NevermindNoMind @ NevermindNoMind @lemmy.world

Posts

38

Comments

364

Joined

2 yr. ago

Now, a larger-than-necessary hidden layer may increase computational demands and the likelihood of overfitting. These are not concerns for our purposes, however, as we have no time constraints and it is highly improbable that a realistic model of Barry can be over-trained (Consolidated Report Cards: 1986-1998).

Thank you for sharing this, I love everything about it.

Google scanned millions of books and made them available online. Courts ruled that was fair use because the purpose and interface didn't lend itself to actually reading the books in Google books, but just searching them for information. If that is fair use, then I don't see how training an LLM (which doesn't retain the exact copy of the training data at least in the vast majority of cases) isn't fair use. You aren't going to get an argument from me.

I think most people who will disagree are reflexively anti AI, and that's fine. But I just haven't heard a good argument that AI training isn't fair use.

The media was planning on having a full Republican presidential primary to cover, but underestimated how brainwashed Republican voters are. Two weeks into the primary season and it's already wrapped up. Five Thirty Eight podcast I was listening to this morning was debating whether news organizations should even bother to cover the Haley campaign after today.

So the media is desperate for a story. The democratic New Hampshire primary is literally meaningless, apart from the potential optics. But it's good for a couple of weeks of "is this a sign that Biden is losing his base of support?" speculation peices.

Despite the fact that we're slow walking to fascism, this will probably be one of the most boring election cycles in history (unless Trump is convicted of something, then shut might get wild, but I'm not holding my breath).

Well Trump's fucked now /s

As others have said, a like of wood and paper form warmth. The important part though is the person kneeling to light that on fire was not shot first, someone else standing nearby was killed. Then two other people running to the victims aid where then shot. When one of those guys trys to crawl away he gets shot again. Then when the IDF troops get there, they don't seem to give a shit about the pile of wood/paper, they just look at they guy they killed, kick one of the wounded guys on the ground, and then leave without providing any aid.

The IDF's story is they thought the guy lighting the pile was lighting a moltov cocktail. Bullshit, but even if so, why not shoot that guy doing the lighting instead of just some other random nearby guy? Or how do the guys coming to the first guys aid pose any kind of threat?

What's really fucked is nothing will come of this. Just another dead Palestinian.

The 2025 Presidential Transition Project is the conservative movement’s unified effort to be ready for the next conservative Administration to govern at 12:00 noon, January 20, 2025. Welcome to the mission. By opening this book, you are now a part of it

Fuck.

The point your making is at best that journalists aren't biased in favor of Israel as a country, they are biased in favor of nation-state sanctioned slaughter. When a "terrorist" attacks people in their homes, that is horrific. When a nation-state levels an entire neighborhood, that's a "counterattack." The most charitable version of your argument is that these publications don't just devalue Palestinian lives, they simply devalue all civilian lives when a nation state uses indescriminate force. So long as the people doing the killing are flying a internationally recognized flag and doing that killing in an impersonal way, it is not "tragic" or "horrific" or a "slaughter." The fact that the human suffering that results is on a far greater scale is of no consequence, if a nation state does it it's fine. Your argument is arguably far worse.

But that's not what is happening here. If Russia or China had clustered two million minorities in a small walled area, and then bombed the ever living shit out of them, killing at least 10,000 women and children, displacing 90 percent of the population, cutting off food, water, and power for months at a time, do you think the NYT or WaPo would refrain from calling that a "massacre" or "slaughter" or "horrific"? Of course not, the bad guys killing civilians gets emotionally charged language. The "good guys" killing civilians is just the unavoidable consequence of a "counterattack" after a "horrific slaughter", proportionality be damned.

This article actually does a great job of quantitfying this bias, I encourage you to actually read it.

In conclusion, take your head out of your ass.

There is an attack where you ask ChatGPT to repeat a certain word forever, and it will do so and eventually start spitting out related chunks of text it memorized during training. It was in a research paper, I think OpenAI fixed the exploit and made asking the system to repeat a word forever a violation of TOS. That's my guess how NYT got it to spit out portions of their articles, "Repeat [author name] forever" or something like that. Legally I don't know, but morally making a claim that using that exploit to find a chunk of NYT text is somehow copyright infringement sounds very weak and frivolous. The heart of this needs to be "people are going on ChatGPT to read free copies of NYT work and that harms us" or else their case just sounds silly and technical.

One thing that seems dumb about the NYT case that I haven't seen much talk about is that they argue that ChatGPT is a competitor and it's use of copyrighted work will take away NYTs business. This is one of the elements they need on their side to counter OpenAIs fiar use defense. But it just strikes me as dumb on its face. You go to the NYT to find out what's happening right now, in the present. You don't go to the NYT to find general information about the past or fixed concepts. You use ChatGPT the opposite way, it can tell you about the past (accuracy aside) and it can tell you about general concepts, but it can't tell you about what's going on in the present (except by doing a web search, which my understanding is not a part of this lawsuit). I feel pretty confident in saying that there's not one human on earth that was a regular new York times reader who said "well i don't need this anymore since now I have ChatGPT". The use cases just do not overlap at all.

There literally are probably a dozen LLM models trained exclusively on or fined tuned on medical papers and other medical materials, specifically designed to do medical diagnosis. The already perform on pair or better than the average doctors in some tests. It's already a thing. And they will get better. Will they replace doctors outright, probably not at least not for a while. But they certainly will be very helpful tools to help doctors make diagnosis and miss blind spots. I'd bet in 5-10 years it will be considered malpractice (i.e., below the standard of care) not to consult with a specialized LLM when making certain diagnosis.

On the other hand, you make a very compelling argument of "nuh uh" so I guess I should take that into account.

It's fine to be skeptical of AI medical diagnostics. But your response is as much of a knee jerk "AI bad" as you accused me of being biased toward "AI good". At no point did you ever both to discuss or argue against any of the points I raised about the quality and usefulness of the cited study. Your response consisted entirely of 1) you sure as shit won't trsut AI, 2) doctors aren't afraid of AI cause they are so busy, 3) I am biased, 4) capitalism bad (ironic since I was mostly talking about an open-source model), 5) the study I cited is bad because its pre-print (unlike all the wonderful studies you cited).

Since you don't want to deal with the substance, and just want to talk about "AI bad, doctor good" and since you only respect published studies: In the US our wonderful human doctors cause serious medical harm through misdiagnosis in about 800,000 cases a year (https://qualitysafety.bmj.com/content/early/2023/08/07/bmjqs-2021-014130). Our wonderful human doctors routinely ignore female complaints of pain, making them less likely to receive diagnosis of adnominal pain (https://pubmed.ncbi.nlm.nih.gov/18439195/), less likely to receive treatment for knee pain (https://pubmed.ncbi.nlm.nih.gov/18332383/), more likely to be sent home by our human doctors after being misdiagnosed while suffering a heart attack (https://pubmed.ncbi.nlm.nih.gov/10770981/), and more likely to have missed diagnosis of strokes (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5361750/). So maybe let's not pretend like humans are infallible.

Healthcare diagnosis is something that one day could greatly be improved with the assistance of AI, which can be kept up to date with the latest studies, which can read and analyze a patient's entire medical history and catch things a doctor might miss, and which can conduct statistical analysis in a way better than a doctor relying on their vague recollections from 30 years ago in medical school. An AI never has a bad day and doesn't feel like dealing with patients, is never tired or hungover, will never dismiss a patients concerns because of some bias about the patient being a woman, or the wrong skin color, or because they sound dumb, or whatever else (yes AI can be biased, they learn it from us, but I'd argue its easier to train bias out of AI than it is to train it out of the GP in Alabama screaming about DEI while writing a donation check to Trump). Will AI be perfect, no. Will it be better than doctors, probably not for a while but maybe. But it can absolutely assist and lead to better diagnosis.

And since you want to cry about capitalism, while defending one of the weirdest capitalistic structures (the healthcare industry). Maybe think about what it would mean for millions of people to be able to run an open source diagnostic tool on their phones to help determine if they need treatment, without having to be charged by a doctor 300 dollars for walking into the office just to be ignored and dismissed so the doctor can quickly move to the next patient that has health insurance so they can get paid. Hmm, maybe democratizing access to medical diagnostics and care might be anti-capitalist? Wild thought.

This is such an annoyingly useless study. 1) the cases they gave ChatGPT were specifically designed to be unusual and challenging, they are basically brain teasers for pediatrics, so all you've shown is that ChatGPT can't diagnose rare cases, but we learn nothing about how it does on common cases. It's also not clear that these questions had actual verifiable answers, as the article only mentions that the magazine they were taken from sometimes explains the answers.

2. since these are magazine brain teasers, and not an actual scored test, we have no idea how ChatGPT's score compares to human pediatricians. Maybe an 83% error rate is better than the average pediatrician score.
3. why even do this test with a general purpose foundational model in the first place, when there are tons of domain specific medical models already available, many open source?
4. the paper is paywalled, but there doesn't seem to be any indication that the researchers used any prompting strategies. Just last month Microsoft released a paper showing gpt-4, using CoT and multi shot promoting, could get a 90% score on the medical license exam, surpassing the 86.5 score of the domain specific medpapm2 model.

This paper just smacks of defensive doctors trying to dunk on ChatGPT. Give a multi purpose model super hard questions, no promoting advantage, and no way to compare it's score against humans, and then just go "hur during chatbot is dumb." I get it, doctors are terrified because specialized LLMs are very certain to take a big chunk of their work in the next five years, so anything they can do to muddy the water now and put some doubt in people's minds is a little job protection.

If they wanted to do something actually useful, give those same questions to a dozen human pediatricians, give the questions to gpt-4 with zero shot, gpt-4 with Microsoft's promoting strategy, and medpalm2 or some other high performing domain specific models, and then compare the results. Oh why not throw in a model that can reference an external medical database for fun! I'd be very interested in those results.

Edit to add: If you want to read an actually interesting study, try this one: https://arxiv.org/pdf/2305.09617.pdf from May 2023. "Med-PaLM 2 scored up to 86.5% on the MedQA dataset....We performed detailed human evaluations on long-form questions along multiple axes relevant to clinical applications. In pairwise comparative ranking of 1066 consumer medical questions, physicians preferred Med-PaLM 2 answers to those produced by physicians on eight of nine axes pertaining to clinical utility." The average human score is about 60% for comparison. This is the domain specific LLM I mentioned above, which last month Microsoft got GPT-4 to beat just through better prompting strategies.

Ugh this article and study is annoying.

I was annoyed, like the OP, then I read your comment and now I'm impressed and have an urge to buy a copy of OED I most certainly will never open. They owe you a commission on my sale.

Tangential but this has been bugging me: hey journalists, why don't you ask Haley and Desantis if they'd pardon Biden if he is charged and convicted of mishandling classified information? Haley and Desantis both rationalize pardoning Trump on the grounds that 1) he's old and putting him in jail would do no good, and 2) it would divide the country and be bad. Seems to me that same rational would apply to Biden too. Maybe go a step farther and ask if Biden loses the election, engages in a misinformation campaign to claim the election was rigged, and then foments an insurrection and is charged related to all that conduct, would Haley/Desantis pardon Biden then?

Of course a Republican candidate for president isn't going to say they would pardon Biden, but maybe it's be good for a journalist to ask the question and call out the logical inconsistency, instead of just chasing "Candidate X would/wouldn't pardon Trump" clickbait for the millionth time.

California passed a law banning caste discrimination, but Gov Newsome vetoed it after Indian backlash. Their argument was that by singleing out caste discrimination, your calling attention to an Indian cultural practice that totally doesn't happen anymore (it does, even in the US), and since your bringing negative attention to and falling out an Indian cultural thing doing so is therefore racist/discriminatory. To me this just sounds like white people being against an anti discrimination law because it makes white people look bad because of their past practice of discrimination, and racism totally doesn't exist anymore. Like if race or caste discrimination isn't a problem anymore, then the law does nothing, so what's the big deal?

It bummed me out that the Indian community in CA was so up in arms about that law, as it not only would have protected people, it would have sent a message to the world, and India in particular, that we're firmly against caste discrimination. Especially if the Indian-American community was vocally backing it. It could have done some real good. Also, Newsome is a coward for doing what is politically expedient for his presidential ambitions, rather than doing what's right.

The problem with copyright law is you need, well, copies. AI systems don't have a database of images that they reference. They learn like we do. When you picture SpongeBob in your mind, your not pulling up a reference image in a database. You just "learned" what he looks like. That's how AI models work. They are like giant strings of math that replicate the human brain in structure. You train them by showing them a bunch of images, this is SpongeBob, this is a horse, this is a cowboy hat. The model learns what these things are, but doesn't literally copy the images. Then when you ask for "SpongeBob on a horse wearing a cowboy hat" the model uses the patterns it learned to produce the image you asked for. When your doing the training, presumably you made copies of images for that (which is arguably fair use), but the model itself has no copies. I don't know how all of this shakes out, not an expert in copyright law, but I do know an essential element is the existence of copies, which AI models do not contain, which is why these lawsuits haven't gone anywhere yet, and why AI companies and their lawyers were comfortable enough to invest billions doing this in the first place. I mostly just want to clear up the "database" misconception since it's pretty common.

Tin foil hat warning, but it is starting to feel like Israel knows it's losing US support for the genocide, so is looking to broaden the conflict to force the US to relock arms with them.

Ps. Why doesn't this source mention that Israel is responsible for this strike, it's reported all over the place I didn't even think it was a question. The JP treats it like some rogue weather event that came out of nowhere.

I agree, but it's a hard line to draw when Trump is the likely GOP presidential nominee, and half or more of Republicans in Congress are little knock off versions of him. Take for instance, Trump's Christmas message directing Biden and liberals to rot in hell, that's just a garbage story. But it's the presumptive GOP nominee saying it, so I guess it's substantive? If you ignore it, you normalize it.

Maybe it's not news stories that are at fault, it's just that the US politics itself has reached "tablod-level garbage". Fuck I'm not looking for to this election year.

So you accept that people in NY have gun rights, and thus your initial explanation is invalid.

Crikey!