hooande

Medical debt is different. The legal system frowns on people running up credit card debt to pay for PS5s or nice vacations with no intention of ever paying it back. That's tantamount to theft. Most medical debt is involuntary and necessary to survive. It doesn't make sense for it to have the same penalties as other forms of credit.

In general in the US, life saving or emergency medical care is administered without regard for the patient's ability to pay. Hospitals are already subsidized or compensated in various ways for this. The real issue is preventative or precautionary care. If Americans had that for free, like with the NHS, there would be fewer $XXX,XXX debts later in life.

LLMs were specifically trained to emulate human interaction patterns. Of course we sound like them at times. It's the things we can do that they can't that are relevant.

If I study Einstein and learn to do a really good impression, the statement "Einstein often sounds like karmacondon" will be true. That does not make me Einstein.

In my experience a non technical founder should have one of the following:

1) Access to capital, normally through family or a family friend. I've worked at several companies where the main thing the CEO brought to the table was that someone trusted him or her enough to invest millions of dollars.

2) At least 5 years working a 9-5 job in the target industry and the associated social connections and experience. This eliminates most college students, sadly.

3) Something unique that enables the execution of the idea. This is normally a relationship or insider knowledge. The answer to "Why you?" can't be "Because I had the idea".

The most common exception I see to this list is when both founders have the same level of passion for solving a given problem. If you have to explain the opportunity and get someone else interested in it, it could be a tough road.

That said, don't lose hope. It's a big world. People meet and things happen.

This really seems like "DALL-E", but for videos. I can make cool/funny videos for my friends, but after a while the novelty wears off.

All of the AI generated media has this quality where I can immediately tell that it's ai, and that becomes my dominant thought. I see these things on social media and think "oh, another ai pic" and keep scrolling. I've yet to be confused about whether something is ai generated or real for more than several seconds.

Consistency and continuity still seem to be a major issues. It would be very difficult to tell a story using Sora because details and the overall style would change from scene to scene. This is also true of the newest image models.

Many people think that Sora is the second coming, and I hope it turns out to have a major impact on all of our lives. But right now it's looking to have about the same impact that DALL-E has had so far.

Debugging is a problem. But the real problem I'm seeing is our expectations as software developers. We're used to being able to fix any problem that we see. If a div is misaligned or a column of numbers is wrong we can open the file, find the offending lines of code and FIX it.

Machine learning is different because every implementation has a known error rate. If your application has a measured 80% accuracy then 20% of cases WILL have an error. You don't know which 20% and you don't get to choose. There's no way to notice a problem and immediately fix it, like you can with almost every other kind of engineering. At best you can expand your dataset, incorporate new models, fix actual bugs in the code. Doing those things could increase the accuracy up to, say, 85%. This means there will be fewer errors overall, but the one that you happened to notice may or may not still be there. There's no way to directly intervene.

I see a lot of people who are new to the field struggle with this. There are many ways to improve models and handle edge cases. But not being able to fix a problem that's in front of you takes some getting used to.

Alternative theory: ChatGPT was a runaway hit product that sucked up a lot of the organization's resources and energy. Sam and Greg wanted to roll with it and others on the board did not. They voted on it and one side won.

There isn't a bigger, more interesting story here. This is in fact a very common story that plays out at many software companies. The board of openai ended up making a decision that destroyed billions of dollars worth of brand value and good will. That's all there is to it.

Why wouldn't Ilya come out and say this? Why wouldn't any of the other people who witnessed the software behave in an unexpected way say something?

I get that this is a "just for fun" hypothesis, which is why I have just for fun questions like what incentive does anyone have to keep clearly observed ai risk a secret during such a public situation?

For real. It's like, did you see Oppenheimer? There's a reason they put the military in charge of that.

This is one of the most insightful comments I've seen on this whole situation.

The people working there would know if they were getting close to AGI. They wouldn't be so willing to quit, or to jeopardize civilization altering technology, for the sake of one person. This looks like normal people working on normal things, who really like their CEO.

OpenAI existed for years before ChatGPT. Granted, at much smaller size and with hundreds fewer employees.

I imagine that the board wants to go back to that or something like it.

If this were true they never would have had talks to bring him back. That's the opposite of steadfast commitment to principles. If Sam wronged them or the company in a significant way they never should have let him back in the building.

The board's decisions may or may not turn out to be correct in hindsight. But it's very difficult to say that this was a good example of leadership or decision making.

In all these years I'd never seen this. Ironic, but not surprising, that according to this account Snowden did exactly what he accused the US government of doing: mass collecting data with no authorization or purpose and then using it to accuse someone he disagreed with of crimes.

These companies have hired focus groups, marketing experts, psychologists, and countless design teams to get people hooked on their platform. Are we really surprised?

I don't think that this is the problem. From what I know, hackernews doesn't hire any marketing experts, psychologists or focus groups. It doesn't even support images. I was more addicted to this site than any other, despite the lack of psychological tricks.

And there are many sites that DO employ full psychological warfare teams that I completely ignore. Tinder seemed fully committed to forcing repetitive user engagement. And I dropped that site after about two days, psychologists or not. If all it took to force engagement was a certain list of UI tricks, every funded social site would be able to do it.

I don't think there's a single root cause or an off switch for social media. This phenomenon is here to stay, for better or worse. I think that we will adapt as a species but there's no going back.

"You do not provide an API key. Please enter your openai key"

I'm not entering my openai key to a random website. My billing is tied to that.

This really makes it difficult to demo the product.

This idea seems like it's 100% about distribution. If I owned an extreme sports rental shop, I would have an incentive to get more people to go out and participate in the activities. If the app was well made, reputable and secure I would consider putting up a sign in my store or whatnot.

People that run equipment rental stores probably have a facebook group or professional association. If you can befriend someone influential in one of those, it might be a good place to get started.

It's gotten better for everyone in the last few months. It used to be a nightmare, but I haven't seen a timeout or rate limit error in a long time.

There is no clear answer. It's debatable among experts.

The grandparent post seems to believe that the issue is algorithmic complexity and programming aptitude. Personally, I think that all the major LLMs are using the same basic transformer architecture with relatively minor differences in code.

GPT is trained on more data with more parameters than any open source model. The size does matter, far more than the software does. In my experience with data science, the best programmers in the world can only do so much if they are operating with 1/10th the scale of data. That applies to any problem.

The difference is that an LLM isn't 1000 different intelligences. It's one intelligence, being asked to pretend to be 1,000 different people. Every instance is the essentially the same weights trained on essentially the same data. The difference in perspective doesn't resemble that of the difference between any two humans.

Humans love to think of multi agent systems as being like a team of people. It's much more like a writer imagining different characters and how they would respond. When George RR Martin imagines all 500 characters in Game of Thrones, there is a lot of diversity of perspective and thought there. But all of that is coming from one intelligence and doesn't represent a collaboration in any traditional sense.

The problem with this is that it has no memory across the different contexts. An analogy would be giving one page of a five page document to five different people, then taking it away and asking them to collaborate. While they can each give more attention to their individual page, none of them can see the whole picture and a lot of information will be lost when trying to communicate.

You can use multiple agents, or split a lot of information across multiple requests to one agent. The result is the same. Some problems require a full understanding of the whole picture.

I think this was overall positive as a social phenomenon. It did not result in a superconductor future, but it got a lot of people interested in science and practical experiments. The biggest cost was several labs doing work they wouldn't have done otherwise. And that seems fairly minor when compared to the cultural exposure that their work received.

It's unfortunate that it probably won't work out. But this was fun. Good story, good experience.

why don't they just repro their own experiment? If they can do it twice, then it isn't a fluke. at worst it's something specific to their lab. if they can't do it a second time, then the issue is settled.

I'm sure it will take a lot of time and money to run everything again. but all of earth seems willing to give them whatever resources they need.

Such a strong dismissal of another comment should come with some kind of objective evidence. Is there any way to prove that fine tuning is as valuable as you say it is?

This will create an excellent search engine but a terrible reasoning machine.

There are a lot of ways to search through docs and support tickets now. The ability of an LLM to draw inferences and summarize all of that information comes from being trained on a very large amount of data with billions of parameters. The data can be highly specialized. There just needs to be several thousand gigs of it for the AI to do things that are rare and useful.

This assumes that their technology would be similar to ours. They could have used biological mechanical parts, made from plant or animal material. They could have had radically different theories of mechanics or construction that better fit the state of the planet at their time.

In general the statement "They couldn't have been civilized because they aren't exactly like us" limits what we look for and how we look for it.

Physics and gravity are behind any kind of predictable motion. But you don't need to understand those things at all to be a successful surfer. Even though surfing is entirely about physics and calculable predictions, performing the act doesn't require any detailed knowledge of either topic.

It's the same with calculus and almost everything you mentioned. People can create algorithms, make statistics calculations and financial predictions, build robots, etc. All without any knowledge of calculus of any kind.

The skills of all of those things are based on calculus like surfing is based on physics. Related, but not in the sense of practical application. Knowledge of the math that underlies the math that underlies the thing is neither required nor sufficient for actually doing the thing.

I thought that regulation was based on the idea that it's difficult to start new, competing social networks. But the market seems to be taking care of that. Why regulate twitter if any miscellanous billionaire can start a competitive service?

Regulation means that companies have to follow arbitrary rules, including rules that you may not like. Wait until every social site is legally required to ban you if you violate a given rule.

The social networking space seems to be working itself out It's just a slow process.

what was the implied threat?

Data quality like you're describing just doesn't matter. GPT is trained on PEBIBYTES of data. Any individual reddit thread is an atom in a drop in a bucket. All of reddit is < 2% of the total training data, regardless of quality.

Yes, the correct thing to do is get more data. Much more.

Do you have a source link for this?