> Testers preferred o3-mini's responses to o1-mini 56% of the time
I hope by this they don't mean me, when I'm asked 'which of these two responses do you prefer'.
They're both 2,000 words, and I asked a question because I have something to do. I'm not reading them both; I'm usually just selecting the one that answered first.
That prompt is pointless. Perhaps as evidenced by the essentially 50% response rate: it's a coin-flip.
> A Redding Police Department officer in 2021 was charged with six misdemeanors after being accused of accessing CLETS to set up a traffic stop for his fiancée's ex-husband, resulting in the man's car being towed and impounded, the local outlet A News Cafe reported. Court records show the officer was fired, but he was ultimately acquitted by a jury in the criminal case. He now works for a different police department 30 miles away.
When people say "I don't see why privacy matters, I have nothing to hide" this is the exact kind of edge-case I always think about.
The problem with a lot of these massive surveillance systems is that a lot of people end up with access to them, and some of those people may not be trustworthy.
In this case, your ex-wife gets engaged to a cop and now they're abusing their access to databases to cause you harm.
(That note that "he now works for a different police department 30 miles away" is such a toxic aspect of American policing: cops who get fired for stuff like this inevitably end up in the exact same job somewhere else.)
I think there's two different things going on here:
"DeepSeek trained on our outputs and that's not fair because those outputs are ours, and you shouldn't take other peoples' data!" This is obviously extremely silly, because that's exactly how OpenAI got all of its training data in the first place - by scraping other peoples' data off the internet.
"DeepSeek trained on our outputs, and so their claims of replicating o1-level performance from scratch are not really true" This is at least plausibly a valid claim. The DeepSeek R1 paper shows that distillation is really powerful (e.g. they show Llama models get a huge boost by finetuning on R1 outputs), and if it were the case that DeepSeek were using a bunch of o1 outputs to train their model, that would legitimately cast doubt on the narrative of training efficiency. But that's a separate question from whether it's somehow unethical to use OpenAI's data the same way OpenAI uses everyone else's data.
I work on the Near Earth Object Surveyor space telescope (NEO Surveyor) writing simulation code which predicts which objects we will see. This one has drummed up a bit of interest due to its (relatively) high chance of impact. I actually spent quite a bit of time yesterday digging through archive images trying trying to see if it was spotted on some previous times it came by the Earth (no luck unfortunately). Since we saw it so briefly, our knowledge of its orbit is not that great, and running the clock back to 2016 for example ended up with a large chunk of sky where it could have been, and it is quite small. We will almost certainly see it again with NEO Surveyor years before its 2032 close encounter. I have not run a simulation for it, but I would not be surprised if LSST (a large ground telescope survey which is currently coming online) to catch it around the same time NEO Surveyor does.
Our knowledge of the diameter of this object is a bit fuzzy, because of surface reflectivity, small shiny things can appear as bright as dark large things. This is one of the motivations of making the NEO Surveyor an IR telescope, since IR we see black body emission off of the objects, which is mostly size dependent, and only weakly albedo dependent.
There is an even tinier chance that if it misses the Earth in 2032, it could hit the moon. I haven't run the numbers precisely for that one, but it impacted a few times in some monte-carlo simulations.
If anyone is interested in orbit dynamics, I have open sourced some of the engine we are using for observation predictions: https://github.com/Caltech-IPAC/kete
It is relatively high precision, though JPL Horizons has more accurate gravitational models and is far better suited for impact studies. My code is primarily for predicting when objects will be seen in telescopes.
The schadenfreude is very real right now. I have difficulty putting to words my level of antipathy towards Altman, and I hope to watch gleefully as this all blows up in his smug face.
This is absolutely hilarious! :)
ClosedAI scraped human content without asking and they explained why this was acceptable... but when the outputs of their training corpus is scraped, it is THEIR dataset and this is NOT acceptable!
Oh, the irony! :D
I shared a few screenshots of DeepSeek answering using ChatGPT's output in yesterday's article!
https://semking.com/deepseek-china-ai-model-breakthrough-sec...
What I find the most comical about this is that the whole situation could be loosely summarized as "OpenAI is losing its job to AI."
Sublime Text developer here, thank you for all the praise! I'm looking forward to what we can accomplish this year. If you have any questions I'd be happy to answer.
Oh I see, so training on copyrighted content is fine unless it's your AI model...
I'm taking this opportunity to Ask HN: what do you think of the new Bitwarden browser extension?
Sure it looks more modern and a few things are better.
But personaly I HATE the new "copy" button.
With the old version there was a button for each field : one to copy the login, one to copy the password, one to copy the TOTP.
Now there's just a single button that will display a list of options to choose from depending on what you want to copy.
So instead of copying a field with one click, now I need to do one click, go on the right option, and another click.
Even worse: if the account contains only one field, the copy button will still display the list of options, with just one option.
How could nobody think that when the user want to copy something from a list, and this list contains only one item, the right thing to do is to copy this single thing, not ask them what they want to copy...
Here’s a take I haven’t seen yet:
If training and inference just got 40x more efficient, but OpenAI and co. still have the same compute resources, once they’ve baked in all the DeepSeek improvements, we’re about to find out very quickly whether 40x the compute delivers 40x the performance / output quality, or if output quality has ceased to be compute-bound.
90% of the comments in this thread make it clear that knowing about technology does not in any way qualify someone to think correctly about markets and equity valuations.
People passing cynical comments at Google need to understand that at a big co like Google, something like this doesn't 'just happen'. It probably happened because some passionate L6/L7 engineer wanted to do it and pushed through the bureaucracy to get approvals for it, probably largely on their own time (by which I mean that this was at best a side-project for them and at worst a distraction that was losing them favor with their bosses). At every point in the process, they probably had to justify what they were doing to their leads, to lawyers, to privacy reviewers, who had no real stake in it and so had nothing to lose by saying No. They almost certainly won't receive any career progress out of this and would risk a setback if something slips through the cracks (such as some unredacted proprietary information).
They did it because they felt it was the right thing to do. Good things happen through the actions of individuals like this. We should acknowledge and celebrate it when they do, anti-big-tech cynicism can wait.
This interview with DeepSeek founder and CEO Liang Wenfeng, also co-founder of the hedge fund backing DeepSeek, might shed some light on the question: https://www.chinatalk.media/p/deepseek-ceo-interview-with-ch...
Some relevant excerpts:
“Because we believe the most important thing now is to participate in the global innovation wave. For many years, Chinese companies are used to others doing technological innovation, while we focused on application monetization — but this isn’t inevitable. In this wave, our starting point is not to take advantage of the opportunity to make a quick profit, but rather to reach the technical frontier and drive the development of the entire ecosystem.”
“We believe that as the economy develops, China should gradually become a contributor instead of freeriding. In the past 30+ years of the IT wave, we basically didn’t participate in real technological innovation. We’re used to Moore’s Law falling out of the sky, lying at home waiting 18 months for better hardware and software to emerge. That’s how the Scaling Law is being treated.
“But in fact, this is something that has been created through the tireless efforts of generations of Western-led tech communities. It’s just because we weren’t previously involved in this process that we’ve ignored its existence.”
“We do not have financing plans in the short term. Money has never been the problem for us; bans on shipments of advanced chips are the problem.”
“In the face of disruptive technologies, moats created by closed source are temporary. Even OpenAI’s closed source approach can’t prevent others from catching up. So we anchor our value in our team — our colleagues grow through this process, accumulate know-how, and form an organization and culture capable of innovation. That’s our moat.
“Open source, publishing papers, in fact, do not cost us anything. For technical talent, having others follow your innovation gives a great sense of accomplishment. In fact, open source is more of a cultural behavior than a commercial one, and contributing to it earns us respect. There is also a cultural attraction for a company to do this.”
I love the animation when you click "No" on "Do you want a new Pebble?". So extra.
Thank you, Google. You didn't have to, but you did. We are all are extraordinarily grateful.
I wrote a blog post about our plans to bring Pebble back, sustainably. https://ericmigi.com/blog/why-were-bringing-pebble-back
We got our original start on HN (https://news.ycombinator.com/item?id=3827868), it's a pleasure to be back.
If you're interested in getting a new Pebble, check out https://rePebble.com
fyi Yann LeCun, Chief AI Scientist at Meta, said:
“To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones.’ DeepSeek has profited from open research and open source (e.g., PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”
[1] https://www.forbes.com/sites/luisromero/2025/01/27/chatgpt-d...
I often like to sum it up as: "Thinking gangs observed in prison represent human households."
This goes along with an imagined mockumentary, narrated by a David Attenborough figure, ex: "In the living room, a confrontation is brewing. The Big Boss has informed the bottom-tier Henchmen Homies that it is time for lights-out, but they are not ready to leave the television and they resent the privileges of the older Made Man, who is permitted to remain. This is a problem for the Big Boss, since directly punishing them might undermine his claim to control the group as a whole. Fortunately for him, his Enforcer is arriving from the kitchen, attracted by the noise..."
It’s worthwhile to “go above and beyond” for individuals who will help you, who may exist in a company… but never for the company itself. A company is no less and no more than a pile of someone else’s money that will do literally anything, including destroy your life, to become a bigger pile.
You should do a good job for individuals who will repay you later on. Companies themselves these days can sod off—they stand for nothing.
The description of DeepSeek reminds me of my experience in networking in the late 80s - early 90s.
Back then a really big motivator for Asynchronous Transfer Mode (ATM) and fiber-to-the-home was the promise of video on demand, which was a huge market in comparison to the Internet of the day. Just about all the work in this area ignored the potential of advanced video coding algorithms, and assumed that broadcast TV-quality video would require about 50x more bandwidth than today's SD Netflix videos, and 6x more than 4K.
What made video on the Internet possible wasn't a faster Internet, although the 10-20x increase every decade certainly helped - it was smarter algorithms that used orders of magnitude less bandwidth. In the case of AI, GPUs keep getting faster, but it's going to take a hell of a long time to achieve a 10x improvement in performance per cm^2 of silicon. Vastly improved training/inference algorithms may or may not be possible (DeepSeek seems to indicate the answer is "may") but there's no physical limit preventing them from being discovered, and the disruption when someone invents a new algorithm can be nearly immediate.
I've been in high tech for 30 years, and I've been laid off many times, most often from failed start ups. I _strongly_ disagree with a fully cynical response of working only to contract, leveraging job offers for raises, etc.
There are a few reasons for this, but the most concrete is that your behavior in this job has an impact on getting the next one. The author is correct that exemplary performance will not save you from being laid off, but when layoffs come your next job often comes from contacts that you built up from the current job, or jobs before. If people know you are a standout contributor then you will be hired quickly into desirable roles. If people think you are a hired gun who only does the bare minimum that next role will be harder to find.
On top of that, carrying around bitterness and cynicism is just bad for you. Pride in good work and pleasure in having an impact on customers and coworkers is good for you. Sometimes that means making dumb business decisions like sacrificing an evening to a company that doesn't care, but IMO that sort of thing is worth it now and then.
To be sure, don't give your heart away to a company (I did that exactly once, never again) because a company will never love you back. But your co-workers will.
I do confirm that i explicitly tested this with my super unused facebook account, just stating that i was testing restrictions on talking about Linux, the text was: """I don't often (or ever) post anything on Facebook, but when I do, it's to check if they really, as announced on hckrnews, are restricting discussing Linux. So here's a few links to trigger that: https://www.qubes-os.org/downloads/ ... https://www.debian.org/releases/stable/""" and indeed within seconds I got the following warning: """ We removed your post The post may use misleading links or content to trick people to visit, or stay on, a website. """. This is one massive wow considering how much Facebook runs on Linux.
The thing that bothers me most about layoffs due to “financial difficulties” is when you observe management wasting absurd amounts of money on something in one year, then announcing the following year that they have to make cuts to baseline, “low level” employees that don’t cost much at all.
This kind of managerial behavior seriously kills employee motivation, because it both communicates that 1) no one has job security and 2) that management is apparently incapable of managing money responsibly.
“Sorry, we spent $200k on consultants and conferences that accomplished nothing, so now we have to cut an employee making $40k” really erodes morale in ways that merely firing people doesn’t.
This is the type of thing that'd normally show up on CDC's Morbidity and Mortality Weekly Report which has been published weekly since 1960 (my understanding is this is without fail).
But unfortunately the current administration has decided an ideological purification is more important than keeping the American public apprised of threats to their health.
So it wasn't published last week, and probably won't be this week either. "Politics don't matter" though ;) Bummer!
“Jackson: In Vanguard’s view, crypto is more of a speculation than an investment. This is at the root of our decision to not offer crypto products, whether our own or others. With equities, you own a share of a company that produces goods or services, and many also pay dividends. With bonds, you get a stream of interest payments. Commodities are real assets that meet consumption needs, have inflation-hedging properties, and can play a role in certain portfolios. While crypto has been classified as a commodity, it’s an immature asset class that has little history, no inherent economic value, no cash flow, and can create havoc within a portfolio.“
Well said, this has been my own thinking on crypto, nice to see it so succinctly summarized.
30 years of their home name Office brand, known by pretty much every person that's ever had a computer
let's get rid of that, and make the unreliable bullshit generator the main brand instead
certainly a courageous decision
Hate that if you use Word for Mac, you now have Copilot next to your cursor with no way of turning it off.
I just want to use Word. I like its print layout features better than Pages. I don't want to switch. Just let me write and leave me alone. Now they're jamming AI down my throat without any opt-out mechanism.
More like "It's not a crime if a corporation does it." The US treats corporations with kid gloves. They get away with so much, and in those rare cases where a regulatory body does anything, it's often just a strongly worded letter, a warning, a threat to one day send a strongly worded letter, or, in really rare, extreme cases, a company official gets called in to say a few words in front of Congress. Nobody goes to jail, fines get whittled down to nothing in appeal after appeal, and the government usually finally just accepts a pinky-swear to never be bad again.
Contrast that to how individuals get the hammer put down on them by the justice system if they so much as fart in the wrong room. The lesson from the last 50 years should be: If you want to commit a crime and get away with it, do it as a corporation.
For those who haven't realized it yet, Deepseek-R1 is better than claude 3.5 and better than OpenAI o1-pro, better than Gemini.
It is simply smarter -- a lot less stupid, more careful, more astute, more aware, more meta-aware, etc.
We know that Anthropic and OpenAI and Meta are panicking. They should be. The bar is a lot higher now.
The justification for keeping the sauce secret just seems a lot more absurd. None of the top secret sauce that those companies have been hyping up is worth anything now that there is a superior open source model. Let that sink in.
This is real competition. If we can't have it in EVs at least we can have it in AI models!
If we actually care about public health, we should act as if both the lab leak and zoonosis theories are correct. We should take laboratory biosecurity, wet markets, the bush meat trade and intensive livestock management equally seriously as threats. We should do this because we have no idea where the next pandemic - and there will be a next pandemic - will come from.
It seems fairly clear to me that a lot of people are much more concerned about finding someone to blame for the last pandemic than about preparing for or preventing the next pandemic.