codelord

Even if the content was 100% AI generated (which is the furthest thing from reality today) human engagement with the content is a powerful signal that can be used by AI to learn. It would be like RLHF with free human annotation at scale.

As an Ex-OpenAI employee I agree with this. Most of the top ML talent at OpenAI already have left to either do their own thing or join other startups. A few are still there but I doubt if they'll be around in a year. The main successful product from OpenAI is the ChatGPT app, but there's a limit on how much you can charge people for subscription fees. I think soon people expect this service to be provided for free and ads would become the main option to make money out of chatbots. The whole time that I was at OpenAI until now GOOG has been the only individual stock that I've been holding. Despite the threat to their search business I think they'll bounce back because they have a lot of cards to play. OpenAI is an annoyance for Google, because they are willing to burn money to get users. Google can't as easily burn money, since they already have billions of users, but also they are a public company and have to answer to investors. But I doubt if OpenAI investors would sign up to give more money to be burned in a year. Google just needs to ease off on the red tape and make their innovations available to users as fast as they can. (And don't let me get started with Sam Altman.)

American and multi-national companies often build such factories in low-income countries. It's not that no one in the US knows how to build factories. It's purely a cost/regulation question.

So basically you can find qualified workers everywhere. There's a valid question about the cost. But to say the problem is lacking "skilled workers" is laughable as most of these skills are machine dependent and can only be learned on the job after the factory is built.

"manufacturing expertise exists primarily in China, Vietnam, Cambodia, and other countries" lol. What's this skill that Cambodians have that Americans can't learn or can't be automated by robots? Can they juggle 4 wrenches in the air simultaneously?

Fade in/out animation.

Having worked in the game industry in the past it's amusing to see people talk about the greed of game developers. You have no idea! You have no idea what an effort it takes to ship a game. An album is the work of one or a few individuals for a relatively short period of time with very little cost. Because of that you can have services like Spotify that allow pretty much access to all the music ever created for a 10$ fee. The math just doesn't work with video games. Video games take an army of developers and years of work to make. Game development is one of the hardest and worst paying professions in tech. Most people in the industry are there not for the money but because of their passion for the profession. Most games fail to pay for their production costs despite all the effort that goes into them. Companies who have had a few mega successes have to make enough money out of their popular titles to be able to pay for all the other titles that fail to pay for themselves. Please don't complain about video game prices!

"If writing down your ideas always makes them more precise and more complete, then no one who hasn't written about a topic has fully formed ideas about it."

Apparently, even writing it down didn't help the author with this flawed deduction.

I wish people stopped equating AI with Adobe's content policy.

I thought this is using AI so I was gonna dismiss it. But then saw the No AI sign and immediately signed up. Seriously though why is "No AI" a "feature" worth mentioning on top?

If you are on TikTok and you don't think CCP has all your personal data TikTok has collected from you, I have a bridge to sell you.

How to start Google? You can't. That holds for 99.999999% of the people. The rest 0.000001% aren't wasting their time reading Paul Graham's essays.

We live in a world where proven maniacs (e.g. Putin) have access to an arsenal of nuclear weapons that can essentially make the earth uninhabitable for all humans. That's a very real possibility (with no ifs and buts and maybes) that exists now and we have learned to live with it. Yet somehow the hypothetical scenario of a human exterminator super-intelligent AI is getting all the coverage.

30% for distribution is fine, 5% for more than half the cost of development is not?

Fortnite is also a top selling mobile game developed with UE5. You can make great mobile games with UE5 now. Mobile hardware right now is comparable to last gen consoles.

IMO Unreal Engine is the best deal available and fits >90% of use cases for game developers. Unless you are building something for the web or low powered mobile VR I wouldn't even consider anything else. For PC and console games UE5 provides incredible amount of tools and flexibility. It's also great for building 2D/3D mobile games. People who say complexity of Unreal has stopped them from using it have gotten it wrong. UE5 provides you with a lot of tools, you don't have to use them all. But if you are thinking of building something more complex than a hello world example, you'd realize that the additional tools that UE5 provides you greatly save your time.

If you are a total beginner you can use Blueprints to write the game logic and use the existing out-of-the box tools. If you are a more experienced programmer you can use C++ to build custom components/plugins to get more customization.

I remember a time that game engines were these precious secret tools that you had to pay millions of dollars to get a license for. Now you can get the full source of UE5 on Github for free. And you pay something like 5% after 1 million dollars of revenue. This is just a no-brainer folks. IMO 5% is totally deserved and justified. In fact it's a bargain and you save money by paying Epic 5% compared to anything else out there. Use UE5 unless you have a really really really good reason not to.

IMO AI is more underrated than overhyped. The scale of value that AI can bring maybe larger than what internet brought. But, product design and engineering havn't caught up with the science yet. I think we are too narrowly looking at AI. LLMs are cool, but investors should look beyond that. ChatGPT wrappers aren't the next big thing. The fact that LLMs and image generation models work as good as they do now, should give investors a signal that the science of AI is approaching a tipping point where it's finally good enough to be incorporated into products. I see potential in 10 years time for a new FAANG, 5 trillion dollar companies with heavy reliance on AI that bring automation to various aspects of our lives.

As a Comcast customer who's paying 5x what I was paying in Europe for half the bandwidth I want this to succeed. However let's not celebrate a win before actually delivering the service to the customers. There's more to building a business than seed funding it. The pessimistic in me would say if there was a viable path to providing high-speed internet with low cost in the US surely companies with a lot on the line like Netflix, Google, Amazon, etc. would make that happen. If you think it's all Comcast profit margins you can always go and buy Comcast stocks to get your share of that profit. They are doing well but not spectacularly well.

As someone who did a PhD and published and read many papers (in ML) I believe vast majority of papers in ML are misleading. If in fact every paper that claims a better performance over state of the art was true we would have solved AI by now. You see all sorts of problems when you dig deeper into the technical details of peer reviewed publications (even in top tier conferences) including misleading baselines, statistical insignificance of improvements, overfitting to test data, and in some cases just flat-out fabrication of results.

I hope this is not as bad in medicine and health related research. But just thinking that some paper can be used against you in court to claim billions of dollars in damages makes me uneasy. Peer reviewed paper != science. Peer review is a crude filter on research that can both accept bad work and reject good work. There must be a higher bar for something that can be used in the court of law. Some sort of scientific consensus must be needed at least.

It's easy to dismiss this because screw J&J. But I think we are all paying for these lawsuits through our insurances and taxes and higher drug prices.

Not saying these lawsuits don't have merits, but I think there must be a higher bar for what is presented as evidence.

Good to know I was under the impression that to draw 2 triangles I need to copy paste the whole code twice. I have been a graphics programmer for over 25 years. I'm not even commenting on the API. The toy example certainly can be reduced in code length even given the current API.

385 lines for a triangle: https://github.com/facebook/igl/blob/main/samples/desktop/Ti... When you are trying to sell a wrapper you want your hello world example as small as possible not as comprehensive as possible.

I think if the hardware is good and even more importantly developer tools are easy to use and better than competition this could be successful. It doesn't really matter what Apple shows in their demo what the use cases are. Does anyone share heart-beats on Apple watch? No. Yet, Apple Watch is a best selling product.

IMHO the prospect of an AI winter is 0%. As someone who has done research in ML I think ML technology is moving forward much faster than we anticipated. ChatGPT shouldn’t work based on what we knew. It’s incredible that it works. Makes you think what other things that shouldn’t work we should scale up and see if they would work. And then there are things that we think they should work. Each new paper or result opens the door for many more ideas. And there are massive opportunities in applying what we already have to all industries.

You can absolutely build high precision ML models. Using a transformer LM to sum numbers is dumb because the model makes little assumptions about the data by design, you can modify the architecture to optimize for this type of number manipulation or you can change the problem to generating code for summing values. In fact Google is using RL to optimize matmul implementations. That’s the right way of doing it.

I'm not sure if that technical difference matters for any practical purposes. Viruses are also not alive, but they kill much bigger and more complex organisms than themselves, use them as a host to spread, mutate, and evolve to ensure their survival, and they do all that without having any intent. A single virus doesn't know what it's doing. But it really doesn't matter. The end result is as if it has an intent to live and spread.

I can't imagine we'll have banknotes and coins in 100 years from now. Cash is neither efficient nor environmentally friendly. The efforts around helping the marginalized groups should be focused on getting them a debit card not keeping cash around.

-Charge your car at work in the sun and use it to power your home at night.

This doesn't make any sense. If you don't need the capacity, why do you wanna carry a ton of battery everywhere with you wasting energy? Buy a smaller battery for the car. Keep another battery installed at home. The battery at home is also more flexible to draw power at the optimal time for the grid.

Not to mention that this constant charging and recharging would decrease the lifetime of the battery, causing more environmentally unfriendly waste and inefficiency.

I wish /member/fire was implemented.

Yes. Let's make a lottery based economy. That would boost the productivity and make the nation prosperous. This is like fixing a pothole by creating a supermassive black hole.

I wonder what the perspective of other people who run businesses or manage teams are on this subject? I have managed teams at top tech companies in bay area in the past (although switched back to an engineering position recently). Regardless of the company, I have always wondered why everyone around me is working so slowly. Things that a good engineer should be able to do in a day gets done in a month or more. When working as an engineer people always commented that I'm very efficient. But honestly I think the reason is a lot of people just barely put 10 hours a week in the current 5 days a week routine. Lots of times I feel like you could just let 60-70% of people go at companies I've worked at and literally nothing changes. My experience has been that a handful of high output individuals at each company are responsible for 90% of things that gets done. So I wonder what happens if we reduce this even more? Probably nothing, because high output individuals usually have intrinsic motivations and will continue putting more hours than is required.

It's interesting to me people make all sort of assertions about why men or women are dominant in an industry without anyway to falsify their hypothesis. I provide some data points having lived in several countries that may offer some insight. Draw your own conclusions.

I got my undergrad degree in Iran (a relatively conservative society with an ultra-conservative even anti-women government), where the ratio of women to men in our computer science class was roughly 60/40 in favor of women. The ratio was similar all across the board in STEM and non-STEM fields as more women were admitted and graduated from universities than men. To get admission in Iran everyone has to go through the same standardized test nationally. Women were scoring better and better in the nationalized test than men every year until they passed men the year I went to school. (The trend continued later even prompting the government to introduce legislation to cap the ratio of women at least in certain programs.)

When I went to Europe for grad school (to a very progressive university in perhaps the most progressive European country) to my surprise I had only one female classmate who was a non-European international student in a class of 40.

I let you judge what caused this discrepancy. But I hope we can all agree the issue is much more complicated than some want to make you believe.

I haven't read the paper yet, no comment on the content. But it's amusing that more than 30% of references are self-citation.