingenuous to say it’s finding attribution to any images you give it.
Finding similar pictures in an embedding space does not mean any of those pictures are part of the attribution chain any more than anything else.
I think it’s misleading to say Tesla faked the driving.
They were running one off software, purpose trained for the route, showing the golden run that was successful.. yes. But it was computer vision and self driving that they demonstrated. It just wasn’t their production system.
1) it makes the engineers happy to be able to speak about the interesting projects they are working on and give back to the wider community.
2) it serves as a venue for attracting other talented candidates who are likewise minded on working on technical problems.
3) when I was employed of FB it was a relatively flat hierarchy, which is to say there weren’t that many higher ups to convince that this should be done.
This thread was a delightful read.
Just NOT in twitter. I gave up on twitter and signed out of it years ago and refuse to sign back in.
I spent a good hour of my life looking for ways to read this thread. I personally know Yishan and value the opinions he cares to share so I new this would be interesting if I could just manage to read it.
Replacing the url to nitter.net helped but honestly it was most cohesive in threadreaderapp although it missed some of the referenced sidebar discussions (like the appeal to Elon to not waste his mental energy on things that aren’t real atom problems).
That you’ve never felt this way does not preclude that others may.
Texts do not convey emotion. Allow verification of nuanced points with simple interjections and interrogatives. The limitation of threads of conversation to simplify discourse through a more direct dialog and rebuttals.
I have had relationships due specifically because texts are really incomplete and poor means of conversation and the other side fails to pick up and acknowledge the severity of the situations I was trying to describe.
That is entirely untrue. So many student films, home movies, YouTube posts get shown before human review.
You seem to be wildly underestimating the complexity of the task and the long tail of content creation. There are 1.9 billion daily users.
You’re asking to REQUIRE a human content moderation system that reviews 100% of the continent that 1.9 billion users can create a day? Before that content can get shown? I don’t see a feasible way of creating a system like that which will serve the needs of those 1.9 billion users in a way that will seem remotely tolerable to them, that can timely share their content.
Wow… someone just nuked this right off of the front page.
As a black developer... this all just makes me a little sad.
Terms like master/slave or whitelist/blacklist are common parlance and this attempt to remove these phrases for less well understood phrases just seems like an attempt to whitewash history for little to no gain.
I'm a firm believer that language is dynamic and the meaning of words is the meanings we currently associate to them.