Some interferences even expose a way to set the “temperature” - higher values of that mean more randomized (feels creative) output, lower values mean less randomness. A temperature of 0 will make the model deterministic.
You know how Google’s new feature called AI Overviews is prone to spitting out wildly incorrect answers to search queries? In one instance, AI Overviews told a user to use glue on pizza to make sure the cheese won’t slide off (pssst…please don’t do this.)...
No, it’s simply contradicting the claim that it is possible.
We literally don’t know how to fix it. We can put on bandaids, like training on “better” data and fine-tune it to say “I don’t know” half the time. But the fundamental problem is simply not solved yet.
It does not perform very well when asked to answer a stack overflow question. However, people ask questions differently in chat than on stack overflow. Continuing the conversation yields much better results than zero shot.
Also I have found ChatGPT 4 to be much much better than ChatGPT 3.5. To the point that I basically never use 3.5 any more.
I understand the motivation behind this opinion and would like to see testing of beauty products on animals outlawed. But pigs with lipstick is not really what you take the most issues with, is it? It’s about giving rabbits cancer so we can test new cancer drugs on them. Assuming we make that illegal, how do you propose new cancer treatments should be tested?
No one likes animal trials, most of all the researchers themselves who work with the animals. For example researchers cannot take any vacation during the trial. In fact someone needs to be in the lab at least once a day, including Sundays and public holidays.
Also animal trials are expensive.
Research on alternatives is progressing. It’s not like there is a big conspiracy of sociopaths that get off on animal suffering and want to keep the status quo because of that. It’s simply really really really hard to simulate a body to the necessary level.
Assuming a young adult develops, idk breast cancer or something. Your sister or your daughter or you maybe. Should we treat it?
If we don’t they’ll die.
But careful, it will cost a hundred rats and a few rabbits their lives.
You’d be surprised how many diseases were fatal once. Aside from fixing broken bones or something like that, drinking tea, suffering and praying it will get better are your only options for the vast majority of diseases if you truly want to live with a clear mind.
Models and simulations are already used to develop new drugs. But the tech is simply not there yet to rely on that exclusively.
Voluntary human trials are already a requirement as well. The problem is that there are not many cancer patients that are beyond saving and still in the condition to take part in such trials.
I don’t know how it is in other countries, but in Germany it’s illegal to pay someone to take part in clinical trials. You are paid money to compensate you for your time and travel costs, but it’s not proper pay. And there is very good reason for that. If you make medical trials a viable income, you will inevitably get poor people to take part in them. You may consider that more ethical than animal trials. I do not.
The comparison with risky jobs is not valid, because we do our best to make those jobs safer. A trial for a new drug inevitably involves getting infected with the disease the drug is supposed to cure.
If you want to look at it from such a fundamentalist angle, sure, animal testing is immortal. You’d only be able to test new drugs on terminally ill patients then.
If you’re willing to humor me, let me take you on a tanget. I promise it’ll make sense: Do you agree that CO2 emissions are fundamentally wrong (leading to a mass extinction event, etc)?
(I will continue this argument under the assumption that we can all agree on that) And do you concede that these emissions are, for the foreseeable decade(s) inseparable from modern human life? Not that they are a basic necessity to survive, but that you and I are indirectly causing such emissions in one way or another for every day that we are alive and continue with our day to day actions (heating, cooking, buying stuff, transportation, etc). This may change in the future, but let’s focus on today.
(Again I assume that you are not the 0.1% of the population that lives without any modern amenities (you have some way of writing comments on the Internet for example), and will continue my argument) Given these two basic building blocks of our mutual understanding of the world:
Neither you nor me have committed suicide. So there is a reason that we continue on living, despite our continued existence being linked to habitat destruction and animal deaths. We are working towards a better future and try to change that, but for some reason we consider our current lives more important than the lives of animals that are threatened as a consequence of our existence. I don’t know why, and you probably neither. I guess it’s just some deeply rooted desire for survival.
Oh, btw, I am actually curious what your answer is to the 100 rats question someone else posted in the comments. Or maybe rephrased a bit: is there any number of rats (or rabbits or fish or dogs) whose deaths you’re unwilling to accept and that makes you say “no, take my sibling/partner/parents instead”?
If you’re a suitable candidate I’m pretty sure they already explicitly ask you if you’d like to participate.
I’m curious what methods you suggest to make medical trials more safe. In e.g. construction we can regulate personal protection equipment and mandate machines to do the heavy lifting instead of the workers. We can mandate more time off and corporate fitness programs to keep the people healthy. But how would you make a safer clinical trial?
Oh, I had hoped for some proper thoughts on the matter. The first suggestion is too vague and the other two are already the status quo. In fact I have heard (anecdotally) someone got treated for something that, with 99.9% certainty, was unrelated to the clinical trial they, were participating in. But it occurred during the trial, so their health care costs were covered in full by the clinical trial. And if you ever witness that participants of a clinical trial were not fully informed you should report it. The ethics committee and lawmakers take that extremely seriously.
Edit: to better explain my previous point about safety. When we talk about a usual job being dangerous what we mean is that you’re supposed to do A, but B might happen and hurt you (build the scaffolding for a house, but pinch your hand when connecting two pieces). Therefore we mandate PPE, maximum working hours, machine assurance, etc. This is possible because the actual job is tangential to the risk associated with it.
A clinical trial is, going from beginning to end: we have simulated this drug in the computer and tested it on cells. Now we need to check for interactions with other parts of the body. For statistically significant results we need 50 animals, we put cancer in them, wait two weeks and then start treating them like we would treat a human who has this cancer. We vary dose and when to give it, maybe the mixture of compounds if the drug is not just a single active component. A lot of the animals will have to be put down when it becomes apparent that this configuration of the drug does not work. But we have a better understanding of the working of the drug in the body now. After that trials move on to human patients. First we start with people who are sick and for whom the current method of treatment did not work. They will die soon anyway, but there is a small chance that the new drug will work on them. Again we vary dose etc, but now we know much better what range of dose is useful. This results in much more difficult to handle data, because taking a few random people will introduce wild variations in confounding factors. But it’s a necessary step to show that the drug works in humans, because we can’t move on to testing the new drug on people for which the old method of treatment would have worked. After this trial is done it is finally time to try the new drug out on people who come to the hospital to seek treatment. The doctor may offer you the chance to participate in a trial for a new medicine if they think you’re a fitting candidate. This trial will test the medicine for the first time against a proper sample of the population. Only now can we say for certain how much better it is than the old drug (or maybe it’s worse) and tease out details from the data (e.g. It’s usually better, but it’s worse if the person has a cold and is overweight when starting treatment. Or it causes severe allergic reactions for people who have asthma that is triggered by grass pollen)
It’s important to note that at every step of the process a drug can fail testing. Researchers want the drug to fail early, because every step costs money and time. When we get better simulations or artificial organs to test on they will be used, because it’s so much faster and cheaper than going to animal trials with a promising drug, only to find out after three months of hard work that it doesn’t work.
Now, the safety concerns in clinical trials is that we have a current drug that works, and we have a new drug that may work. Is it ok to not treat someone with the known working one, just to see if the maybe working one helps? Most people say no. The danger is inherent in the thing, which is why we have such a lengthy process. There is no PPE you can wear to reduce the effects of cancer when the trial requires you to have cancer. We must get to the stage of “it’s most likely working better than the current treatment” before starting testing on otherwise healthy humans.
Fair enough. It would put a stop on developing new medicine for a while (5-20 years maybe?), but I can understand the opinion that “what’s done is done, we just should not continue doing that”.
Fair enough. If you can recognize that you have a strong opinion based on ethics, and are willing to read up on how things are currently done and what the problems are (both with the current way and with the way that your ethics would like it to be) thats fine.
Let me ask you this: are you opposed to professional fighting? Boxing, wrestling, wwe, etc?
I’m not a sports guy (at least watching sports, I do exercise weekly) and would barely notice if those would no longer exist tomorrow. So I am certainly not one to defend their existence.
And yes, I am super critical of professional sport and how much these people hurt themselves. In German we have a saying: “Sport ist Mord”, sports is murder. I think in the broad population it’s also used as an excuse if you’re lazy and don’t want to exercise, but for me it appropriately hits on the problem of professional sport. Some are better than others, for example I have not heard of many negative consequences from swimming on a professional level. But I think the problems that people get from playing rugby on a professional level are absurd. There are measurable levels of IQ drop after a few years of working as an athlete. I have absolutely no idea why anyone would willingly do that.
One difference is that in order to get to such a level you need talent and need to be into it from a young age. Yes, some people can lift their family out of poverty with it. But not because they needed some quick money.
A better comparison to paying big money for participants of clinical trials than sports is selling your kidney. You only need one, technically, so it’s safe on paper. But it’s a surgery that comes with some inherent risks to your life. And there is a reason we usually have two.
And again, the injury is tangential to the performance. In clinical trials a sizable fraction of the “patients” die (cells, animals, humans. The earlier in the trial the bigger this fraction. Animal test are there to hopefully have the number be zero when we get to human trials), until we know what dose is effective and safe at the same time.
Has anyone else had this problem? I updated my openSuse Tumbleweed today (restart required), and now none of my games work. Most are through Steam, so at first I thought it was that. But I have Cyberpunk through GOG launched with Heroic. And even more, Alien Arena has the same problem, so it’s not even a proton issue (I did...
Ich kann die Beweggründe hinter dem Artikel verstehen und stimme auf der Metaebene auch zu, dass teurer Wohnraum ein großes gesellschaftliches Problem ist.
Aber ich hab das Gefühl, dass sich die Autorin darüber aufregt, dass Leute Lösungsvorschläge machen, wenn sie ihr Problem beschreibt. Ist ok, wenn man nur Mitleid haben und nichts an der Situation ändern will. Aber Alternativen aufzeigen ist mMn auch eine legitime Antwort.
Wenn ich unbedingt auf Fiji Urlaub machen will (es muss kein 5 Sterne Hotel sein, ich will einfach nur ein Bett und da an den Strand) aber nichts in meinem Budget finde, dann ist es doch völlig legitim, dass Leute sagen “in Italien war ich Mal an einem so geilen Strand, da vergisst du komplett in welchem Land du offiziell bist, du bist einfach im Himmel”.
Most games are apparently built like this. But there are two solutions: as you’ve said, making the UI configurable is one of them. (In KSP you can move the navball left and right on the bottom edge of the screen) Or the UI placement can be made with modern ultra wide screens in mind. W.g. the minimap is placed 40px from the right edge, unless the screen is an ultrawide one, then the UI is restructed to a box that is equivalent to a 21:9 (or 16:9) screen.
Yeah, but the YouTube culture used to not use swearwords/unwanted words as a result of this. I have recently seen a few videos, where they said prn and sx, including cutting the vowels from the audio. This change in how to deal with the restrictions imposed by the platform is what I don’t understand.
Edit: I have edited my comment above to be more precise.
But My Leftovers (lemmy.world)
Gemini doesn't share Google search's AI advice on pizza cheese solutions? (sh.itjust.works)
CEO of Google Says It Has No Solution for Its AI Providing Wildly Incorrect Information (futurism.com)
You know how Google’s new feature called AI Overviews is prone to spitting out wildly incorrect answers to search queries? In one instance, AI Overviews told a user to use glue on pizza to make sure the cheese won’t slide off (pssst…please don’t do this.)...
Anon reinvents something dark (sh.itjust.works)
17 cringe-worthy Google AI answers demonstrate the problem with training on the entire web (www.tomshardware.com)
These are 17 of the worst, most cringeworthy Google AI overview answers:...
So I was buying new rack... (sopuli.xyz)
And I looked at what I can add to have free shipping. After I told my mum that they have unicycles she told me: “It will be f-ing awesome, buy it.”...
LPT Do it. (mander.xyz)
Humans need to stop being cruel (lemmy.ca)
[Solved] Tumbleweed update killed my gaming
Has anyone else had this problem? I updated my openSuse Tumbleweed today (restart required), and now none of my games work. Most are through Steam, so at first I thought it was that. But I have Cyberpunk through GOG launched with Heroic. And even more, Alien Arena has the same problem, so it’s not even a proton issue (I did...
Autism (mander.xyz)
Wohnungsnot in Städten: Ein total normaler Lebensentwurf (www.taz.de) German
Which do you prefer? A triple monitor setup? Or one single 32:9 ultrawide monitor?
I have a triple monitor setup right now, but am thinking of trying a single ultrawide monitor. What is your preference?
vampire rule (lemmy.world)
Mozilla Firefox Adds AI-Powered RTX Video (blogs.nvidia.com)